Unknown W. Brackets
8a715f7b67
Don't forget a tex is unreliable until decimate.
2013-02-10 14:26:17 -08:00
Unknown W. Brackets
ca943f070c
Don't delete cache invalidated textures.
...
Note: still decimates.
2013-02-10 14:20:59 -08:00
Unknown W. Brackets
f77a4985f1
Keep track of whether a texture is reliable or not.
2013-02-10 14:20:58 -08:00
Unknown W. Brackets
d56f3055b6
Don't check the power of two padding in texcache.
2013-02-10 14:20:58 -08:00
Unknown W. Brackets
f15c1839ac
Use backoff to invalidate textures as well.
2013-02-10 14:20:58 -08:00
Unknown W. Brackets
5f3f3b8716
Speed up TextureCache::Invalidate() in debug.
2013-02-10 13:11:29 -08:00
Henrik Rydgard
87c9aa99c2
Let's specify MAX_LOD whether it works or not on gles 2...
2013-02-10 12:13:35 +01:00
Henrik Rydgard
db3f01044d
Turn down texturecache logging a bit. Fix unfinished optimization.
2013-02-09 21:32:02 +01:00
Henrik Rydgard
aef2376b34
Merge branch 'master' into framebuffer-texture
...
Conflicts:
GPU/GLES/TextureCache.cpp
2013-02-09 21:18:46 +01:00
Henrik Rydgard
4868b5041c
Fall back to glGenerateMipmap on OpenGL ES 2.0 for now, explanation in comments.
...
Enable 4x aniso by default, just because. This might be turned into an option later.
2013-02-09 20:21:10 +01:00
Henrik Rydgard
2fcea2a7e6
Fix mipmap-related slowdown caused by excessive texture decoding
2013-02-09 19:25:36 +01:00
Henrik Rydgard
bf7e5a4115
Mipmapping! Fixes issue #614
2013-02-08 00:04:34 +01:00
Henrik Rydgard
76acd62e6e
Try to adjust texture size for fbo dimension
2013-02-02 12:37:41 +01:00
Henrik Rydgard
a0c0d6a977
Initial support for texturing from framebuffers.
2013-02-01 00:18:23 +01:00
Henrik Rydgard
02936d5090
Don't use palette addr to build the cache key if the texture doesn't use one.
2013-02-01 00:02:50 +01:00
Unknown W. Brackets
e8519a8a9b
Walk a pointer in texture reading too.
2013-01-31 00:49:18 -08:00
Unknown W. Brackets
63ef98a994
Use pointer walking and sizeInRAM for texture hash.
2013-01-31 00:30:48 -08:00
Henrik Rydgard
3b681d3854
Make GLES_GPU own texturecache
2013-01-30 20:40:26 +01:00
Henrik Rydgard
8b180513cb
Assorted GPU cleanup/fixing (fbo, vbo draw type)
2013-01-26 21:39:36 +01:00
Henrik Rydgard
14d26141a5
TextureCache_Invalidate now does a range match instead of just checking if the texture address was in range.
2013-01-22 22:03:39 +01:00
Unknown W. Brackets
140fd171ef
Fix some 64-bit type conversion warnings.
2013-01-19 13:48:20 -08:00
Henrik Rydgard
6ab6045baf
Throw away unused FBOs. This gets rid of extreme slowness of Star Soldier on Nexus S for whatever reason.
2013-01-11 02:00:51 +01:00
Henrik Rydgard
7390f2b5f6
Some realtime profiling stuff. Enable FZ (flush-to-zero) mode.
2013-01-11 00:13:39 +01:00
Henrik Rydgard
7a7c92edeb
Oops (forgot to remove a debug statement).
2013-01-06 18:17:01 +01:00
Henrik Rydgard
26a5bfcd5d
Avoid setting texture filter over and over again.
2013-01-06 17:44:14 +01:00
Henrik Rydgard
88e921b074
Use invalidatecachehint in dcachewritebackinvalidateall too. Count invalidations.
2013-01-06 12:27:01 +01:00
Henrik Rydgard
66a098c863
Some simple optimizations to texturecache
2013-01-06 12:11:47 +01:00
Unknown W. Brackets
adf8c71745
Recheck writebackall'd textures more frequently.
...
Since we're dealing with texture upload speed (time), basing this off
number of times invalidated (also time usually) seems to make sense.
This helps lazy games that just use writebackall and expect their
graphics to update.
2013-01-05 17:24:19 -08:00
Unknown W. Brackets
3861ce0376
Use sceKernelDcacheWritebackAll() as a hint only.
2013-01-05 17:07:19 -08:00
raven02
b94bcad288
Fall back to original filter logic when filter option is off
2012-12-29 09:48:46 +08:00
raven02
e407ec4314
User define linear filtering
2012-12-29 00:39:46 +08:00
Henrik Rydgard
8c267b0105
Didn't mean to remove these calls
2012-12-22 00:38:17 +01:00
Henrik Rydgard
e42af096c8
Merge
2012-12-21 23:43:48 +01:00
Henrik Rydgard
6b3ac02dae
Now not flushing at every drawcall by mistake...
2012-12-21 22:52:09 +01:00
Henrik Rydgard
d663e28bde
More work and optimization. Still not quite there.
2012-12-21 21:49:09 +01:00
Henrik Rydgard
d4cc0bada7
Revert broken optimization in TextureCache (there are still memory savings to be had though)
2012-12-01 10:17:33 +01:00
Henrik Rydgard
aa04eaec31
TextureCache: Cleanup, memory savings
2012-11-30 22:32:50 +01:00
Henrik Rydgard
37f28ab4eb
Buildfix (?)
2012-11-28 17:09:19 +01:00
Henrik Rydgard
7a7548a70f
Warning fixes etc
2012-11-28 16:20:38 +01:00
Henrik Rydgard
5fb8e6dfb0
DXT textures now mostly working-ish (work left to do on alpha for DXT3/5)
2012-11-28 16:12:29 +01:00
Henrik Rydgard
870ea6628b
Reasonably correct viewport handling. Optional debug stats overlay (not always 100% working).
2012-11-26 17:35:08 +01:00
Sacha
3c903dda24
Improve portability to future platforms. Make use of new USING_GLES2 define throughout PPSSPP. Also use ARM definition where suitable. Remove some redundancy with includes.
2012-11-26 13:25:14 +10:00
Henrik Rydgard
1a7ca88df4
Fix some black textures
2012-11-26 01:21:14 +01:00
Henrik Rydgard
d0f829353d
Support FIXA and FIXB blend factors as well as possible.
2012-11-25 15:49:37 +01:00
Unknown W. Brackets
8971ff7e99
Don't log texture cache evictions.
2012-11-24 11:00:29 -08:00
Unknown W. Brackets
11868f26a9
Cache textures based on more parameters.
...
This makes Tales of Eternia run much faster.
2012-11-24 10:58:10 -08:00
Unknown W. Brackets
8e09869ac2
Invalidate the texcache without restarting loop.
2012-11-24 10:43:16 -08:00
Henrik Rydgard
b964516669
Fix clear mode, turn Alpha Test back on
2012-11-23 12:43:31 +01:00
Henrik Rydgard
d141d205f8
Delete old textures, to prevent memory usage from growing indefinitely.
2012-11-22 23:07:15 +01:00
Henrik Rydgard
d572d40c36
More warning fixing
2012-11-18 23:35:02 +01:00
Unknown W. Brackets