qf/quakeforge
0
0
Fork 0
mirror of https://git.code.sf.net/p/quake/quakeforge synced 2024-12-11 21:31:30 +00:00
Code Issues Projects Releases Packages Wiki Activity
quakeforge/libs/video/renderer/gl/gl_rsurf.c

893 lines
21 KiB
C
Raw Normal View History

initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
/*
gl_rsurf.c
surface-related refresh code
Copyright (C) 1996-1997 Id Software, Inc.
Copyright (C) 2000 Joseph Carter <knghtbrd@debian.org>
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to:
Free Software Foundation, Inc.
59 Temple Place - Suite 330
Boston, MA 02111-1307, USA
*/
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
re-arrange things so __attribute__ and __builtin_expect are properly autoconfiscated so rcsid will continue to work with gcc 3.3
2003-01-15 15:31:36 +00:00
First pass at updating nq particles. Compiles. Crashes. Taniwha wanted a look. Flee in terror. Also has bunches of cleanups, including string.h/strings.h ifdefs, whitespace, and removal of pointless comments.
2001-05-09 05:41:34 +00:00
#ifdef HAVE_STRING_H
# include <string.h>
#endif
#ifdef HAVE_STRINGS_H
# include <strings.h>
#endif
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Handle alloca "correctly". Use AC_FUNC_ALLOCA and the #ifdef mess suggested by the autoconf docs (hidden in qfalloca.h).
2013-01-22 05:09:41 +00:00
#include "qfalloca.h"
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
#include <math.h>
emptiness shuffled.
2001-05-11 20:50:16 +00:00
#include <stdio.h>
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
no header includes cvar.h now
2001-05-31 03:41:35 +00:00
#include "QF/cvar.h"
Remove client.h from sw_rsurf.c. Prepare gl_rsurf.c for same.
2001-05-21 03:39:41 +00:00
#include "QF/render.h"
gl_rsurf is now client clean
2001-05-21 23:09:46 +00:00
#include "QF/sys.h"
[scene] Rename libQFentity to libQFscene And start working on scene management.
2021-07-24 05:19:52 +00:00
#include "QF/scene/entity.h"
This is changing more then I wanted to at once, however. The major change is that we no longer require libGL to even exist on the system at compile time for the GL targets, we dynamicly link to the libGL of choice at run time. (This probably breaks most non-linux systems, and all GL targets except -glx, some fixup will be needed.) (This also kills glquake, dead dead DEAD! GONE FOREVER! WHEE!) Some gl_draw cleanup. Commented out equake alias model occlusion test stuff, very experimental. Added the .lo and .la patterns to the .gitignore files. Some minor sbar cleanup. (We don't use the disc in use symbol for anything.)
2001-06-24 09:25:55 +00:00
#include "QF/GL/defines.h"
#include "QF/GL/funcs.h"
Split lightmaps off from surface drawing.
2002-07-23 19:57:47 +00:00
#include "QF/GL/qf_lightmap.h"
This is changing more then I wanted to at once, however. The major change is that we no longer require libGL to even exist on the system at compile time for the GL targets, we dynamicly link to the libGL of choice at run time. (This probably breaks most non-linux systems, and all GL targets except -glx, some fixup will be needed.) (This also kills glquake, dead dead DEAD! GONE FOREVER! WHEE!) Some gl_draw cleanup. Commented out equake alias model occlusion test stuff, very experimental. Added the .lo and .la patterns to the .gitignore files. Some minor sbar cleanup. (We don't use the disc in use symbol for anything.)
2001-06-24 09:25:55 +00:00
#include "QF/GL/qf_rmain.h"
big protototype cleanup. Now, except for a few cases, all non-static prototypes are in headers files.
2002-11-05 19:12:51 +00:00
#include "QF/GL/qf_rsurf.h"
This is changing more then I wanted to at once, however. The major change is that we no longer require libGL to even exist on the system at compile time for the GL targets, we dynamicly link to the libGL of choice at run time. (This probably breaks most non-linux systems, and all GL targets except -glx, some fixup will be needed.) (This also kills glquake, dead dead DEAD! GONE FOREVER! WHEE!) Some gl_draw cleanup. Commented out equake alias model occlusion test stuff, very experimental. Added the .lo and .la patterns to the .gitignore files. Some minor sbar cleanup. (We don't use the disc in use symbol for anything.)
2001-06-24 09:25:55 +00:00
#include "QF/GL/qf_sky.h"
Header and variable declaration cleanups, tiny bit of whitespace.
2001-09-01 08:57:04 +00:00
#include "QF/GL/qf_textures.h"
#include "QF/GL/qf_vid.h"
This is changing more then I wanted to at once, however. The major change is that we no longer require libGL to even exist on the system at compile time for the GL targets, we dynamicly link to the libGL of choice at run time. (This probably breaks most non-linux systems, and all GL targets except -glx, some fixup will be needed.) (This also kills glquake, dead dead DEAD! GONE FOREVER! WHEE!) Some gl_draw cleanup. Commented out equake alias model occlusion test stuff, very experimental. Added the .lo and .la patterns to the .gitignore files. Some minor sbar cleanup. (We don't use the disc in use symbol for anything.)
2001-06-24 09:25:55 +00:00
(Optionally) Use glTexSubImage2D for lightmap updates. Doesn't give as large a boost as I'd hoped, but will matter far more as other optimizations happen. New cvars gl_lightmap_align does nothing yet, and gl_lightmap_subimage only works in mode 0 and 1. Mode 2 (should be fastest mode on most cards) will eventully get implemented...
2001-08-02 02:18:04 +00:00
#include "compat.h"
Take the first step towards render plugins. No clients link. Even if they did, nothing would work.
2012-02-14 08:28:09 +00:00
#include "r_internal.h"
[renderer] Move frame rendering out a layer. This is a step towards high-level unification of the renderers, as far as possible keeping only actual low-level implementation details in the individual renderers (some higher level stuff, eg shadows, is expected to be per-renderer as some things are just not feasible to implement in all renderers). However, the idea is to move the high-level functionality into scene rendering.
2022-03-07 14:32:44 +00:00
#include "vid_gl.h"
(Optionally) Use glTexSubImage2D for lightmap updates. Doesn't give as large a boost as I'd hoped, but will matter far more as other optimizations happen. New cvars gl_lightmap_align does nothing yet, and gl_lightmap_subimage only works in mode 0 and 1. Mode 2 (should be fastest mode on most cards) will eventully get implemented...
2001-08-02 02:18:04 +00:00
Clean up global symbols for the gl renderer. Where possible, symbols have been made static, prefixed with gl_/GL_ or moved into the code shared by all renderers. This will make doing plugins easier but done now for link testing.
2012-02-17 09:33:07 +00:00
static instsurf_t *waterchain = NULL;
static instsurf_t **waterchain_tail = &waterchain;
static instsurf_t *sky_chain;
static instsurf_t **sky_chain_tail;
build the texture chains such that they get rendered front to back. seems to only get about 0.17% speedup in overkill (with just about every thing off), but it might make more of a difference in-game.
2001-11-05 06:20:35 +00:00
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
typedef struct glbspctx_s {
mod_brush_t *brush;
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
animation_t *animation;
vec4f_t *transform;
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
float *color;
} glbspctx_t;
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
#define CHAIN_SURF_F2B(surf,chain) \
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
({ \
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
instsurf_t *inst = (surf)->instsurf; \
Make gl_rsurf.c compile without warnings with "gcc (GCC) 4.4.5 20101112 (Red Hat 4.4.5-2)".
2012-01-19 19:44:26 +00:00
if (__builtin_expect(!inst, 1)) \
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
inst = get_instsurf (); \
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
inst->surface = (surf); \
*(chain##_tail) = inst; \
(chain##_tail) = &inst->tex_chain; \
*(chain##_tail) = 0; \
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
inst; \
})
build the texture chains such that they get rendered front to back. seems to only get about 0.17% speedup in overkill (with just about every thing off), but it might make more of a difference in-game.
2001-11-05 06:20:35 +00:00
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
#define CHAIN_SURF_B2F(surf,chain) \
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
({ \
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
instsurf_t *inst = (surf)->instsurf; \
Make gl_rsurf.c compile without warnings with "gcc (GCC) 4.4.5 20101112 (Red Hat 4.4.5-2)".
2012-01-19 19:44:26 +00:00
if (__builtin_expect(!inst, 1)) \
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
inst = get_instsurf (); \
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
inst->surface = (surf); \
inst->tex_chain = (chain); \
(chain) = inst; \
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
inst; \
})
build the texture chains such that they get rendered front to back. seems to only get about 0.17% speedup in overkill (with just about every thing off), but it might make more of a difference in-game.
2001-11-05 06:20:35 +00:00
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
static gltex_t **r_texture_chains;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
static int r_num_texture_chains;
static int max_texture_chains;
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
static instsurf_t *static_chains;
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
static instsurf_t *free_instsurfs;
static instsurf_t *alloced_instsurfs;
static instsurf_t **alloced_instsurfs_tail = &alloced_instsurfs;
#define NUM_INSTSURFS (64 * 6) // most brush models are simple boxes.
static inline instsurf_t *
get_instsurf (void)
{
instsurf_t *instsurf;
if (!free_instsurfs) {
int i;
free_instsurfs = calloc (NUM_INSTSURFS, sizeof (instsurf_t));
for (i = 0; i < NUM_INSTSURFS - 1; i++)
Make a start on bsp rendering. Still nothing being rendered: still in the process of building the display lists, but I'm making good progress. Get this into git before something goes wrong :)
2012-01-08 01:29:38 +00:00
free_instsurfs[i]._next = &free_instsurfs[i + 1];
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
}
instsurf = free_instsurfs;
Make a start on bsp rendering. Still nothing being rendered: still in the process of building the display lists, but I'm making good progress. Get this into git before something goes wrong :)
2012-01-08 01:29:38 +00:00
free_instsurfs = instsurf->_next;
instsurf->_next = 0;
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
//build the chain of allocated instance surfaces so they can all be freed
//in one go
*alloced_instsurfs_tail = instsurf;
Make a start on bsp rendering. Still nothing being rendered: still in the process of building the display lists, but I'm making good progress. Get this into git before something goes wrong :)
2012-01-08 01:29:38 +00:00
alloced_instsurfs_tail = &instsurf->_next;
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
return instsurf;
}
static inline void
release_instsurfs (void)
{
if (alloced_instsurfs) {
*alloced_instsurfs_tail = free_instsurfs;
free_instsurfs = alloced_instsurfs;
alloced_instsurfs = 0;
alloced_instsurfs_tail = &alloced_instsurfs;
}
}
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
void
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
gl_R_AddTexture (texture_t *tx)
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
{
int i;
if (r_num_texture_chains == max_texture_chains) {
max_texture_chains += 64;
r_texture_chains = realloc (r_texture_chains,
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
max_texture_chains * sizeof (gltex_t *));
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
for (i = r_num_texture_chains; i < max_texture_chains; i++)
r_texture_chains[i] = 0;
}
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
gltex_t *tex = tx->render;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
r_texture_chains[r_num_texture_chains++] = tex;
tex->tex_chain = NULL;
tex->tex_chain_tail = &tex->tex_chain;
}
void
[models] Move brush data into its own struct This is a big step towards a cleaner api. The struct reference in model_t really should be a pointer, but bsp submodel(?) loading messed that up, though that's just a matter of taking more care in the loading code. It seems sensible to make that a separate step.
2021-02-01 10:31:11 +00:00
gl_R_InitSurfaceChains (mod_brush_t *brush)
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
{
if (static_chains)
free (static_chains);
[models] Move brush data into its own struct This is a big step towards a cleaner api. The struct reference in model_t really should be a pointer, but bsp submodel(?) loading messed that up, though that's just a matter of taking more care in the loading code. It seems sensible to make that a separate step.
2021-02-01 10:31:11 +00:00
static_chains = calloc (brush->nummodelsurfaces, sizeof (instsurf_t));
[util] Make bsp_t counts size_t and other bsp data counts unsigned, and clean up the resulting mess.
2021-08-01 12:54:05 +00:00
for (unsigned i = 0; i < brush->nummodelsurfaces; i++)
[models] Move brush data into its own struct This is a big step towards a cleaner api. The struct reference in model_t really should be a pointer, but bsp submodel(?) loading messed that up, though that's just a matter of taking more care in the loading code. It seems sensible to make that a separate step.
2021-02-01 10:31:11 +00:00
brush->surfaces[i].instsurf = static_chains + i;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
release_instsurfs ();
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
}
void
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_R_ClearTextures (void)
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
{
r_num_texture_chains = 0;
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Colormod & alpha for brushmodels, excluding skypoly emittors.
2002-07-07 06:15:36 +00:00
// BRUSH MODELS ===============================================================
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Minor rendering cleanup, also may help rendering speed when r_wateralpha = 1, by playing with draw order.
2001-11-20 09:41:15 +00:00
static void
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
R_RenderFullbrights (void)
{
Header and variable declaration cleanups, tiny bit of whitespace.
2001-09-01 08:57:04 +00:00
float *v;
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
int i, j;
glpoly_t *p;
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
instsurf_t *sc;
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
gltex_t *tex;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
for (i = 0; i < r_num_texture_chains; i++) {
if (!(tex = r_texture_chains[i]) || !tex->gl_fb_texturenum)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
continue;
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
qfglBindTexture (GL_TEXTURE_2D, tex->gl_fb_texturenum);
for (sc = tex->tex_chain; sc; sc = sc->tex_chain) {
if (sc->transform) {
qfglPushMatrix ();
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
qfglLoadMatrixf ((vec_t*)&sc->transform[0]);//FIXME
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
}
if (sc->color)
qfglColor4fv (sc->color);
for (p = sc->surface->polys; p; p = p->next) {
qfglBegin (GL_POLYGON);
for (j = 0, v = p->verts[0]; j < p->numverts;
j++, v += VERTEXSIZE)
{
qfglTexCoord2fv (&v[3]);
qfglVertex3fv (v);
}
qfglEnd ();
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
if (sc->transform)
qfglPopMatrix ();
if (sc->color)
qfglColor3ubv (color_white);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
}
}
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
static inline void
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
R_RenderBrushPoly_3 (msurface_t *surf)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
float *v;
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
int i;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[renderer] Move frame rendering out a layer. This is a step towards high-level unification of the renderers, as far as possible keeping only actual low-level implementation details in the individual renderers (some higher level stuff, eg shadows, is expected to be per-renderer as some things are just not feasible to implement in all renderers). However, the idea is to move the high-level functionality into scene rendering.
2022-03-07 14:32:44 +00:00
gl_ctx->brush_polys++;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Clean up and repair of the run time linking, does NOT work until deek commits his gl_funcs.c.
2001-06-26 02:26:46 +00:00
qfglBegin (GL_POLYGON);
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
v = surf->polys->verts[0];
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
for (i = 0; i < surf->polys->numverts; i++, v += VERTEXSIZE) {
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qglMultiTexCoord2fv (gl_mtex_enum + 0, &v[3]);
qglMultiTexCoord2fv (gl_mtex_enum + 1, &v[5]);
qglMultiTexCoord2fv (gl_mtex_enum + 2, &v[3]);
qfglVertex3fv (v);
}
qfglEnd ();
}
static inline void
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
R_RenderBrushPoly_2 (msurface_t *surf)
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
{
float *v;
int i;
[renderer] Move frame rendering out a layer. This is a step towards high-level unification of the renderers, as far as possible keeping only actual low-level implementation details in the individual renderers (some higher level stuff, eg shadows, is expected to be per-renderer as some things are just not feasible to implement in all renderers). However, the idea is to move the high-level functionality into scene rendering.
2022-03-07 14:32:44 +00:00
gl_ctx->brush_polys++;
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qfglBegin (GL_POLYGON);
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
v = surf->polys->verts[0];
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
for (i = 0; i < surf->polys->numverts; i++, v += VERTEXSIZE) {
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qglMultiTexCoord2fv (gl_mtex_enum + 0, &v[3]);
qglMultiTexCoord2fv (gl_mtex_enum + 1, &v[5]);
qfglVertex3fv (v);
}
qfglEnd ();
}
static inline void
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
R_RenderBrushPoly_1 (msurface_t *surf)
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
{
float *v;
int i;
[renderer] Move frame rendering out a layer. This is a step towards high-level unification of the renderers, as far as possible keeping only actual low-level implementation details in the individual renderers (some higher level stuff, eg shadows, is expected to be per-renderer as some things are just not feasible to implement in all renderers). However, the idea is to move the high-level functionality into scene rendering.
2022-03-07 14:32:44 +00:00
gl_ctx->brush_polys++;
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qfglBegin (GL_POLYGON);
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
v = surf->polys->verts[0];
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
for (i = 0; i < surf->polys->numverts; i++, v += VERTEXSIZE) {
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qfglTexCoord2fv (&v[3]);
qfglVertex3fv (v);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
Clean up and repair of the run time linking, does NOT work until deek commits his gl_funcs.c.
2001-06-26 02:26:46 +00:00
qfglEnd ();
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
}
static inline void
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
R_AddToLightmapChain (glbspctx_t *bctx, msurface_t *surf, instsurf_t *sc)
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
{
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
int maps;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// add the poly to the proper lightmap chain
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
sc->lm_chain = gl_lightmap_polys;
gl_lightmap_polys = sc;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// check for lightmap modification
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
for (maps = 0; maps < MAXLIGHTMAPS && surf->styles[maps] != 255; maps++)
if (d_lightstylevalue[surf->styles[maps]] != surf->cached_light[maps])
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
goto dynamic;
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
if ((surf->dlightframe == r_framecount) || surf->cached_dlight) {
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
dynamic:
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
if (r_dynamic) {
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
gl_R_BuildLightMap (bctx->transform, bctx->brush, surf);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
}
}
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
void
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_R_DrawWaterSurfaces (void)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
int i;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
instsurf_t *s;
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
msurface_t *surf;
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
float wateralpha = max (vr_data.min_wateralpha, r_wateralpha);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
if (!waterchain)
return;
Take the first step towards render plugins. No clients link. Even if they did, nothing would work.
2012-02-14 08:28:09 +00:00
if (wateralpha < 1.0) {
Clean up and repair of the run time linking, does NOT work until deek commits his gl_funcs.c.
2001-06-26 02:26:46 +00:00
qfglDepthMask (GL_FALSE);
Take the first step towards render plugins. No clients link. Even if they did, nothing would work.
2012-02-14 08:28:09 +00:00
color_white[3] = wateralpha * 255;
glColor cleanup
2001-08-30 18:24:19 +00:00
qfglColor4ubv (color_white);
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
i = -1;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
for (s = waterchain; s; s = s->tex_chain) {
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
gltex_t *tex;
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
surf = s->surface;
[renderer] Remove namehack.h Finally. I never liked it (felt bad adding it in the first place), and it has caused confusion with function and global variable names, but it did let me get the render plugins working.
2022-03-14 05:31:23 +00:00
if (s->transform) {
qfglPushMatrix ();
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
qfglLoadMatrixf ((vec_t*)&s->transform[0]);//FIXME
[renderer] Remove namehack.h Finally. I never liked it (felt bad adding it in the first place), and it has caused confusion with function and global variable names, but it did let me get the render plugins working.
2022-03-14 05:31:23 +00:00
}
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
tex = surf->texinfo->texture->render;
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
if (i != tex->gl_texturenum) {
i = tex->gl_texturenum;
Clean up and repair of the run time linking, does NOT work until deek commits his gl_funcs.c.
2001-06-26 02:26:46 +00:00
qfglBindTexture (GL_TEXTURE_2D, i);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
GL_EmitWaterPolys (surf);
[renderer] Remove namehack.h Finally. I never liked it (felt bad adding it in the first place), and it has caused confusion with function and global variable names, but it did let me get the render plugins working.
2022-03-14 05:31:23 +00:00
if (s->transform) {
qfglPopMatrix ();
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
waterchain = NULL;
build the texture chains such that they get rendered front to back. seems to only get about 0.17% speedup in overkill (with just about every thing off), but it might make more of a difference in-game.
2001-11-05 06:20:35 +00:00
waterchain_tail = &waterchain;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Take the first step towards render plugins. No clients link. Even if they did, nothing would work.
2012-02-14 08:28:09 +00:00
if (wateralpha < 1.0) {
Clean up and repair of the run time linking, does NOT work until deek commits his gl_funcs.c.
2001-06-26 02:26:46 +00:00
qfglDepthMask (GL_TRUE);
glColor cleanup
2001-08-30 18:24:19 +00:00
qfglColor3ubv (color_white);
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
Make gl_rsurf.c compile without warnings with "gcc (GCC) 4.4.5 20101112 (Red Hat 4.4.5-2)".
2012-01-19 19:44:26 +00:00
static void
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
DrawTextureChains (int disable_blend, int do_bind)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
int i;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
instsurf_t *s;
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
msurface_t *surf;
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
gltex_t *tex;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Optimize fullbright testing for brushes, assorted other cleanups, test for functions built into current GL drivers, as well as the old extensions, and add incorrectly rendering BGRA support. Whee!
2004-02-24 20:50:55 +00:00
if (gl_mtex_active_tmus >= 2) {
QF, now complete with underbrights, thanks to Grievre. Play with gl_overbright and gl_doublebright. That's an order.
2004-02-15 00:02:04 +00:00
// Lightmaps
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
qglActiveTexture (gl_mtex_enum + 1);
qfglEnable (GL_TEXTURE_2D);
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
qfglBindTexture (GL_TEXTURE_2D, gl_R_LightmapTexture ());
QF, now complete with underbrights, thanks to Grievre. Play with gl_overbright and gl_doublebright. That's an order.
2004-02-15 00:02:04 +00:00
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
// Base Texture
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
qglActiveTexture (gl_mtex_enum + 0);
qfglTexEnvf (GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
for (i = 0; i < r_num_texture_chains; i++) {
tex = r_texture_chains[i];
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
if (!tex)
continue;
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
qfglBindTexture (GL_TEXTURE_2D, tex->gl_texturenum);
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
Optimize fullbright testing for brushes, assorted other cleanups, test for functions built into current GL drivers, as well as the old extensions, and add incorrectly rendering BGRA support. Whee!
2004-02-24 20:50:55 +00:00
if (tex->gl_fb_texturenum && gl_mtex_fullbright) {
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
qglActiveTexture (gl_mtex_enum + 2);
qfglEnable (GL_TEXTURE_2D);
qfglBindTexture (GL_TEXTURE_2D, tex->gl_fb_texturenum);
QF, now complete with underbrights, thanks to Grievre. Play with gl_overbright and gl_doublebright. That's an order.
2004-02-15 00:02:04 +00:00
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
for (s = tex->tex_chain; s; s = s->tex_chain) {
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
surf = s->surface;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
if (s->transform) {
qfglPushMatrix ();
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
qfglLoadMatrixf ((vec_t*)&s->transform[0]);//FIXME
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
}
if (s->color && do_bind)
qfglColor4fv (s->color);
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
R_RenderBrushPoly_3 (surf);
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
if (s->transform)
qfglPopMatrix ();
if (s->color && do_bind)
qfglColor3ubv (color_white);
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
}
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
qglActiveTexture (gl_mtex_enum + 2);
qfglDisable (GL_TEXTURE_2D);
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qglActiveTexture (gl_mtex_enum + 0);
} else {
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
for (s = tex->tex_chain; s; s = s->tex_chain) {
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
surf = s->surface;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
if (s->transform) {
qfglPushMatrix ();
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
qfglLoadMatrixf ((vec_t*)&s->transform[0]);//FIXME
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
}
if (s->color && do_bind)
qfglColor4fv (s->color);
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
R_RenderBrushPoly_2 (surf);
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
if (s->transform)
qfglPopMatrix ();
if (s->color && do_bind)
qfglColor3ubv (color_white);
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
}
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
qglActiveTexture (gl_mtex_enum + 0);
}
}
QF, now complete with underbrights, thanks to Grievre. Play with gl_overbright and gl_doublebright. That's an order.
2004-02-15 00:02:04 +00:00
// Turn off lightmaps for other entities
qglActiveTexture (gl_mtex_enum + 1);
qfglDisable (GL_TEXTURE_2D);
Fix a bunch of state leaks that made multitextured fullbrights smear the world. Now just have one (known) issue with bmodels with fullbright textures not rendering their base texture. Floating fullbrights are nifty! Also a number of optimizations, which need wider performance testing.
2004-02-21 05:36:19 +00:00
// Reset mode for default TMU
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
qglActiveTexture (gl_mtex_enum + 0);
qfglTexEnvf (GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE);
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
} else {
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
if (disable_blend)
qfglDisable (GL_BLEND);
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
for (i = 0; i < r_num_texture_chains; i++) {
tex = r_texture_chains[i];
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
if (!tex)
continue;
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
if (do_bind)
qfglBindTexture (GL_TEXTURE_2D, tex->gl_texturenum);
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
for (s = tex->tex_chain; s; s = s->tex_chain) {
if (s->transform) {
qfglPushMatrix ();
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
qfglLoadMatrixf ((vec_t*)&s->transform[0]);//FIXME
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
}
R_RenderBrushPoly_1 (s->surface);
if (s->transform)
qfglPopMatrix ();
if (s->color && do_bind)
qfglColor3ubv (color_white);
}
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
}
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
if (disable_blend)
qfglEnable (GL_BLEND);
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
}
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
}
Reorder a lot of code, removing tons of ifs. multitexture paths are a bit more sane now. Also optimize QSG2 color handling, and fix some QSG2 bugs. Not that anyone cares, nobody uses it anyways. Should be faster, but isn't here. *grmbl*
2004-02-22 06:00:15 +00:00
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
static void
clear_texture_chains (void)
{
int i;
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
gltex_t *tex;
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
Create a list of texture chains. This moves the texture chains out of the world model and into a separate list, allowing other brush models to registter their textures.
2011-12-15 06:07:01 +00:00
for (i = 0; i < r_num_texture_chains; i++) {
tex = r_texture_chains[i];
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
if (!tex)
continue;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
tex->tex_chain = NULL;
tex->tex_chain_tail = &tex->tex_chain;
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
}
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
tex = r_notexture_mip->render;
Move the surface chain links out of the surfaces. The links are now in "instance surfaces". For non-instanced models (world, doors, plats etc (ie, world and its sub-models)), there will be one instance surface per model surface. However, for instanced models (ammo boxes etc), there will be many, dynamically allocated (not yet implemented). This commit gets the static instance surfaces working.
2011-12-17 10:14:14 +00:00
tex->tex_chain = NULL;
tex->tex_chain_tail = &tex->tex_chain;
Implement the instance surface allocation/freeing. Not actually used yet, so it's harmless at this stage.
2011-12-17 10:42:19 +00:00
release_instsurfs ();
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
gl_lightmap_polys = 0;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
static inline void
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
chain_surface (glbspctx_t *bctx, msurface_t *surf)
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
{
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
instsurf_t *sc;
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
if (surf->flags & SURF_DRAWTURB) {
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
sc = CHAIN_SURF_B2F (surf, waterchain);
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
} else if (surf->flags & SURF_DRAWSKY) {
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
sc = CHAIN_SURF_F2B (surf, sky_chain);
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
} else {
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
texture_t *tx;
gltex_t *tex;
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
if (!surf->texinfo->texture->anim_total)
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
tx = surf->texinfo->texture;
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
else
[sw] Take advantage of the ECS for edge rendering This fixes the segfault due to the world entity not actually existing, without adding a world entity. It takes advantage of the ECS in that the edge renderer needs only the world matrix, brush model pointer, and the animation frame number (which is just 0/1 for brush models), thus the inherent SOA of ECS helps out, though benchmarking is needed to see if it made any real difference. With this, all 4 renderers are working again.
2022-10-25 10:36:09 +00:00
tx = R_TextureAnimation (bctx->animation->frame, surf);
[model] Isolate renderer specific texture data This cleans up texture_t and possibly even improves locality of reference when running through texture chains (not profiled, and not actually the goal).
2021-01-19 04:05:39 +00:00
tex = tx->render;
[renderer] Remove tinst from msurface_t Its sole purpose was to pass the newly allocated instsurf when chaining an instance model (ammo box, etc) surface, but using expresion statements removes the need for such shenanigans, and even makes msurface_t that little bit smaller (though a separate array would be much better for cache coherence). More importantly, the relevant code is actually easier to understand: I spent way too long working out what tinst was for and why it was never cleared.
2021-07-13 07:02:47 +00:00
sc = CHAIN_SURF_F2B (surf, tex->tex_chain);
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
R_AddToLightmapChain (bctx, surf, sc);
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
}
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
sc->transform = bctx->transform;
sc->color = bctx->color;
Split out chain_surface() from visit_node().
2011-12-15 04:49:53 +00:00
}
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
void
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
gl_R_DrawBrushModel (entity_t e)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
Use R_CullSphere for rotated bsp models.
2002-08-23 21:14:43 +00:00
float dot, radius;
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
transform_t transform = Entity_Transform (e);
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
msurface_t *surf;
Nuke qboolean from orbit I never liked it, but with C2x coming out, it's best to handle bools properly. I haven't gone through all the uses of int as bool (I'll leave that for fixing when I encounter them), but this gets QF working with both c2x (really, gnu2x because of raw strings).
2023-06-13 09:06:11 +00:00
bool rotated;
Header and variable declaration cleanups, tiny bit of whitespace.
2001-09-01 08:57:04 +00:00
vec3_t mins, maxs;
[renderer] Clean up entity_t to a certain extent This is the first step towards component-based entities. There's still some transform-related stuff in the struct that needs to be moved, but it's all entirely client related (rather than renderer) and will probably go into a "client" component. Also, the current components are directly included structs rather than references as I didn't want to deal with the object management at this stage. As part of the process (because transforms use simd) this also starts the process of moving QF to using simd for vectors and matrices. There's now a mess of simd and sisd code mixed together, but it works surprisingly well together.
2021-03-09 14:52:40 +00:00
mat4f_t worldMatrix;
[renderer] Tidy up entity component access Since entity_t has a pointer to the registry owning the entity, there's no need to access a global to get at the registry. Also move component getting closer to where it's used.
2022-10-25 03:53:30 +00:00
renderer_t *renderer = Ent_GetComponent (e.id, scene_renderer, e.reg);
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
model_t *model = renderer->model;
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
mod_brush_t *brush = &model->brush;
glbspctx_t bspctx = {
brush,
[renderer] Tidy up entity component access Since entity_t has a pointer to the registry owning the entity, there's no need to access a global to get at the registry. Also move component getting closer to where it's used.
2022-10-25 03:53:30 +00:00
Ent_GetComponent (e.id, scene_animation, e.reg),
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
renderer->full_transform,
renderer->colormod,
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
};
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
Transform_GetWorldMatrix (transform, worldMatrix);
[renderer] Clean up entity_t to a certain extent This is the first step towards component-based entities. There's still some transform-related stuff in the struct that needs to be moved, but it's all entirely client related (rather than renderer) and will probably go into a "client" component. Also, the current components are directly included structs rather than references as I didn't want to deal with the object management at this stage. As part of the process (because transforms use simd) this also starts the process of moving QF to using simd for vectors and matrices. There's now a mess of simd and sisd code mixed together, but it works surprisingly well together.
2021-03-09 14:52:40 +00:00
if (worldMatrix[0][0] != 1 || worldMatrix[1][1] != 1
|| worldMatrix[2][2] != 1) {
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
rotated = true;
Use R_CullSphere for rotated bsp models.
2002-08-23 21:14:43 +00:00
radius = model->radius;
#if 0 //QSG FIXME
if (e->scale != 1.0)
radius *= e->scale;
#endif
Update vec3_t/vec4f_t hacks to work with clang Still work with gcc, of course, and I still need to fix them properly, but now they're actually slightly easier to find as they all have vec_t and FIXME on the same line.
2022-03-30 15:07:20 +00:00
if (R_CullSphere (r_refdef.frustum, (vec_t*)&worldMatrix[3], radius)) {//FIXME
Use R_CullSphere for rotated bsp models.
2002-08-23 21:14:43 +00:00
return;
[renderer] Remove more old fields from entity_t The only transform related field remaining is old_origin. This also brings the renderer closer to using simd (lots of stuff to fix still, though).
2021-03-19 11:18:45 +00:00
}
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
} else {
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
rotated = false;
[renderer] Remove more old fields from entity_t The only transform related field remaining is old_origin. This also brings the renderer closer to using simd (lots of stuff to fix still, though).
2021-03-19 11:18:45 +00:00
VectorAdd (worldMatrix[3], model->mins, mins);
VectorAdd (worldMatrix[3], model->maxs, maxs);
Colormod & alpha for brushmodels, excluding skypoly emittors.
2002-07-07 06:15:36 +00:00
#if 0 // QSG FIXME
Use R_CullSphere for rotated bsp models.
2002-08-23 21:14:43 +00:00
if (e->scale != 1.0) {
VectorScale (mins, e->scale, mins);
VectorScale (maxs, e->scale, maxs);
}
Colormod & alpha for brushmodels, excluding skypoly emittors.
2002-07-07 06:15:36 +00:00
#endif
[mathlib] Remove frustum global It should never have been there and is now in the refdef (not its final home: it should probably be part of the camera).
2022-03-19 03:33:12 +00:00
if (R_CullBox (r_refdef.frustum, mins, maxs))
Use R_CullSphere for rotated bsp models.
2002-08-23 21:14:43 +00:00
return;
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[renderer] Clean up use of vup/vright/vpn This moves the common camera setup code out of the individual drivers, and completely removes vup/vright/vpn from the non-software renderers. This has highlighted the craziness around AngleVectors with it putting +X forward, -Y right and +Z up. The main issue with this is it requires a 90 degree pre-rotation about the Z axis to get the camera pointing in the right direction, and that's for the native sw renderer (vulkan needs a 90 degree pre-rotation about X, and gl and glsl need to invert an axis, too), though at least it's just a matrix swizzle and vector negation. However, it does mean the camera matrices can't be used directly. Also rename vpn to vfwd (still abbreviated, but fwd is much clearer in meaning (to me, at least) than pn (plane normal, I guess, but which way?)).
2022-03-14 00:34:24 +00:00
vec4f_t relviewpos = r_refdef.frame.position - worldMatrix[3];
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
if (rotated) {
[renderer] Clean up r_origin and modelorg So far, in gl and glsl, but viewposition is much clearer than r_origin (despite being the same thing), and modelorg is just confusing (I think it's the view position relative to the current model).
2022-03-07 18:47:36 +00:00
vec4f_t temp = relviewpos;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[renderer] Clean up r_origin and modelorg So far, in gl and glsl, but viewposition is much clearer than r_origin (despite being the same thing), and modelorg is just confusing (I think it's the view position relative to the current model).
2022-03-07 18:47:36 +00:00
relviewpos[0] = dotf (temp, worldMatrix[0])[0];
relviewpos[1] = dotf (temp, worldMatrix[1])[0];
relviewpos[2] = dotf (temp, worldMatrix[2])[0];
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
// calculate dynamic lighting for bmodel if it's not an instanced model
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
if (brush->firstmodelsurface != 0 && r_dlight_lightmap) {
[util] Make bsp_t counts size_t and other bsp data counts unsigned, and clean up the resulting mess.
2021-08-01 12:54:05 +00:00
for (unsigned k = 0; k < r_maxdlights; k++) {
Take the first step towards render plugins. No clients link. Even if they did, nothing would work.
2012-02-14 08:28:09 +00:00
if ((r_dlights[k].die < vr_data.realtime)
|| (!r_dlights[k].radius))
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
continue;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
vec4f_t lightorigin;
[renderer] Remove more old fields from entity_t The only transform related field remaining is old_origin. This also brings the renderer closer to using simd (lots of stuff to fix still, though).
2021-03-19 11:18:45 +00:00
VectorSubtract (r_dlights[k].origin, worldMatrix[3], lightorigin);
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
lightorigin[3] = 1;
[models] Move brush data into its own struct This is a big step towards a cleaner api. The struct reference in model_t really should be a pointer, but bsp submodel(?) loading messed that up, though that's just a matter of taking more care in the loading code. It seems sensible to make that a separate step.
2021-02-01 10:31:11 +00:00
R_RecursiveMarkLights (brush, lightorigin, &r_dlights[k], k,
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
brush->hulls[0].firstclipnode);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
}
Clean up and repair of the run time linking, does NOT work until deek commits his gl_funcs.c.
2001-06-26 02:26:46 +00:00
qfglPushMatrix ();
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
gl_R_RotateForEntity (Transform_GetWorldMatrixPtr (transform));
qfglGetFloatv (GL_MODELVIEW_MATRIX, (vec_t*)&renderer->full_transform[0]);
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
qfglPopMatrix ();
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
surf = &brush->surfaces[brush->firstmodelsurface];
Grievre's lightmap/fullbright multitexture patch (fbs not tested due to lack of drivers supporting 3 tmus)
2004-02-14 05:10:29 +00:00
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// draw texture
[util] Make bsp_t counts size_t and other bsp data counts unsigned, and clean up the resulting mess.
2021-08-01 12:54:05 +00:00
for (unsigned i = 0; i < brush->nummodelsurfaces; i++, surf++) {
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// find which side of the node we are on
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
plane_t *plane = surf->plane;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[renderer] Clean up r_origin and modelorg So far, in gl and glsl, but viewposition is much clearer than r_origin (despite being the same thing), and modelorg is just confusing (I think it's the view position relative to the current model).
2022-03-07 18:47:36 +00:00
dot = DotProduct (relviewpos, plane->normal) - plane->dist;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// draw the polygon
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
if (((surf->flags & SURF_PLANEBACK) && (dot < -BACKFACE_EPSILON)) ||
(!(surf->flags & SURF_PLANEBACK) && (dot > BACKFACE_EPSILON))) {
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
chain_surface (&bspctx, surf);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
}
}
Colormod & alpha for brushmodels, excluding skypoly emittors.
2002-07-07 06:15:36 +00:00
// WORLD MODEL ================================================================
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
static inline void
visit_leaf (mleaf_t *leaf)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
// deal with model fragments in this leaf
if (leaf->efrags)
Make R_StoreEfrags const-correct. R_StoreEfrags does not need to modify its parameter or anything to which it directly refers, so make the pointer single-level and constant.
2010-12-03 04:28:58 +00:00
R_StoreEfrags (leaf->efrags);
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
static inline int
get_side (mnode_t *node)
{
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
// find which side of the node we are on
[renderer] Clean up use of vup/vright/vpn This moves the common camera setup code out of the individual drivers, and completely removes vup/vright/vpn from the non-software renderers. This has highlighted the craziness around AngleVectors with it putting +X forward, -Y right and +Z up. The main issue with this is it requires a 90 degree pre-rotation about the Z axis to get the camera pointing in the right direction, and that's for the native sw renderer (vulkan needs a 90 degree pre-rotation about X, and gl and glsl need to invert an axis, too), though at least it's just a matrix swizzle and vector negation. However, it does mean the camera matrices can't be used directly. Also rename vpn to vfwd (still abbreviated, but fwd is much clearer in meaning (to me, at least) than pn (plane normal, I guess, but which way?)).
2022-03-14 00:34:24 +00:00
vec4f_t org = r_refdef.frame.position;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
return dotf (org, node->plane)[0] < 0;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
the inline went wandering by mistake. gives a tiny (< 0.5%) gain, but a gain none the less
2004-02-23 04:46:55 +00:00
static inline void
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
visit_node (glbspctx_t *bctx, mnode_t *node, int side)
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
{
int c;
msurface_t *surf;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
code cleanups and possibly minor optimisations
2001-09-07 21:15:08 +00:00
// sneaky hack for side = side ? SURF_PLANEBACK : 0;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
side = (~side + 1) & SURF_PLANEBACK;
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
// draw stuff
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
if ((c = node->numsurfaces)) {
[model] Move visframe out of msurface_t One more step towards BSP thread-safety. This one brought with it a very noticeable speed boost (ie, not lost in the noise) thanks to the face visframes being in tightly packed groups instead of 128 bytes apart, though the sw render's boost is lost in the noise (but it's very fill-rate limited).
2022-05-22 07:31:24 +00:00
int surf_id = node->firstsurface;
surf = bctx->brush->surfaces + surf_id;
for (; c; c--, surf++, surf_id++) {
if (r_face_visframes[surf_id] != r_visframecount)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
continue;
code cleanups and possibly minor optimisations
2001-09-07 21:15:08 +00:00
// side is either 0 or SURF_PLANEBACK
if (side ^ (surf->flags & SURF_PLANEBACK))
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
continue; // wrong side
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
chain_surface (bctx, surf);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
}
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
}
static inline int
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
test_node (glbspctx_t *bctx, int node_id)
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
{
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
if (node_id < 0)
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
return 0;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
if (r_node_visframes[node_id] != r_visframecount)
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
return 0;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
mnode_t *node = bctx->brush->nodes + node_id;
[mathlib] Remove frustum global It should never have been there and is now in the refdef (not its final home: it should probably be part of the camera).
2022-03-19 03:33:12 +00:00
if (R_CullBox (r_refdef.frustum, node->minmaxs, node->minmaxs + 3))
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
return 0;
return 1;
}
static void
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
R_VisitWorldNodes (glbspctx_t *bctx)
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
{
Remove the hard-coded bsp maximum depths. The depth limits in the gl and glsl renderers and in the trace code really bothered me, but then the fix hit me: at load-time, recurse the trees normally and record the depth in the appropriate place. The node stacks can then be allocated as necessary (I chose to add a paranoia buffer of 2, but I expect the maximum depth will rarely be used).
2012-11-28 12:29:03 +00:00
typedef struct {
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
int node_id;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
int side;
Remove the hard-coded bsp maximum depths. The depth limits in the gl and glsl renderers and in the trace code really bothered me, but then the fix hit me: at load-time, recurse the trees normally and record the depth in the appropriate place. The node stacks can then be allocated as necessary (I chose to add a paranoia buffer of 2, but I expect the maximum depth will rarely be used).
2012-11-28 12:29:03 +00:00
} rstack_t;
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
mod_brush_t *brush = bctx->brush;
Remove the hard-coded bsp maximum depths. The depth limits in the gl and glsl renderers and in the trace code really bothered me, but then the fix hit me: at load-time, recurse the trees normally and record the depth in the appropriate place. The node stacks can then be allocated as necessary (I chose to add a paranoia buffer of 2, but I expect the maximum depth will rarely be used).
2012-11-28 12:29:03 +00:00
rstack_t *node_ptr;
rstack_t *node_stack;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
int node_id;
int front;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
int side;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
node_id = 0;
Remove the hard-coded bsp maximum depths. The depth limits in the gl and glsl renderers and in the trace code really bothered me, but then the fix hit me: at load-time, recurse the trees normally and record the depth in the appropriate place. The node stacks can then be allocated as necessary (I chose to add a paranoia buffer of 2, but I expect the maximum depth will rarely be used).
2012-11-28 12:29:03 +00:00
// +2 for paranoia
[models] Move brush data into its own struct This is a big step towards a cleaner api. The struct reference in model_t really should be a pointer, but bsp submodel(?) loading messed that up, though that's just a matter of taking more care in the loading code. It seems sensible to make that a separate step.
2021-02-01 10:31:11 +00:00
node_stack = alloca ((brush->depth + 2) * sizeof (rstack_t));
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
node_ptr = node_stack;
while (1) {
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
while (test_node (bctx, node_id)) {
mnode_t *node = bctx->brush->nodes + node_id;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
side = get_side (node);
front = node->children[side];
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
if (test_node (bctx, front)) {
node_ptr->node_id = node_id;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
node_ptr->side = side;
node_ptr++;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
node_id = front;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
continue;
}
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
if (front < 0) {
mleaf_t *leaf = bctx->brush->leafs + ~front;
if (leaf->contents != CONTENTS_SOLID) {
visit_leaf (leaf);
}
}
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
visit_node (bctx, node, side);
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
node_id = node->children[side ^ 1];
}
if (node_id < 0) {
mleaf_t *leaf = bctx->brush->leafs + ~node_id;
if (leaf->contents != CONTENTS_SOLID) {
visit_leaf (leaf);
}
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
}
if (node_ptr != node_stack) {
node_ptr--;
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
node_id = node_ptr->node_id;
mnode_t *node = bctx->brush->nodes + node_id;
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
side = node_ptr->side;
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
visit_node (bctx, node, side);
[model] Move plane info into mnode_t, and visframe out The main goal was to get visframe out of mnode_t to make it thread-safe (each thread can have its own visframe array), but moving the plane info into mnode_t made for better data access patters when traversing the bsp tree as the plane is right there with the child indices. Nicely, the size of mnode_t is the same as before (64 bytes due to alignment), with 4 bytes wasted. Performance-wise, there seems to be very little difference. Maybe slightly slower. The unfortunate thing about the change is the plane distance is negated, possibly leading to some confusion, particularly since the box and sphere culling functions were affected. However, this is so point-plane distance calculations can be done with a single 4d dot product.
2022-05-22 02:18:32 +00:00
node_id = node->children[side ^ 1];
make R_RecursiveWorldNode non-recursive (needs a new name, now :). 0.4% speedup in null-renderer gl: unknown for sw and sw32 (could be slightly slower, left recursive version in but #if 0ed out jic)
2004-02-15 03:46:55 +00:00
continue;
}
break;
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
void
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_R_DrawWorld (void)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
animation_t animation = {};
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
glbspctx_t bctx = { };
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
fix skybox/skydom map vis issues. I hope this works on all drivers.
2001-10-03 02:51:30 +00:00
sky_chain = 0;
build the texture chains such that they get rendered front to back. seems to only get about 0.17% speedup in overkill (with just about every thing off), but it might make more of a difference in-game.
2001-11-05 06:20:35 +00:00
sky_chain_tail = &sky_chain;
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
if (!gl_sky_clip) {
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_R_DrawSky ();
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
bctx.brush = &r_refdef.worldmodel->brush;
bctx.animation = &animation;
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
R_VisitWorldNodes (&bctx);
[gl] Draw sky chains before brush entities Drawing sky chains first ensures that sky surfaces correctly block parts of the map that should not be visible (by writing the correct depth to the depth buffer when doing box or dome skies). Writing brush models first means that the models (ammo boxes etc) could be visible when they should not be.
2022-03-06 23:40:39 +00:00
gl_R_DrawSkyChain (sky_chain);
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
if (r_drawentities) {
[renderer] Support multiple entity queues While there's currently only the one still, this will allow the entities to be multiply queued for multi-pass rendering (eg, shadows). As the avoidance of putting an entity in the same queue more than once relies on the entity id, all entities now come from the scene (which is stored in cl_world in the client code for nq and qw), thus the extensive changes in the clients.
2022-03-04 16:48:10 +00:00
for (size_t i = 0; i < r_ent_queue->ent_queues[mod_brush].size; i++) { \
[scene] Make entity_t just an entity id for ECS This puts the hierarchy (transform) reference, animation, visibility, renderer, active, and old_origin data in separate components. There are a few bugs (crashes on grenade explosions in gl/glsl/vulkan, immediately in sw, reasons known, missing brush models in vulkan). While quake doesn't really need an ECS, the direction I want to take QF does, and it does seem to have improved memory bandwidth a little (uncertain). However, there's a lot more work to go (especially fixing the above bugs), but this seems to be a good start.
2022-10-23 01:32:09 +00:00
entity_t ent = r_ent_queue->ent_queues[mod_brush].a[i]; \
[renderer] Remove currententity from non-sw renderers Really needed only for vulkan, but removing currententity from gl and glsl made testing easier.
2021-07-22 06:39:28 +00:00
gl_R_DrawBrushModel (ent);
Merge brush model with world model rendering. With this, non-mtex global fog now works properly :).
2011-12-17 12:45:52 +00:00
}
}
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
gl_R_FlushLightmaps ();
[gl] Call gl_R_CalcLightmaps after sub-model brushes have been queued This fixes the sub-models not being dynamically lit.
2022-05-10 08:04:57 +00:00
[renderer] Merge the fog support code GL still has its own functions for enabling and disabling fog while rendering, but GLSL doesn't need such (thanks to the shaders), nor will vulkan (and the software renderers don't support fog).
2022-03-07 17:10:47 +00:00
if (!Fog_GetDensity ()
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
|| (gl_fb_bmodels && gl_mtex_fullbright)
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
|| gl_mtex_active_tmus > 1) {
// we have enough active TMUs to render everything in one go
// or we're not doing fog
DrawTextureChains (1, 1);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Restore the missing lightmaps in non-mtex mode. I forgot to blend in the lightmaps in the fogless non-mtex path.
2011-12-16 00:21:39 +00:00
if (gl_mtex_active_tmus <= 1)
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_R_BlendLightmaps ();
Restore the missing lightmaps in non-mtex mode. I forgot to blend in the lightmaps in the fogless non-mtex path.
2011-12-16 00:21:39 +00:00
[cvar] Make cvars properly typed This is an extremely extensive patch as it hits every cvar, and every usage of the cvars. Cvars no longer store the value they control, instead, they use a cexpr value object to reference the value and specify the value's type (currently, a null type is used for strings). Non-string cvars are passed through cexpr, allowing expressions in the cvars' settings. Also, cvars have returned to an enhanced version of the original (id quake) registration scheme. As a minor benefit, relevant code having direct access to the cvar-controlled variables is probably a slight optimization as it removed a pointer dereference, and the variables can be located for data locality. The static cvar descriptors are made private as an additional safety layer, though there's nothing stopping external modification via Cvar_FindVar (which is needed for adding listeners). While not used yet (partly due to working out the design), cvars can have a validation function. Registering a cvar allows a primary listener (and its data) to be specified: it will always be called first when the cvar is modified. The combination of proper listeners and direct access to the controlled variable greatly simplifies the more complex cvar interactions as much less null checking is required, and there's no need for one cvar's callback to call another's. nq-x11 is known to work at least well enough for the demos. More testing will come.
2022-04-23 03:22:45 +00:00
if (gl_fb_bmodels && !gl_mtex_fullbright)
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
R_RenderFullbrights ();
} else {
if (gl_mtex_active_tmus > 1) {
// textures and lightmaps in one pass
// black fog
// no blending
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_Fog_StartAdditive ();
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
DrawTextureChains (1, 1);
// buf = fTL + (1-f)0
// = fTL
} else {
// texture pass + lightmap pass
// no fog
// no blending
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_Fog_DisableGFog ();
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
DrawTextureChains (1, 1);
// buf = T
// black fog
// blend: buf = zero, src (non-overbright)
// FIXME overbright broken?
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_Fog_EnableGFog ();
gl_Fog_StartAdditive ();
gl_R_BlendLightmaps (); //leaves blending as As, 1-As
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
// buf = I*0 + buf*I
// = T*C
// = T(fL + (1-f)0)
// = fTL
}
// fullbright pass
// fog is still black
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
R_RenderFullbrights ();
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
// buf = aI + (1-a)buf
// = aC + (1-a)fTL
// = a(fG + (1-f)0) + (1-a)fTL
// = afG + (1-a)fTL
// = f((1-a)TL + aG)
Rename the public symbols for gl, glsl and sw32. Evil hack, but it does the job.
2012-02-22 07:32:34 +00:00
gl_Fog_StopAdditive (); // use fog color
Get global fog working for multi-pass rendering. For the world textures, anyway. Entity multi-pass rendering still needs to be verified.
2011-12-12 05:35:53 +00:00
qfglDepthMask (GL_FALSE); // don't write Z
qfglBlendFunc (GL_ONE, GL_ONE);
// draw black polys
qfglColor4f (0, 0, 0, 1);
DrawTextureChains (0, 0);
// buf = I + buf
// = C + f((1-a)TL + aG)
// = (f0 + (1-f)F) + f((1-a)TL + aG)
// = (1-f)F + f((1-a)TL + aG)
// = f((1-a)TL + aG) + (1-f)F
// restore state
qfglColor4ubv (color_white);
qfglBlendFunc (GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
qfglDepthMask (GL_TRUE);
}
clear_texture_chains ();
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
Clean up global symbols for the gl renderer. Where possible, symbols have been made static, prefixed with gl_/GL_ or moved into the code shared by all renderers. This will make doing plugins easier but done now for link testing.
2012-02-17 09:33:07 +00:00
model_t *gl_currentmodel;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Split lightmaps off from surface drawing.
2002-07-23 19:57:47 +00:00
void
[renderer] Clean up r_bsp a little Move r_pcurrentvertbase into the sw renderer, cleaning up gl's use of (not really needed there). Not ready to move r_bsp into the main bin yet as there are linking issues since only the low-level code references any of its symbols.
2022-03-17 04:09:20 +00:00
GL_BuildSurfaceDisplayList (mod_brush_t *brush, msurface_t *surf)
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
{
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
float s, t;
Header and variable declaration cleanups, tiny bit of whitespace.
2001-09-01 08:57:04 +00:00
float *vec;
Clean up all the "set but not used" warnings. gcc on my system is failing to treat this specific warning as an error :/
2011-06-19 01:48:02 +00:00
int lindex, lnumverts, i;
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
glpoly_t *poly;
Header and variable declaration cleanups, tiny bit of whitespace.
2001-09-01 08:57:04 +00:00
medge_t *pedges, *r_pedge;
[renderer] Clean up r_bsp a little Move r_pcurrentvertbase into the sw renderer, cleaning up gl's use of (not really needed there). Not ready to move r_bsp into the main bin yet as there are linking issues since only the low-level code references any of its symbols.
2022-03-17 04:09:20 +00:00
mvertex_t *vertex_base = brush->vertexes;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
// reconstruct the polygon
[models] Move brush data into its own struct This is a big step towards a cleaner api. The struct reference in model_t really should be a pointer, but bsp submodel(?) loading messed that up, though that's just a matter of taking more care in the loading code. It seems sensible to make that a separate step.
2021-02-01 10:31:11 +00:00
pedges = gl_currentmodel->brush.edges;
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
lnumverts = surf->numedges;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// draw texture
[util] Make hunk (optionally) thread-safe For now, the functions check for a null hunk pointer and use the global hunk (initialized via Memory_Init) if necessary. However, Hunk_Init is available (and used by Memory_Init) to create a hunk from any arbitrary memory block. So long as that block is 64-byte aligned, allocations within the hunk will remain 64-byte aligned.
2021-07-28 06:01:45 +00:00
poly = Hunk_Alloc (0, sizeof (glpoly_t) + (lnumverts - 4) *
Renderer merge continues. gl_rlight, gl_rmisc, gl_rmain, and gl_rsurf the current victims.
2001-05-13 00:28:50 +00:00
VERTEXSIZE * sizeof (float));
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
poly->next = surf->polys;
poly->flags = surf->flags;
surf->polys = poly;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
poly->numverts = lnumverts;
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
mtexinfo_t *texinfo = surf->texinfo;
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
for (i = 0; i < lnumverts; i++) {
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
lindex = gl_currentmodel->brush.surfedges[surf->firstedge + i];
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
if (lindex > 0) {
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
r_pedge = &pedges[lindex];
[renderer] Clean up r_bsp a little Move r_pcurrentvertbase into the sw renderer, cleaning up gl's use of (not really needed there). Not ready to move r_bsp into the main bin yet as there are linking issues since only the low-level code references any of its symbols.
2022-03-17 04:09:20 +00:00
vec = vertex_base[r_pedge->v[0]].position;
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
} else {
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
r_pedge = &pedges[-lindex];
[renderer] Clean up r_bsp a little Move r_pcurrentvertbase into the sw renderer, cleaning up gl's use of (not really needed there). Not ready to move r_bsp into the main bin yet as there are linking issues since only the low-level code references any of its symbols.
2022-03-17 04:09:20 +00:00
vec = vertex_base[r_pedge->v[1]].position;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
s = DotProduct (vec, texinfo->vecs[0]) + texinfo->vecs[0][3];
s /= texinfo->texture->width;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
t = DotProduct (vec, texinfo->vecs[1]) + texinfo->vecs[1][3];
t /= texinfo->texture->height;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
VectorCopy (vec, poly->verts[i]);
poly->verts[i][3] = s;
poly->verts[i][4] = t;
// lightmap texture coordinates
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
s = DotProduct (vec, texinfo->vecs[0]) + texinfo->vecs[0][3];
t = DotProduct (vec, texinfo->vecs[1]) + texinfo->vecs[1][3];
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
s -= surf->texturemins[0];
[renderer] Be more consistent with msurface_t names Use "surf" everywhere in gl, glsl and vulkan. Not worried about sw/sw32 at this stage (I don't poke around in there anywhere near as much).
2021-07-15 12:38:12 +00:00
t -= surf->texturemins[1];
[gl] Use a scrap for lightmaps This gives a rather significant speed boost to timedemo demo1: from about 2300-2360fps up to 2520-2600fps, at least when using multi-texture. Since it was necessary for testing the scrap, gl got the ability to set the console background texture, too.
2022-05-10 15:27:55 +00:00
s += surf->lightpic->rect->x * 16 + 8;
t += surf->lightpic->rect->y * 16 + 8;
s /= 16;
t /= 16;
poly->verts[i][5] = s * surf->lightpic->size;
poly->verts[i][6] = t * surf->lightpic->size;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
}
// remove co-linear points - Ed
[model] Clean out some useless surface flags The gl water warp ones have been useless since very early on due to not doing water warp in gl (vertex warping just didn't work well), and the recent water warp implementation doesn't need those hacks. The rest of the removed flags just aren't needed for anything. SURF_DRAWNOALPHA might get renamed, but should be useful for translucent bsp surfaces (eg, vines in ad_tears).
2022-05-22 14:38:18 +00:00
if (!gl_keeptjunctions) {
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
for (i = 0; i < lnumverts; ++i) {
vec3_t v1, v2;
float *prev, *this, *next;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
prev = poly->verts[(i + lnumverts - 1) % lnumverts];
this = poly->verts[i];
next = poly->verts[(i + 1) % lnumverts];
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
VectorSubtract (this, prev, v1);
VectorNormalize (v1);
VectorSubtract (next, prev, v2);
VectorNormalize (v2);
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
// skip co-linear points
# define COLINEAR_EPSILON 0.001
run indent over all the .c files using qw's .indent.pro. The real reason for this is I mistakenly did so while making some other changes (which I made sure were NOT in the checkin:)
2001-02-26 06:48:02 +00:00
if ((fabs (v1[0] - v2[0]) <= COLINEAR_EPSILON) &&
(fabs (v1[1] - v2[1]) <= COLINEAR_EPSILON) &&
(fabs (v1[2] - v2[2]) <= COLINEAR_EPSILON)) {
int j;
for (j = i + 1; j < lnumverts; ++j) {
int k;
initial checkin of most recent newtree and nuq(?) source
2001-02-19 21:15:25 +00:00
for (k = 0; k < VERTEXSIZE; ++k)
poly->verts[j - 1][k] = poly->verts[j][k];
}
--lnumverts;
// retry next vertex next time, which is now current vertex
--i;
}
}
}
poly->numverts = lnumverts;
}
Reference in a new issue Copy permalink