- fixed: Visplane checks should only compare the plane flags that are relevant for rendering and mask out the rest.
- floatified FTransform and made the visplane checks a bit less verbose by moving the comparison as an operator into FTransform.
Note that this operator needs forceinline on Visual Studio so that it won't get called as a function.
The only reason this even existed was that ZDoom's original VC projects used __fastcall. The CMake generated project do not, they stick to __cdecl.
Since no performance gain can be seen by using __fastcall the best course of action is to just remove all traces of it from the source and forget that it ever existed.
- replaced some uses of FRACUNIT with OPAQUE when it was about translucency.
- simplified some overly complicated translucency multiplications in the SBARINFO code.
- Converted P_MovePlayer and all associated variables to floating point because this wasn't working well with a mixture between float and fixed.
Like the angle commit this has just been patched up to compile, the bulk of work is yet to be done.
Patched up everything so that it compiles without errors again. This only addresses code related to some compile error. A large portion of the angle code still uses angle_t and converts back and forth.
- some consolidation in p_map.cpp. PIT_CheckLine and PIT_FindFloorCeiling had quite a bit of redundancy which has been merged.
- čontinued work on FMultiBlockLinesIterator. It's still not completely finished.
Converting a floating point value that is out of range for a signed integer will result in 0x80000000 with SSE math, which is used exclusively for this purpose on modern Visual C++ compilers, so this cannot be used anywhere.
On ARM there's problems with float to unsigned int conversions.
xs_Float does not depend on these
- The original Doom renderer was inclusive for all right edges. This was
fine for the wonky projection it did. This was not fine for a standard
perspective divide, so I had to change walls to be right-edge exclusive
when I changed the projection. I only touched what was needed. Until
now. The right edge is always exclusive now, which should prevent any
more bugs related to mixing the two clusivities incorrectly.
- Since Clang++, G++, and VC++ all support this extension (even though it's
technically officially only part of C99), use it. It lets Clang's array-
bounds checker know that these are meant to be accessed out of their so-called
"bounds".
the camera is on. Mostly, this means testing the distance of the camera to the plane rather
than computing the plane's Z at the camera and comparing that with the camera's Z.
SVN r4220 (trunk)
be used when for walls and floors when the renderer is paletted. The format
is very simple:
rgbtex1 paltex1
rgbtex2 paltex2
...
The first texture is the one to be used normally, and the second is the one
to be used in paletted modes.
The vid_nopalsubstitutions cvar can be used to ignore this lump.
SVN r3311 (trunk)
GETPALOOKUP. The end result is that there is a minimum distance around you where light
amplification stops and it gets no brighter. Should this scale with visibility? I can't say.
So, yeah, it turns out all these years ago, I made this out to be harder than it really is.
SVN r3224 (trunk)
- added Sector_SetTranslucent special so set translucency of portal planes at run time.
- added 'additive' information for portal planes. This is no-op at the moment because the flat drawers can't handle additive translucency yet though.
SVN r3149 (trunk)
does not include xtoviewangle[centerx] in the mirroring so that the two columns at the center
of the screen do not map to the same angle. (BTW, this array is only used for the sky drawing.)
SVN r2487 (trunk)
it happened outside the moved actor's Tick function. This got particularly
obvious with moving skybox viewpoints (See Daedalus's MAP21 intro for a good
example.)
SVN r2059 (trunk)
former used fistp, which is not portable across platforms, so cannot be
used in the play simulation. They were only suitable for the renderer.
xs_Float.h also has a very fast float->fixed conversion, so FLOAT2FIXED
uses that now.
(And I also learned that the FPU's round to nearest is not the rounding I
learned in grade school but actually Banker's Rounding. I had no idea.)
(Also, also, the only thing that could have made quickertoint faster than
toint was that it stored a 32-bit int. I never timed them, and I doubt in
practice there was any real difference between the two.)
- Changed atan2f to atan2. Using floats is not a win, because the result is
returned as a double on the x87 stack, which the caller then needs to cast
down to a float using fst/fld.
SVN r1990 (trunk)
are scaled to the height of a 200 pixel tall sky. Skies taller than 240
use the same scale as a 240 tall sky but are shifted down to make the
top of the texture align with the top of the screen when looking fully up.
Thus, by using a sky texture with a height of 240 or more pixels, the sky
will be drawn with square pixels instead of the vertically elongated ones
imposed by Doom's native 320x200 resolution.
SVN r1978 (trunk)
- Improved sky stretching a bit: It now only stretches the sky as tall as it
needs to be: 228 pixels, not 256. It no longer stretches horizontally,
either.
The reason it stretches to 228 and not 200 pixels is because Doom shifted
its sky texture down 28 pixels. By stretching to 228 pixels, we can keep
the sky tiled at the same height on the horizon. Skies 200 pixels tall
(or more) will continue to tile at the center of the screen when looking
directly ahead.
SVN r1976 (trunk)
unsigned integer that can use all 32 bits. They must therefore use
the unsigned mul instruction rather than the signed imul instruction.
- Fixed several signed/unsigned comparison and possibly uninitialized
variable warnings flagged by GCC.
SVN r1965 (trunk)
size textures at any scale. I also tried doing sky scrolling on the sky
cylinder, but that didn't look so good, so I left it in screen space.
SVN r1932 (trunk)
completely ignore them, either).
- Separated light level fixing out of player_t's fixedcolormap parameter.
Using a fixed light level (e.g. PowerTorch) will no longer wipe out
colored lighting.
- Moved the blending rectangle drawing into a separate discrete stage, since
doing it while copying the 3D view window to the display now blends
underneath the weapon instead of on top of it.
- Consolidated the special colormaps into a single 2D table.
- Tweaked the special colormaps slightly to make the true color results more
closely match the paletted approximations.
- fb_d3d9_shaders.h was getting unwieldy, so I moved the shaders out of the
executable and into zdoom.pk3. Shaders are still precompiled so I don't need
to pull in a dependancy on D3DX.
- Added a few more shaders to accomodate drawing weapons with all the in-game
lighting models. These are accessed with the new DrawTexture tags
DTA_SpecialColormap and DTA_ColormapStyle.
- Player weapon sprites are now drawn using Direct3D and receive all the
benefits thereof.
SVN r1858 (trunk)
- Fixed: P_LineAttack may not check the puff actor for MF6_FORCEPAIN because
it's not necessarily spawned yet.
- Fixed: FWeaponSlots::PickNext/PrevWeapon must be limited to one iteration
through the weapon slots. Otherwise they will hang if there's no weapons
in a player's inventory.
SVN r1659 (trunk)
* scalex_top
* scaley_top
* scalex_mid
* scaley_mid
* scalex_bottom
* scalex_bottom
- Added sidedef versions of the linedef flags wrapmidtex and clipmidtex (via
UDMF; names are the same). If the flag is set on the line, it applies to
both sides. Otherwise, each side can control them individually.
SVN r1645 (trunk)
- Removed #pragma warnings from cmdlib.h and fixed the places where they were
still triggered.
These #pragmas were responsible for >90% of the GCC warnings that were not
listed in VC++.
- Fixed one bug in the process: DSeqNode::m_Atten was never adjusted when the
parameter handling of the sound functions for attenuation was changed.
Changed m_Atten to a float and fixed the SNDSEQ parser to set proper values.
Also added the option to specify attenuation with direct values in addition
to the predefined names.
SVN r1583 (trunk)
so that all files are included by a central one instead of compiling
each one separately. This speeds up the compilation process by 25%
when doing a complete rebuild in Visual C.
- Cleaned up more header dependencies.
SVN r1226 (trunk)
registers AMD64 provides, this routine still needs to be written as self-
modifying code for maximum performance. The additional registers do allow
for further optimization over the x86 version by allowing all four pixels
to be in flight at the same time. The end result is that AMD64 ASM is about
2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM.
(For further comparison, AMD64 C and x86 C are practically the same for
this function.) Should I port any more assembly to AMD64, mvlineasm4 is the
most likely candidate, but it's not used enough at this point to bother.
Also, this may or may not work with Linux at the moment, since it doesn't
have the eh_handler metadata. Win64 is easier, since I just need to
structure the function prologue and epilogue properly and use some
assembler directives/macros to automatically generate the metadata. And
that brings up another point: You need YASM to assemble the AMD64 code,
because NASM doesn't support the Win64 metadata directives.
- Added an SSE version of DoBlending. This is strictly C intrinsics.
VC++ still throws around unneccessary register moves. GCC seems to be
pretty close to optimal, requiring only about 2 cycles/color. They're
both faster than my hand-written MMX routine, so I don't need to feel
bad about not hand-optimizing this for x64 builds.
- Removed an extra instruction from DoBlending_MMX, transposed two
instructions, and unrolled it once, shaving off about 80 cycles from the
time required to blend 256 palette entries. Why? Because I tried writing
a C version of the routine using compiler intrinsics and was appalled by
all the extra movq's VC++ added to the code. GCC was better, but still
generated extra instructions. I only wanted a C version because I can't
use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit
of a pain. (It's a pain because Linux and Windows have different calling
conventions, and you need to maintain extra metadata for functions.) So,
the assembly version stays and the C version stays out.
- Removed all the pixel doubling r_detail modes, since the one platform they
were intended to assist (486) actually sees very little benefit from them.
- Rewrote CheckMMX in C and renamed it to CheckCPU.
- Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache
only for AMD processors, so we must not use it on other architectures, or
we end up overwriting the L1 cache line size with 0 or some other number
we don't actually understand.
SVN r1134 (trunk)
arbitrary point. It has been replaced with a variant that takes a polyobject
as a source, since that was the only use that couldn't be rewritten with the
other variants. This also fixes the bug that polyobject sounds were not
successfully saved and caused a crash when reloading the game. Note that
this is a significant change to how equality of sound sources is determined,
so some things may not behave quite the same as before. (Which would be a
bug, but hopefully everything still sounds the same.)
SVN r1059 (trunk)
that animated icons can be done with it.
- Changed FImageCollection to use a TArray to hold its data.
- Fixed: SetChanHeadSettings did an assignment instead of comparing
the channel ID witg CHAN_CEILING.
- Changed sound sequence names for animated doors to FNames.
- Automatically fixed: DCeiling didn't properly serialize its texture id.
- Replaced integers as texture ID representation with a specific new type
to track down all potentially incorrect uses and remaining WORDs used
for texture IDs so that more than 32767 or 65535 textures can be defined.
SVN r1036 (trunk)
- Grouped the sector plane texture transformation options into a separate
structure and replaced all access to them with wrapper functions.
SVN r1033 (trunk)
