Commit Graph
148 Commits
Author SHA1 Message Date
Henrik Rydgard 01a1438dce Improve depth buffering in D3D by computing a depth-flipped proj matrix but only when needed. 2014-09-08 23:10:23 +02:00
Unknown W. Brackets 1f51fe7843 d3d: Avoid rewriting textures, just swizzle.
Luckily A is in the same place and the same width, so we can do this for
all but framebuffers easily.

Technically we could do it in OpenGL as well.

Small (1-2%) performance improvement in FF2.
2014-08-28 01:20:21 -07:00
Unknown W. Brackets a43c3771a3 More correctly blend when hitting the frame cap.
If we hit too many blits per frame, fall back correctly in the shader.
2014-08-18 23:20:48 -07:00
The DaxandHenrik Rydgard 8a78f5e814 Pause rendering when minimized, when appropriate (hopefully? This is sort of hard to test.). 2014-07-21 10:30:51 +02:00
Unknown W. Brackets 2883988f22 Ignore the high bits of the framebuffer address.
And depth buffer.  On a PSP, these bits don't affect where it renders to
even slightly.
2014-06-21 08:29:38 -07:00
Unknown W. Brackets 2038bc5527 Support alpha / color test masks on desktop/gles3.
Won't work on GLES 2, though...
2014-06-16 00:33:48 -07:00
Unknown W. Brackets f6649794df Respect max texture level in GE debugger preview. 2014-06-15 10:31:16 -07:00
Unknown W. Brackets 030b562eb9 Support render-to-offset within a framebuffer.
Fixes #6324.
2014-06-14 17:08:41 -07:00
Unknown W. Brackets 1d9f7b04fc Implement in-shader blending on gles2 / gl2.
I'm sure it'll be slow, though...
2014-06-13 23:42:59 -07:00
Unknown W. Brackets acab898be0 Keep track of x/y offset for render-to-tex.
Fixes intro in Valkyrie Profile now that we're wrapping correctly.
2014-06-08 14:27:05 -07:00
Unknown W. Brackets cc841bbe4c Apply tex wrap/clamp in shader for render-to-tex.
Fixes graphical artifacts in Wild Arms XF (which depends on how you get
there, because it's based on the size of the framebuffer.)
2014-06-07 13:13:58 -07:00
Unknown W. Brackets 580143e5e2 Disable vertex arrays before depal as well. 2014-05-31 22:07:42 -07:00
Unknown W. BracketsandHenrik Rydgard 29a9ff369e Make a note when we need to reset the shader.
Hmm, not sure this is the cleanest way.
2014-05-27 22:12:25 +02:00
Unknown W. Brackets f660c04983 Optimize double alpha blending for 1/0 textures.
If the blend is 2*a / 1-a, and the alpha is either 1 or 0, we can still
use color doubling instead.  Fixes #3379 (Persona 2.)
2014-05-11 14:17:19 -07:00
Henrik Rydgard 16f4622ac7 Remove cached light variables to reduce work when parsing DLs.
Also removes some duplication between GL and the inactive D3D code.
2014-04-21 12:51:19 +02:00
raven02 abc5c86339 Split out texture level 2014-04-19 06:45:17 +08:00
Unknown W. Brackets 6cfc61665d Switch to a bitmask instead. 2014-04-13 14:45:55 -07:00
Unknown W. Brackets 4c1c694d4b Use flags to avoid hashing textures when unchanged.
If only parameters change (like wrapping or clut, etc.) we don't need to
rehash the data - we know it hasn't changed.

Should reduce the distance between lazy texture hashing on and off.
2014-04-13 14:22:39 -07:00
Henrik Rydgard b174996c1c Add a conservative check that prevents alpha testing in a few cases.
This will become really powerful if we add some code to the vertex decoder
to check for non-full alpha in the vertices, and set gstate_c.vertexFullAlpha if none
is found (probably want to do the reverse, set it to true and clear if any non-255 alpha is found).

Alpha testing is a performance killer on many mobile GPUs so big efforts to
avoid it can be worth it.
2014-03-23 16:32:38 +01:00
Unknown W. Brackets 678237aa6c Improve SSE usage in software transform.
It's actually already pretty decent (unlike the softgpu), but there were a
few places it could use a bit of help.  Speeds up things with hardware
transform off, or areas that need to use software transform.
2014-03-17 23:05:48 -07:00
raven02 04f31c5e8e Add gstate.isSkinningEnabled() 2013-12-16 20:42:45 +08:00
Unknown W. Brackets a2d91ee26d softgpu: Don't mask out so much of fb stride.
Seems like anything from 4 is supported in 8888, most likely it just needs
to align to 16 bytes.  Values above 1024 work, but e.g. 2044 seems buggy.

Fixes the map on Hexyz Force (rendered at 80 stride.)
2013-12-15 11:59:24 -08:00
raven02 2e7609b084 Cleanup ClearmodeDepthMask 2013-12-10 20:48:20 +08:00
Henrik Rydgård e14f17b430 isClutSharedForMipmaps: This was presumably inverted by accident? 2013-12-09 12:46:13 +01:00
Unknown W. Brackets c6a441965e Move shared clut check to gstate. 2013-12-08 23:11:56 -08:00
Unknown W. Brackets 435115441d Reinitialize the GE on sceKernelLoadExec().
Mostly, reset the lists and draw completion, etc.

May be required for Capcom Classic Collection Reloaded to work.
2013-11-28 15:35:12 -08:00
Unknown W. Brackets 7a86ca2890 Mask block transfer stride correctly per tests.
1024 is valid, but nothing higher.

May help graphics in some games.
2013-11-25 01:53:33 -08:00
Henrik Rydgard 8c562a615f Bone matrix loading speedup. Bit of a hack but seems reliable enough. 2013-11-14 14:02:31 +01:00
Henrik Rydgard 7e67476b00 Simple unoptimized software skinning.
Does not take advantage of the possible reduction in state changes yet.
2013-11-13 18:10:57 +01:00
Henrik Rydgard 4b98e0d6d6 Optimize LoadClut a little 2013-11-12 17:06:03 +01:00
Unknown W. Brackets 21deaef35a Invert gstate.getClearModeColorMask().
It makes more sense matching gstate.getColorMask().
2013-11-10 03:20:58 -08:00
Henrik Rydgard 80702109f5 Move gpu vendor detection to native with the rest of the gl init.
Also disable vertical layout for mainscreen entirely.
2013-10-16 00:48:44 +02:00
Unknown W. Brackets d49ad5f114 Use an accessor for depth/frame addresses. 2013-10-06 22:17:53 -07:00
Unknown W. Brackets 3787471791 Add some initial methods to GPUDebugInterface. 2013-09-22 10:22:33 -07:00
Henrik Rydgard d7ae3f88a1 Do Beziers more correctly (although still not tesselating properly).
Snow now visible in SSX.
2013-09-22 09:52:46 +02:00
Henrik Rydgard 01def3b6cc Missed these 2013-09-21 23:44:11 +02:00
Henrik Rydgard 82e5787bbe Preparation for proper spline/bez: Convert control points to a simple format.
The bezier/spline code will no longer need to handle morph and splines,
when it's finally written. This is done by pre-skinning in software and
pre-decoding to get rid of the morph.
2013-09-21 23:37:14 +02:00
Unknown W. Brackets a1f3591960 Android buildfix. 2013-09-20 09:51:43 -07:00
Unknown W. Brackets 7906de26f7 Try to match especially the size of ge contexts.
But also some of the actual data, most of it matches like this.
2013-09-20 00:33:32 -07:00
Unknown W. Brackets ad4bd8ed74 softgpu: small optimization, precompute tex info. 2013-09-15 21:39:28 -07:00
Unknown W. Brackets 68f61e5071 Remove some unused gstate_c values. 2013-09-14 14:36:06 -07:00
Unknown W. Brackets 7b2dea4429 Use accessors for offsetx/y. 2013-09-14 11:06:48 -07:00
Unknown W. Brackets 40e3e8e930 Use a gstate enum, report unhandled blend mode. 2013-09-14 07:36:43 -07:00
raven02 f6c41bfa13 Add u16 getTextureDimension(int level) 2013-09-08 16:09:40 +08:00
Unknown W. Brackets bbf714c361 Provide for GPU init failure, kill a warning. 2013-09-07 22:30:30 -07:00
raven02 e834b1b80d Use GEComparison for getDepthTestFunction() 2013-09-05 13:59:09 +08:00
raven02 7f080b9958 Add isAntiAliasEnabled() 2013-08-30 20:46:36 +08:00
Henrik Rydgard 711d3ff52c Speedup a few 2D games on mobile by eliminating the alpha test in some conditions
That is, if there's no depth testing/writing, and the alpha equation will produce the same result as the alpha test (when the alpha test is >= 0 or != 0 and the blend func is safe), we turn off the alpha test.
2013-08-27 21:48:03 +02:00
Henrik Rydgard 6360fcb0ae Add some stats to get some idea of much alpha tested drawing we do.
Turns out it's one hell of a lot and we really should fix this by
checking both vertices and textures to deduce when we can just skip
alpha testing safely, probably quite often.

I care about this because alpha testing drawing is very expensive
on some mobile chips.
2013-08-27 21:01:17 +02:00
Unknown W. Brackets cd70250d8c Use more accessors, simplify softgpu colortest. 2013-08-24 11:34:38 -07:00