Commit graph

46 commits

Author SHA1 Message Date
Bill Currie
634219ea06 [qfvis] Add set debug prints (disabled)
They were useful for narrowing down why mightsee wasn't being updated.
2021-03-28 21:11:13 +09:00
Bill Currie
ff4cd84891 [qfvis] Use simd vector code
While whether it's any faster is debatable (it's slightly slower, but
many more portals are being tested due to different rounding in the base
vis stage), it's certainly easier to read.
2021-03-28 19:55:47 +09:00
Bill Currie
eb325376b1 [qfvis] Collect base vis culling stats
Specifically, just how many are culled by sphere and winding tests.
2021-03-28 12:17:15 +09:00
Bill Currie
d072a7b99c [qfvis] Add stats for memory usage
Verbosity levels probably need more tweaking, but -v is at least a
little more usable.
2021-03-27 23:04:13 +09:00
Bill Currie
3ef38188ce [qfvis] Add an option to limit the processed portals
It's not documented as I needed it for debugging memory allocations and
it causes qfvis to error out due to unprocessed portals.
2021-03-27 20:59:56 +09:00
Bill Currie
f2b6b23acc [qfvis] Switch to unsigned for various counts 2021-03-27 20:55:15 +09:00
Bill Currie
72280186bf [qfvis] Use cmem for memory management
While the main bulk of the improvement (36s down from 42s for
gmsp3v2.bsp on my i7-6850K) comes from using a high-tide allocator for
the windings (which necessitated using a fixed size), it is ever so
slightly faster than using malloc as the back-end.
2021-03-27 20:30:35 +09:00
Bill Currie
c901fe74f9 [qfvis] Fix pthread portability macros
Those that were defined were incorrectly defined (didn't swallow the
parameter), and portal lock macros were missing.
2021-03-26 15:27:48 +09:00
Bill Currie
6d5ffa9f8e [build] Move to non-recursive make
There's still some cleanup to do, but everything seems to be working
nicely: `make -j` works, `make distcheck` passes. There is probably
plenty of bitrot in the package directories (RPM, debian), though.

The vc project files have been removed since those versions are way out
of date and quakeforge is pretty much dependent on gcc now anyway.

Most of the old Makefile.am files  are now Makemodule.am.  This should
allow for new Makefile.am files that allow local building (to be added
on an as-needed bases).  The current remaining Makefile.am files are for
standalone sub-projects.a

The installable bins are currently built in the top-level build
directory. This may change if the clutter gets to be too much.

While this does make a noticeable difference in build times, the main
reason for the switch was to take care of the growing dependency issues:
now it's possible to build tools for code generation (eg, using qfcc and
ruamoko programs for code-gen).
2020-06-25 11:35:37 +09:00
Bill Currie
34bcf7faab Do a pure/const/noreturn/format attribute pass.
I always wanted these, but as gcc now provides warnings for functions that
could do with such attributes, finding all the functions is much easier.
2018-10-09 12:42:21 +09:00
Bill Currie
88e5adcec6 Make the base vis multi-threaded.
Now multi-threaded qfvis is on par with tyrutils vis (differences usually
<1s, sometimes more, sometimes less).
2013-03-19 11:42:09 +09:00
Bill Currie
cb096c601d Use a per-portal rwlock for portal updates.
This should make qfvis scale a little better with cpu count.
2013-03-18 15:03:11 +09:00
Bill Currie
c824e668ed Rework some of the pthread stuff.
Init/uninit is now separate from portal vising.
The global lock has a better name and is now a rwlock.
Use a separate lock for the stats.
2013-03-18 14:26:52 +09:00
Bill Currie
ffb6d628bd Simplify the pthreads detection macros. 2013-03-18 13:31:35 +09:00
Bill Currie
1c20a49dba Use the recursive set allocator for mightsee.
This completely removes the lock used to protect the set allocation code
while keeping the use of the set api clean.
2013-03-18 13:30:50 +09:00
Bill Currie
a28ec8aa82 Revert "Allocate stack blocks and mightsee in one block."
This reverts commit 1ea79e8626.

Conflicts:
	tools/qfvis/include/vis.h
	tools/qfvis/source/flow.c

I've decided to do reentrant versions of the set allocators and I didn't
particularly like the invasiveness of allocating sets this way.
2013-03-18 12:47:59 +09:00
Bill Currie
ccc432a7ea Give the fields of pstack_t clearer names.
And some comments.
2013-03-17 19:18:38 +09:00
Bill Currie
1ea79e8626 Allocate stack blocks and mightsee in one block.
This bypasses set_new, but completely removes the use of the global lock
from within RecursiveClusterFlow. This seems to give a small speedup: 203
seconds threaded.
2013-03-17 16:37:27 +09:00
Bill Currie
9b10304c2f Make CopyWinding const-correct. 2013-03-15 19:25:24 +09:00
Bill Currie
f80ae52828 Make vis's ClipWinding const-correct. 2013-03-15 15:28:25 +09:00
Bill Currie
77c858060d Add a bunch more statistics.
Now I know why sphere culling was a loss: 78% of all tested target portals
were trimmed by ClipToSeparators (50% eventually clipped away entirely).
2013-03-14 19:43:46 +09:00
Bill Currie
97da7fe31d Document some fields. 2013-03-14 19:43:46 +09:00
Bill Currie
eec87bd61b Remove thread from stack_t.
It really wasn't gaining anything and made reading the code a little
harder.
2013-03-14 19:43:46 +09:00
Bill Currie
5d6df082f2 Move the vis stats vars into thread data.
This should make the stats more reliable when running multi-threaded
(chains is still random, but it seems there are set access issues).
2013-03-14 12:52:40 +09:00
Bill Currie
b9d71218f6 Use sphere culling in the base vis.
Base vis was done first for testing. Optimized base vis is down to ~12.4s
from ~16s (29% faster?).
2013-03-13 21:32:18 +09:00
Bill Currie
d1e65257b6 Implement the cached separators idea from tyrutils.
I think the reason I didn't think of that when I tried to improve qfvis's
performance many years ago is I just simply did not understand
ClipToSeparators. However, the difference caching the separators makes is
phenomenal. Before the change, single threaded qfvis would get stuck on one
particular portal for at least a day (I gave up waiting), but now even a
debug build will complete gmsp3v2.bsp in less than 12 minutes (4 threads on
my quad-core). And that's at level 2! Getting stuck for a day was at level
0.
2013-03-08 22:20:29 +09:00
Bill Currie
dbdfdb6d28 Add support for PRT2 portal files.
These seem to be identical to PRT1-AM but with a different count order in
the header. Taken from tyrutils-0.5.
2013-03-07 18:51:32 +09:00
Bill Currie
299ff8f575 Use set functions for qfvis.
While noticeably slower than the previous expanded set manipulation code,
this is much easier to read. I can worry about optimizing the set code when
I get qfvis behaving better.
2013-03-07 11:06:55 +09:00
Bill Currie
a2f2d4d949 "Check" for the availability of pthreads.
Unfortunately, just because the header is there doesn't mean anything will
actually work :(. Also, the check is based on the host vendor/os for now.
Yes, it's rather lame but it will do for now.

With this, QF will build on an almost fresh ps3toolchain install. Only two
"fixes" are needed:
 o  In $PS3DEV/ppu/powerpc64-ps3-elf: ln -s ../include sys-include
 o  libsamplerate cross-built and installed.
2012-08-19 13:40:42 +09:00
Bill Currie
23a38738fc Massive whitespace cleanup.
Lots of trailing whitespace and otherwise blank lines.
2012-05-22 08:23:22 +09:00
Bill Currie
bc1b483525 Nuke the rcsid stuff.
It's pretty useless in git.
2012-04-22 10:56:32 +09:00
Bill Currie
91e65b6c80 Rename mplane_t to plane_t and clean up the mess.
I got rather tired of there being multiple definitions of mostly compatible
plane types (and I need a common type anyway). dplane_t still exists for
now because I want to be careful when messing with the actual bsp format.
2011-11-28 20:54:34 +09:00
Bill Currie
142defe9c0 Parameter consistency fixes.
Make the params for FreeWinding and CopyWinding consistent with those in
qfbsp. This fixes some doxygen warnings while I think about how best to
handle the duplicate code.
2010-10-13 20:52:07 +09:00
Bill Currie
a51e888a1b Nuke MAX_OSPATH and clean up the mess. 2010-08-25 13:31:08 +09:00
Bill Currie
0dfff8fd58 ignore stuff 2010-08-07 10:42:09 +00:00
Bill Currie
371a0b8e75 support old-style portal files again 2004-02-02 05:44:46 +00:00
Bill Currie
7073afc0a4 port in OQ's detail, hint and skip brush/texture enhancements 2003-02-04 23:26:26 +00:00
Bill Currie
e81a0e2095 qfvis and qflight are still copyright Id 2002-09-25 01:51:58 +00:00
Bill Currie
ded572b31f various var cleanups 2002-09-23 22:54:28 +00:00
Bill Currie
ee61eaebbb don't do threading if only 1 thread is used and add another state to
vstatus_t for better portal state checking
2002-09-22 21:54:41 +00:00
Bill Currie
7d7e578b71 white and grey space cleanup 2002-09-21 21:27:38 +00:00
Bill Currie
544d4d6ff8 fix the missing LOCK and UNLOCK defs (found by accident:) and use a dstring
for building the compressed vis data
2002-09-20 19:03:06 +00:00
Bill Currie
24a9b48127 rewrite bspfile.[ch] so they work off an object rather than global data.
The direct filesystem access is also removed. Fix up qfvis and qflight to
compile with the new bspfile. qfvis seems to work, qflight probably doesn't
2002-09-19 05:35:17 +00:00
Ragnvald Maartmann-Moe IV
6376ebb253 Whitespace. 2002-09-19 02:37:52 +00:00
Ragnvald Maartmann-Moe IV
fed01ddd3e .gitignores for qf map tools 2002-08-25 23:16:48 +00:00
Ragnvald Maartmann-Moe IV
a3af2d9fa9 Add Hideo's WIP qflight & qfvis. 2002-08-25 23:06:23 +00:00