libretro/ppsspp
1
0
Fork 0
mirror of https://github.com/libretro/ppsspp.git synced 2024-12-11 10:24:43 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
16,196 Commits 19 Branches 54 Tags 247 MiB
2780eef595
Commit Graph

3225 Commits

Author SHA1 Message Date
Henrik Rydgard
ff758f58ad ARM64: The Return of the Pointer Truncator - The Movie 2015-04-06 18:13:26 +02:00
Henrik Rydgard
db0fd1042f Add "Arm64EmitterTest" which can run on startup to play around with instructions. 2015-04-06 18:13:26 +02:00
Henrik Rydgard
1e9fdf08c5 Temporarily disable vertexjit 2015-04-06 18:13:23 +02:00
Henrik Rydgard
5496b3d3b1 ARM64: Some minor vertex decoder work. Hm, I think SCVTF will actually divide by 128.0, not 127.0 :/ 2015-04-06 18:13:20 +02:00
Henrik Rydgard
a12e448fb4 ARM64: Stub vertex decoder jit, implementing just enough for the cube.elf cube. 2015-04-06 18:13:18 +02:00
Henrik Rydgard
b309c83973 Initial work on ARM64, based on the ARM jit. 2015-04-06 18:13:01 +02:00
Unknown W. Brackets
2450724be2 Make Memory::Memcpy() execute memchecks directly. 
This makes it easier to handle breakpoints in HLE.
 2015-04-05 18:09:20 -07:00
Henrik Rydgård
79fc557ad1 Merge pull request #7626 from sum2012/GPU-minor 
GPU:Avoid "decimate" the usageFlags corrupted memory
 2015-03-23 15:22:08 +01:00
sum2012
3549705195 Use better solution from @Bigpet 2015-03-23 20:29:51 +08:00
Unknown W. Brackets
5fa3bf0eb9 Move the viewport transform to the proj matrix. 
The viewport has an implementation defined maximum size.  On some devices,
it's as low as 2048 or 4096.

Tales of Destiny 2, for example, uses a large viewport, which may be
outside this range at 1x or 2x on such devices.  This makes it work much
better.  It may also have better subpixel precision.
 2015-03-22 23:35:24 -07:00
sum2012
433347e3b6 GPU:Do not "decimate" the usageFlags. 
Fix #7625
 2015-03-23 07:25:44 +08:00
Henrik Rydgård
cf7115aa17 Merge pull request #7599 from unknownbrackets/texscaling 
Avoid constant reload of frequently changed textures
 2015-03-18 18:11:30 +01:00
Unknown W. Brackets
0dda098b2b Oops, forgot to actually reset vertex shader. 2015-03-15 17:05:23 -07:00
Unknown W. Brackets
d00f6adf11 Avoid querying attrib locations in depal runtime. 2015-03-15 14:44:13 -07:00
Unknown W. Brackets
8f8aeaef77 Improve depal shader failure cases. 
Primarily, this handles lost (e.g. Android sleep) better, in that it
recreates the vertex shader.  Previously, this would simply cause all
future depal operations to fail, and to constantly retry compiling and
linking the failed program.
 2015-03-15 13:15:16 -07:00
Unknown W. Brackets
7ee9112425 Avoid constant reload of frequently changed tex. 
If we detect that it's frequently changing, this will cause it to spiral
into an infinite loop of staying frequently changing and reloading to try
to scale.
 2015-03-15 09:10:09 -07:00
Henrik Rydgård
085ecc0a64 Merge pull request #7591 from unknownbrackets/gpu-framebuf 
Better glInvalidateFramebuffer usage, track usage flags better
 2015-03-15 11:18:30 +01:00
Unknown W. Brackets
6b26e1a591 Read pending vertex data before exiting gpu loop. 
Fixes save pictures in Crimson Gem Saga.
 2015-03-14 18:11:24 -07:00
Unknown W. Brackets
0f38a1f459 Invalidate buffers after using for post shaders. 2015-03-14 16:16:32 -07:00
Unknown W. Brackets
e7b830ff61 Invalidate buffers used for download. 2015-03-14 16:11:20 -07:00
Unknown W. Brackets
d8478255d4 Invalidate temp FBOs each frame. 
These are used for depal and render-to-self, as well as shader blending.
 2015-03-14 16:02:04 -07:00
Unknown W. Brackets
7e0489c83e Keep better track of framebuf usage flags. 2015-03-14 15:21:58 -07:00
Unknown W. Brackets
e45daac41c Fix a dumb mistake for shader framebuf fetch. 2015-03-14 00:14:59 -07:00
Henrik Rydgård
33b291a36e Merge pull request #7567 from unknownbrackets/gpu-blend 
Use framebuffer fetch more and on more devices
 2015-03-09 22:24:23 +01:00
Henrik Rydgård
9db4c5f34d Merge pull request #7573 from unknownbrackets/gpu-minor 
Small optimizations to software transform
 2015-03-09 09:59:45 +01:00
Unknown W. Brackets
f0676b0c85 Optimize color tests against zero. 
They seem to be somewhat common.  Speeds up Tales of Phantasia X a bit.
 2015-03-08 19:08:21 -07:00
Unknown W. Brackets
534b06d3ff Correct clearing of texture cache on oom. 2015-03-08 18:38:28 -07:00
Unknown W. Brackets
a82bb31c6f Decimate the fragtest cache less often. 2015-03-08 18:35:53 -07:00
Unknown W. Brackets
cf7a4ab2cf Avoid shader id data that doesn't change shader. 
This may reduce shader switches, or more likely at least total # of
shaders, when alpha test settings are changed but alpha test is not
enabled.
 2015-03-08 18:28:59 -07:00
Unknown W. Brackets
04eb83002c Assume lmode does nothing in throughmode. 2015-03-08 18:03:17 -07:00
Unknown W. Brackets
fb071a066d d3d9: Change vertex attribute logic to match gles. 2015-03-08 18:01:00 -07:00
Unknown W. Brackets
83e4edbc7c Oops, don't affect hardware transform. 2015-03-08 14:19:44 -07:00
Unknown W. Brackets
c7984dd93e Use texture stage 3 for the palette, always. 
It's better to keep a stage dedicated to each thing.  This also makes it
easier to potentially process depal in the shader, if we do that.
 2015-03-08 13:13:04 -07:00
Unknown W. Brackets
39cb74a8d3 Simplify transform in throughmode/sw (rectangles.) 
This should make it a bit faster, a bit less work.
 2015-03-08 13:12:13 -07:00
Unknown W. Brackets
39347efda4 Fix shader blending on ARM devices, oops. 2015-03-08 00:12:29 -08:00
Unknown W. Brackets
eb349a1978 Make sure not to try multiple shader fetch exts. 2015-03-07 22:31:55 -08:00
Unknown W. Brackets
a04af187ea Fix 16-bit indexed rectangles. 
This was causing us to draw double the rectangles.  8-bit indexed
rectangles are already correct.
 2015-03-07 19:52:09 -08:00
Unknown W. Brackets
08a1bb8796 Add support for ARM_shader_framebuffer_fetch. 2015-03-07 16:11:43 -08:00
Unknown W. Brackets
2344760018 Use framebuffer fetch for more accurate blending. 
On devices which support it, framebuffer fetch should give us more correct
blending than copying fbos or using a fallback (such as doubling which may
clamp.)

Not actually tested on any supported device.
 2015-03-07 12:09:27 -08:00
Unknown W. Brackets
9c3ccc97a2 Support EXT_shader_framebuffer_fetch. 
We only supported NV_ before.  Not tested on any supporting device, but
this is primarily found on PowerVR devices.  This may allow them to
perform effects (like min/max/etc.) more accurately and quicker.
 2015-03-07 12:07:53 -08:00
Henrik Rydgård
077eae6255 Merge pull request #7566 from Bigpet/vs2015 
Make compilable under VS2015
 2015-03-07 00:16:50 +01:00
Peter Tissen
3ada988740 capture by value, since the dispatch seems asynchronous 2015-03-06 22:10:10 +01:00
Peter Tissen
f5795a68a7 Make compilable under VS2015 2015-03-06 21:09:45 +01:00
Unknown W. Brackets
98d7afae89 Switch to #pragma once in a few places. 
Doesn't really affect git history much to change these.
 2015-03-02 22:34:51 -08:00
Unknown W. Brackets
818627801d Fix newlines in old merge, oops. 
This is now showing up in git because I touched it, and it's got the wrong
newlines.  Just making it go away.
 2015-03-01 18:35:28 -08:00
Henrik Rydgård
c0450f7f9e Merge pull request #6920 from hrydgard/dx9-depal 
Dx9 depal - needs debugging before merge
 2015-03-01 21:04:36 +01:00
Unknown W. Brackets
d4e74e6998 d3d9: Clarify a comment. 2015-03-01 11:47:39 -08:00
Unknown W. Brackets
105087d367 dx9: Swap depal bgra in shader. 
We don't swap palettes anymore, except in the shader.
 2015-03-01 11:28:06 -08:00
Unknown W. Brackets
3bc954778c dx9: Explicitly restore viewport after depal. 
Otherwise, it doesn't get restored.
 2015-03-01 11:28:06 -08:00
Unknown W. Brackets
bda6f0a4b8 dx9: Reset more state before depal draw. 
Was alpha blending before.
 2015-03-01 11:28:05 -08:00