Commit Graph

168 Commits

Author SHA1 Message Date
Henrik Rydgard
7faddd6100 Consolidate the two TextureScaler copies into one.
Also lets us clean up ColorConv slightly.
2015-01-20 17:32:55 +01:00
Henrik Rydgard
2623a48b4a Centralize color conversion functions in Common/ColorConv. 2015-01-20 17:32:54 +01:00
Unknown W. Brackets
05feac3eb0 Fix a potential divide by zero. 2015-01-19 08:43:43 -08:00
Unknown W. Brackets
340029f333 Avoid a bad buffer access on GE_PRIM_KEEP_PREVIOUS. 2015-01-17 18:43:01 -08:00
Henrik Rydgard
c2500744e1 D3D9 rectangle rendering fix (maxindex) 2015-01-15 23:58:07 +01:00
Unknown W. Brackets
d663cda2de Transform rects using indices.
Not a hge difference.  Minor cleanup.
2015-01-15 12:26:35 -08:00
Henrik Rydgard
2cc1bd327a Make the reporting option more prominent. Also move a few options around.
Very minor change in indexgenerator.
2015-01-11 00:12:32 +01:00
Unknown W. Brackets
5cea64316a Avoid using an offset framebuffer for large tex.
When texturing from the entire 512 height of a framebuffer, we still
should not trigger the offset framebuffer code.  This is expected to help
Danball Senki BOOST but not break Breath of Fire 3.
2015-01-06 20:57:19 -08:00
Unknown W. Brackets
06e9d7680a Handle a case of 480x360 framebuffers.
Region and scissor are 480x360, but viewport is just default (also,
throughmode is on.)  We could add more checks, but just the height being a
bit higher is already unusual, so should be relatively safe.

Fixes #7277.
2015-01-06 00:02:05 -08:00
Unknown W. Brackets
7f7bf937fc Scale UVs in through and non-through.
This also shows them scaled to the w/h in the vertex preview, which is
probably easier to read and matches positions.

Additionally, fixes textures drawn fully not being highlighted at all.
2015-01-03 17:16:16 -08:00
Unknown W. Brackets
7ecff0acd6 Support more screenshot framebuffer formats. 2014-12-31 16:17:23 -08:00
Unknown W. Brackets
962d8c5224 dx9: Take screenshots (fixes #7197.) 2014-12-20 08:31:56 -08:00
Unknown W. Brackets
2e9f492b24 Silence an initialization warning.
Pretty sure it's wrong but this isn't that hot anyway.
2014-12-13 18:36:28 -08:00
Unknown W. Brackets
7ed67ad066 Use floats and SSE for spline colors. 2014-12-13 10:36:09 -08:00
Unknown W. Brackets
e130c1648d Clean up some conversion warnings. 2014-12-13 10:35:47 -08:00
Henrik Rydgård
5d6d552e57 Merge pull request #7175 from xebra/spline_bezier
Improve spline/bezier tesselation
2014-12-13 17:45:15 +01:00
xebra
43e3028add [spline/bezier]Writing vertices to the buffer directly. 2014-12-14 01:08:00 +09:00
Lioncash
dc436aebe8 GPU: Get rid of unused variable warnings 2014-12-12 09:29:41 -05:00
xebra
51a5935d9a [spline/bezier]Bernstein function does not need to calculate at the edges. 2014-12-10 01:50:28 +09:00
xebra
9a83c99e19 [spline/bezier]Precompute bernstein derivative functions. 2014-12-10 01:41:02 +09:00
xebra
8ef4a61bf6 [spline/bezier]Spline subdivision counts should be 2 or higher. Related to #4234 #7111 2014-12-09 02:42:52 +09:00
Lioncash
a962bc5a6c Mark some functions as const 2014-12-07 17:08:13 -05:00
xebra
f6b937fe59 [spline/bezier]Fix sometimes spline disappear in medium quality. Reported on #7111. 2014-12-03 01:49:08 +09:00
xebra
2d59a15ade [spline/bezier]Fix spline tex coords correctly, it seems perfect! 2014-12-03 01:26:40 +09:00
xebra
ad22ffc679 [spline/bezier]Support medium quality. It's half of high quality. High quality is the original. 2014-11-26 12:23:02 +09:00
xebra
8cc7d8c71a [spline/bezier]Fix patch subdivision counts correctly. 2014-11-25 23:47:29 +09:00
xebra
f0ef073307 Fix NormalizeVertices() for indexed vertices. 2014-11-25 21:24:54 +09:00
xSacha
88b1c00a67 MIPS: Complete MIPSX rename for buildfix. 2014-11-19 07:55:44 +10:00
xSacha
3172fa22e2 Integrate new MIPS emitter. 2014-11-18 08:40:57 -08:00
Henrik Rydgard
4422b3deb7 x86: Minor soft-skinning optimization 2014-11-16 17:43:29 +01:00
Sacha
92b0f3b7d9 Minor change. Fix build. 2014-11-13 23:45:45 +10:00
xSacha
57e4088216 Introduce fake vertex decoder JIT as well.
Compiles and links on CI20 but gets unknown crash in GL driver.
2014-11-13 17:10:29 +10:00
Unknown W. Brackets
c02172ebd5 vertexjit: Oops, this wasn't supposed to be there.
Fixes #7081.
2014-11-11 08:31:27 -08:00
Unknown W. Brackets
a838459627 vertexjit: Optimize weights on x64 (not 32-bit.)
A few percent in Gods Eater Burst and Kingdom Hearts.
2014-11-08 15:16:02 -08:00
Unknown W. Brackets
eee3ac79f4 Always clamp in ToRGB[A]?().
Before we only clamped with SSE, better to be consistent.  This may also
be slightly faster.
2014-10-31 09:07:54 -07:00
Unknown W. Brackets
1b12b57e21 Fix reversed consts in hand-neon texhash. 2014-10-29 16:38:40 -07:00
Unknown W. Brackets
faea860c27 Attempt to match the SSE2 version of the hash.
Not tested, don't currently have an ARM device.  May or may not have a
performance hit.
2014-10-29 16:37:45 -07:00
Unknown W. Brackets
3977f5a8ec Add a non-SSE implementation of the QuickTexHash.
This way we can use it as a stable value for texture replacements.
2014-10-29 16:37:44 -07:00
Unknown W. Brackets
2b16b5b79b Use XXH64 on 64-bit systems.
I'm not able to reproduce any real performance difference, but I know
there are some games which do more hashing and it may help there.
2014-10-26 17:49:24 -07:00
Henrik Rydgard
d24abb3af0 More GetPointer cleanup. memCardDirectory->memStickDirectory. 2014-10-19 23:20:51 +02:00
Henrik Rydgard
7cb5e7f53f Some very minor optimizations. Remove little-used stat counter. 2014-10-19 21:12:25 +02:00
Henrik Rydgard
8177b4c43b Avoid an ifdef using PTRBITS 2014-10-12 19:35:55 +02:00
Unknown W. Brackets
358462a7f4 Add a .gitattributes to normalize newlines.
No code changes.
2014-09-26 21:20:50 -07:00
Unknown W. Brackets
cee2827172 Normalize newlines, no code changes.
We really shouldn't let mixed newlines creep into the codebase.
They're annoying.
2014-09-24 23:10:13 -07:00
Unknown W. Brackets
9bcffc5b4d Fix divide by zero in memcpy checks.
Happened in Resistance demo.
2014-09-21 21:36:26 -07:00
Unknown W. Brackets
4210ba44eb Clean up a few more ImmPtr() cases. 2014-09-21 08:34:27 -07:00
Henrik Rydgård
23ade6108d Merge pull request #6917 from unknownbrackets/vertexjit
Add x86 weight ToFloat handling for d3d9
2014-09-18 17:48:22 +02:00
Unknown W. Brackets
967df9c018 vertexjit: Use SSE4 in tc u8/u16 code.
It's not much different for non-SSE 4, might as well be common.
2014-09-17 23:15:24 -07:00
Unknown W. Brackets
5276487611 vertexjit: Use SSE for all weights, u8/u16. 2014-09-17 23:14:36 -07:00
Henrik Rydgard
07c7687052 More code deduplication. Also normalized some line endings. 2014-09-18 00:45:11 +02:00