Eric Anholt [Wed, 20 Apr 2011 18:48:31 +0000 (11:48 -0700)]
intel: Add support for ARB_sampler_objects.
This extension support consists of replacing
"gl_texture_obj->Sampler." with "_mesa_get_samplerobj(ctx, unit)->".
One instance of referencing the texture's base sampler remains in the
initial miptree allocation, where I'm not sure we have a clear
association with any texture unit.
Tested with piglit ARB_sampler_objects/sampler-objects.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Wed, 20 Apr 2011 04:40:48 +0000 (21:40 -0700)]
i965: Add support for NV_conditional_render.
Since we lack hardware support for it, this is a simple matter of
checking _mesa_check_conditional_render at the entrypoints, and
suppressing it for the metaops where it doesn't apply.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Thu, 21 Apr 2011 23:52:21 +0000 (16:52 -0700)]
swrast: Disable glAccum drawing during conditional rendering.
Eric Anholt [Wed, 20 Apr 2011 07:21:01 +0000 (00:21 -0700)]
meta: Don't do conditional rendering on GenerateMipmaps and BlitFramebuffer.
The NV_conditional_render spec calls out specific operations that
conditional rendering applies to, which doesn't include these.
Fixes NV_conditional_render/generatemipmap on swrast.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Mon, 18 Apr 2011 21:28:14 +0000 (14:28 -0700)]
i965: Add support for ARB_texture_compression_rgtc.
Tested with rgtc-teximage-0[12].
EXT_texture_compression_rgtc/fbo-generatemipmap-formats fails in NPOT
just like S3TC does.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Wed, 20 Apr 2011 16:33:59 +0000 (09:33 -0700)]
mesa: Don't assert on the compressed convertformat for GenerateMipmaps.
This assertion doesn't make any sense to me -- the convertFormat is
already something valid (tested above), and the BaseFormat dictated by
convertFormat doesn't matter to the function about to be called (it's
the datatype/comps that were pulled out of convertFormat).
Fixes assertion failure in
GL_EXT_texture_compression_rgtc/fbo-generatemipmap-formats
(still has a rendering failure in NPOT like S3TC does).
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Wed, 20 Apr 2011 16:31:34 +0000 (09:31 -0700)]
mesa: Choose RGTC formats for GL_COMPRESSED_RED, GL_COMPRESSED_RG.
We were falling through to the default R8 and RG88 formats instead of
compressing when possible. Noticed by swrast fbo-blending-formats
actually doing rendering.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
José Fonseca [Sat, 23 Apr 2011 19:25:11 +0000 (20:25 +0100)]
mesa/gdi: Address compiler warnings.
In particular the function prototypes.
José Fonseca [Sat, 23 Apr 2011 17:44:19 +0000 (18:44 +0100)]
mesa/gdi: Remove InitCritSections.cpp.
We are now able to declare global critical sections through other
mechanisms so this is useless code.
José Fonseca [Sat, 23 Apr 2011 12:21:41 +0000 (13:21 +0100)]
Drop fx windows driver.
Irrelevant now that glide driver was removed.
Jakob Bornecrantz [Sat, 23 Apr 2011 15:01:19 +0000 (17:01 +0200)]
glu: Fix linux build
Jakob Bornecrantz [Sat, 23 Apr 2011 14:57:28 +0000 (16:57 +0200)]
glu: Clone the enviroment
José Fonseca [Sat, 23 Apr 2011 12:18:45 +0000 (13:18 +0100)]
docs: Remove MSVS project files.
They were totally broken for several releases.
scons now builds everything the project files built and more, and can be
kept up-to-date with little effort.
José Fonseca [Sat, 23 Apr 2011 12:16:03 +0000 (13:16 +0100)]
scons: Build glu.
José Fonseca [Sat, 23 Apr 2011 11:41:28 +0000 (12:41 +0100)]
scons: Build classic mesa gdi driver.
Build as
scons platform=windows mesagdi
José Fonseca [Sat, 23 Apr 2011 11:23:42 +0000 (12:23 +0100)]
osmesa: Fix Mingw build.
Build as
scons platform=windows osmesa
José Fonseca [Sat, 23 Apr 2011 11:22:59 +0000 (12:22 +0100)]
scons: Build osmesa.
Just type
scons osmesa
José Fonseca [Sat, 23 Apr 2011 11:14:39 +0000 (12:14 +0100)]
mapi/glapi: Avoid warnings about visibility being unsupported on MinGW.
José Fonseca [Sat, 23 Apr 2011 10:49:41 +0000 (11:49 +0100)]
mesa/vf: Remove.
Unused. Probably replaced by translate module.
José Fonseca [Sat, 23 Apr 2011 09:44:43 +0000 (10:44 +0100)]
graw: Replace dead symlinks to delete python statetracker shaders with their contents.
Brian Paul [Fri, 22 Apr 2011 19:37:10 +0000 (13:37 -0600)]
draw: fix point/line/tri flushing bug in vbuf code
Need to reset the point/line/tri functions to point to the "first"
versions whenever we flush vertices. Fixes unfilled polygon rendering
errors seen in demos/samples/logo.c. See comments for more info.
NOTE: This is a candidate for the 7.10 branch.
Marek Olšák [Fri, 22 Apr 2011 18:19:38 +0000 (20:19 +0200)]
r300g: do not advertise color_buffer_float on r300 and r400
It seems they can't do unclamped vertex colors. Tested on RV350.
Marek Olšák [Fri, 22 Apr 2011 16:54:02 +0000 (18:54 +0200)]
r300/compiler: fix up error message
Benjamin Franzke [Mon, 7 Mar 2011 14:10:01 +0000 (15:10 +0100)]
egl/wayland: Follow buffer.attach removal
Benjamin Franzke [Mon, 7 Mar 2011 14:06:31 +0000 (15:06 +0100)]
egl/wayland: Update to buffer.damage addition
Marek Olšák [Fri, 22 Apr 2011 14:55:54 +0000 (16:55 +0200)]
configure.ac: fix up the help text for --enable-gallium-r300
Marek Olšák [Fri, 22 Apr 2011 14:17:18 +0000 (16:17 +0200)]
r300g: disallow A16F,L16F,LA16F,I16F texture formats on DRM<2.8.0
Benjamin Franzke [Thu, 21 Apr 2011 08:51:44 +0000 (10:51 +0200)]
st/egl: Update to wayland protocol,egl changes
Benjamin Franzke [Thu, 21 Apr 2011 08:51:14 +0000 (10:51 +0200)]
egl_dri2: Create wl_buffers for pixmap surfaces
Needed since wayland-egl doesnt do this anymore.
Benjamin Franzke [Thu, 21 Apr 2011 09:14:02 +0000 (11:14 +0200)]
egl/wayland: Add new version parameter for post_drm
Benjamin Franzke [Thu, 21 Apr 2011 07:39:25 +0000 (09:39 +0200)]
egl/wayland: Fix prefixes for interface names
Marek Olšák [Wed, 20 Apr 2011 10:40:40 +0000 (12:40 +0200)]
swrast: fix readpix clamping
Broken with
e5c6a92a12b5cd7db205d72039f58d302b0be9d5. (ARB_color_buffer_float)
Clamping should occur if type != float, otherwise the MSBs of the resulting
pixels are killed off. For example, reading back LUMINANCE = R+G+B can be
greater than 0xff, but the result is naturally masked by 0xff
for UNSIGNED_BYTE, leading to bogus results.
The following bug report seems to want clamping to occur if type == half_float
too. Not sure what's correct.
Bug: [bisected pineview] oglc case pxconv-read failed
https://bugs.freedesktop.org/show_bug.cgi?id=35852
Tested by: Fang Xun <xunx.fang@intel.com>
Reviewed-and-tested-by: Ian Romanick <ian.d.romanick@intel.com>
Kristian Høgsberg [Fri, 22 Apr 2011 03:47:36 +0000 (23:47 -0400)]
WL_bind_wayland_display: clarify that context must be NULL
Ian Romanick [Sat, 16 Apr 2011 05:20:09 +0000 (22:20 -0700)]
i915: Gut all remaining bits of hardware fog
None of this ever gets used. Fog is always calculated by a fragment
program. Even though the fixed-function fog unit is never used, state
updates are still sent to the hardware. Removing those spurious state
updates can't hurt performance.
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 02:25:59 +0000 (19:25 -0700)]
i915: i915_context::vertex_fog is always I915_FOG_NONE, so kill it
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 02:23:13 +0000 (19:23 -0700)]
i915: There's always a fragment program
Fragment programs are generated by core Mesa for fixed-function.
Because of this, there's no reason to handle cases where there is no
fragment program for fog.
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 02:14:36 +0000 (19:14 -0700)]
i915: Delete disabled try_pixel_fog paths
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 02:04:57 +0000 (19:04 -0700)]
mesa: Kill gl_fragment_program::FogOption with fire
All drivers expect this to always be GL_NONE. Don't let there be any
opportunity for a bad value to leak out and infect some unsuspecting
driver. If any driver for hardware that had fixed-function
per-fragment fog (i915 and perhaps some r300-ish) was ever going to
add support, it would have done it by now.
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 01:46:57 +0000 (18:46 -0700)]
i915: gl_fragment_program::FogOption is always GL_NONE so don't check it
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 01:46:43 +0000 (18:46 -0700)]
i965: gl_fragment_program::FogOption is always GL_NONE so don't check it
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 01:45:06 +0000 (18:45 -0700)]
mesa: gl_fragment_program::FogOption is always GL_NONE so don't check it
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Ian Romanick [Sat, 16 Apr 2011 01:38:54 +0000 (18:38 -0700)]
mesa: Fix bugs in ff fragment shader fog handling
This patch fixes two bugs related to fog in the fixed-function
fragment shader generation code.
Fog was only lowered to instructions if MRTs were used. The fragment
shader assembler always lowers "fog option" code to instructions, and
many drivers (e.g., r300) expect this.
When fog lowering did happen, it was after the instruction count was
checked against implementation limits. Since fog lowering may add up
to 5 instructions, a program that was below the limits before lowering
may exceed the limits after lowering.
NOTE: This is a candidate for the stable branches.
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Brian Paul [Thu, 21 Apr 2011 18:59:16 +0000 (12:59 -0600)]
st/mesa: check image size before copy_image_data_to_texture()
We should only copy images into the dest texture if the size is correct.
This fixes a failed assertion when finalizing a texture with mis-defined
mipmap levels such as:
level 0: 32x32
level 1: 8x8
Also, fix incorrect mipmap level used in assertion at the top of
copy_image_data_to_texture().
NOTE: This is a candidate for the 7.10 branch.
Marek Olšák [Thu, 21 Apr 2011 13:13:34 +0000 (15:13 +0200)]
galahad,util: warn on resource target mismatch in copy_region
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Marek Olšák [Thu, 21 Apr 2011 10:54:56 +0000 (12:54 +0200)]
st/mesa: implement CopyBufferSubData using resource_copy_region
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Marek Olšák [Thu, 21 Apr 2011 10:53:16 +0000 (12:53 +0200)]
gallium: add fallback for copying buffers to all drivers
Just to keep drivers working.
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Marek Olšák [Thu, 21 Apr 2011 10:50:00 +0000 (12:50 +0200)]
util: add a simple memcpy path for copying buffers in util_resource_copy_region
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Kenneth Graunke [Fri, 15 Apr 2011 00:21:59 +0000 (17:21 -0700)]
glsl/ast_to_hir: Only use the local 'type' variable as a temporary.
Lots of code (deleted by this patch) tried to make type == result->type,
but not all cases did. Don't pretend; just use result->type.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 19 Apr 2011 22:33:15 +0000 (15:33 -0700)]
i965: Remove dead vertex buffer structs.
We do this OUT_BATCH-style in brw_draw_upload.c.
Kristian Høgsberg [Thu, 14 Apr 2011 14:42:41 +0000 (10:42 -0400)]
wayland-egl: Update to wayland-egl.h changes, drop struct wl_egl_display
Brian Paul [Wed, 20 Apr 2011 18:04:25 +0000 (12:04 -0600)]
mesa: fix void pointer arithmetic warning
Eric Anholt [Mon, 18 Apr 2011 20:45:00 +0000 (13:45 -0700)]
docs: Mention that GLSL 1.30 is started.
Things definitely remaining todo: switch statements, clip distances.
On 965, we also need real integers in the VS, and implementations of
some things like isinf/isnan.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Mon, 18 Apr 2011 20:44:35 +0000 (13:44 -0700)]
docs: Update for Intel ARB_texture_float and ARB_color_buffer_float.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 21:40:09 +0000 (14:40 -0700)]
intel: Add support for ARB_color_buffer_float.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 21:39:47 +0000 (14:39 -0700)]
meta: Add support for ARB_color_buffer_float to _mesa_meta_Clear().
Tested with piglit arb_color_buffer_float-clear.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 22:48:32 +0000 (15:48 -0700)]
meta: Add support for ARB_color_buffer_float to _mesa_meta_DrawPixels.
Tested with piglit arb_color_buffer_float-drawpixels.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 18:00:00 +0000 (11:00 -0700)]
intel: Add support for ARB_texture_float.
For 1 and 2-channel formats the hardware only supports rendering to R
and RG. To do I and L render targets we just call them R and
everything works out. For A, we would need to rewrite the CC to do
the alpha channel's blending on color instead, and send the fragment
alpha down the red channel. For LA, there doesn't seem to be any
hope, because we can't do independent color/alpha blending while
treating the LA surface as RG.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Sat, 16 Apr 2011 01:15:59 +0000 (18:15 -0700)]
intel: Add support for blit copies of >32bpp formats.
The blitter only does up 32bpp at a time, so we handle it by mangling
coordinates and calling the surface 32bpp.
Fixes ARB_texture_rg/fbo-generatemipmap-formats-float with ARB_texture_float.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Mon, 18 Apr 2011 17:53:06 +0000 (10:53 -0700)]
mesa: Add renderbuffer accessors for A, I, L, FLOAT32.
Of these, intel will be using I and L initially, and A once we rewrite
fragment shaders and the CC for rendering to it as R.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Mon, 18 Apr 2011 16:17:00 +0000 (09:17 -0700)]
mesa: Add renderbuffer accessors for MESA_FORMAT_R_FLOAT32, RG_FLOAT32.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 21:18:43 +0000 (14:18 -0700)]
mesa: Add ChooseTexFormat fallbacks for floating point textures.
This covers X_FLOAT16 -> X_FLOAT32, and X -> RGBA_FLOAT32.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 20:36:40 +0000 (13:36 -0700)]
mesa: Add renderbuffer accessors for MESA_FORMAT_RGBA_FLOAT32.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Mon, 18 Apr 2011 16:32:06 +0000 (09:32 -0700)]
swrast: Add LUMINANCE, INTENSITY, LUMINANCE_ALPHA to span asserts.
Fixes ARB_texture_float/fbo-alphatest-formats.
Reviewed-by: Brian Paul <brianp@vmware.com>
Alex Deucher [Wed, 20 Apr 2011 17:09:09 +0000 (13:09 -0400)]
r600g: fix userspace fences again
reinstate
b7617346dcff50a66a10c61b95c33682cf629c9e after the
rework in
6067a2a67f9a7aab2aee051469bea8af03747a95.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Fredrik Höglund [Tue, 19 Apr 2011 22:21:42 +0000 (00:21 +0200)]
r600g: don't flush the dest caches on every draw
Keep track of when the caches are dirty, and only flush them when
the framebuffer state is set and when the context is flushed.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Brian Paul [Tue, 19 Apr 2011 22:13:43 +0000 (16:13 -0600)]
translate: disable clamping of instanced array indexes
This fixes piglit's draw-instanced-divisor test for softpipe on both
the generic and SSE paths. This is temporary until we have the
correct per-array max_index information.
Alex Deucher [Tue, 19 Apr 2011 17:35:19 +0000 (13:35 -0400)]
r600g: add evergreen+ big endian support
Based on Cédric's r6xx/r7xx patch.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cédric Cano [Tue, 19 Apr 2011 17:02:14 +0000 (13:02 -0400)]
r600g: add big endian support for r6xx/r7xx
Signed-off-by: Cedric Cano <ccano@interfaceconcept.com>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Alex Deucher [Tue, 19 Apr 2011 16:03:55 +0000 (12:03 -0400)]
r600c: add evergreen big endian support
Based on Cedric's r6xx/r7xx patch.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cédric Cano [Tue, 19 Apr 2011 15:14:23 +0000 (11:14 -0400)]
r600c: add big endian support for r6xx/r7xx
Signed-off-by: Cedric Cano <ccano@interfaceconcept.com>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Brian Paul [Tue, 19 Apr 2011 15:07:44 +0000 (09:07 -0600)]
Makefile: add missing Scons files
Thierry Vignaud [Tue, 19 Apr 2011 14:34:06 +0000 (08:34 -0600)]
Makefile: add missing .cpp, .ll, .yy files
Signed-off-by: Brian Paul <brianp@vmware.com>
Dave Airlie [Tue, 19 Apr 2011 10:42:48 +0000 (20:42 +1000)]
r600g: add cb support for snorm formats.
Check for signed type and enable SNORM.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Marek Olšák [Tue, 19 Apr 2011 10:28:51 +0000 (12:28 +0200)]
configure.ac: rename --enable-gallium-radeon to --enable-gallium-r300
Also fix up the help string for both r300 and r600.
Marek Olšák [Sun, 17 Apr 2011 21:41:33 +0000 (23:41 +0200)]
r200: enable some extensions
Such as:
- GL_ARB_half_float_pixel
- GL_ARB_vertex_array_object
- GL_APPLE_vertex_array_object
- GL_EXT_gpu_program_parameters
Dave Airlie [Tue, 19 Apr 2011 08:44:43 +0000 (18:44 +1000)]
r600g: fix warnings with piglit texture-rg test.
Fill in the swaps for the formats to fix warnings.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 19 Apr 2011 06:39:50 +0000 (16:39 +1000)]
r600g: make loop const always flush
this needs revisiting, we really don't want to be flushing all 32 of these,
but currently we don't flush any of them, and it seems to have caused a regression
as reported on irc with doom3 on evergreen.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Tom Stellard [Tue, 19 Apr 2011 04:16:14 +0000 (21:16 -0700)]
r300/compiler: Fix dataflow analysis bug with ELSE blocks
Writes within ELSE blocks were being ignored which prevented us from
discovering all possible writers for some register values.
Fixes piglit glsl-fs-raytrace-bug27060
Bryan Cain [Mon, 18 Apr 2011 06:39:08 +0000 (01:39 -0500)]
glsl: fix conversions from uint to bool and from float/bool to uint
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Dave Airlie [Tue, 19 Apr 2011 00:08:18 +0000 (10:08 +1000)]
r600g: add dirty tracking to context reg.
just makes the code more consistent.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 19 Apr 2011 00:04:02 +0000 (10:04 +1000)]
r600g: deinline some large functions.
really at these sort of sizes these are pointless inlines.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 19 Apr 2011 00:00:03 +0000 (10:00 +1000)]
r600g: consolidate r600/evergreen code for resource emission.
These really didn't have much difference, and totally not inline material.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Bas Nieuwenhuizen [Wed, 13 Apr 2011 18:36:15 +0000 (20:36 +0200)]
r600g: don't flush caches if we already did so, even for a subset of the flags
Merging the flushes that are left doesn't seem to give a significant
performance improvement
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Mon, 18 Apr 2011 03:37:10 +0000 (13:37 +1000)]
r600g: attempt to avoid emitting resources that are the same
This just avoids reemitting resources that haven't changed.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Mon, 18 Apr 2011 03:03:06 +0000 (13:03 +1000)]
r600g: modify block to only emit the first few dirty registers. (v2)
This gets me from 2200 to 1978 dwords for a gears frame.
This is due to us having some 32-dwords blocks in the SPI, that we only
modify the first dwords off.
v2: fix dirty reg count from Bas Nieuwenhuizen
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Sun, 17 Apr 2011 07:35:44 +0000 (17:35 +1000)]
r600g: track dirty registers better. (v2)
This is a first step to decreasing the CPU usage, by decreasing how much
stuff we pass to the GPU and hence to the kernel CS checker.
This adds a check to see if the values we need to write are actually dirty,
and avoids writing if they are. However certain register need to always
be written so we add a new flag to say which ones should be always written
if used. (Note this could probably be done cleaner with a larger refactoring,
since I think the CONST_BUFFER_SIZE_PS/VS and CONST_CACHE_PS/VS might
be better off as a special state).
It also moves the need_bo to be a flags on the register now.
With this, a frame of gears goes from emitting 3k dwords to emitting 2k dwords,
and I'm sure it could get a lot smaller.
v2: fix some evergreen dirty bits.
Original patch from: Bas Nieuwenhuizen, I NIHed nearly the same thing
before seeing his patch on the list, oops.
Reviewed-by: Bas Nieuwenhuizen
Signed-off-by: Dave Airlie <airlied@redhat.com>
Ian Romanick [Tue, 19 Apr 2011 00:06:08 +0000 (17:06 -0700)]
glsl: 80-column wrapping and whitespace fixes
Kenneth Graunke [Thu, 14 Apr 2011 21:37:46 +0000 (14:37 -0700)]
i965: Convert 3DPRIMITIVE command from struct-style to OUT_BATCH style.
Most of the newer portions of the code use OUT_BATCH style. I prefer
this style because it offers a clear distinction between a) hardware
messages/structures with a mandatory format, and b) data structures for
our own internal use that we can format however we want.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Tue, 12 Apr 2011 22:42:06 +0000 (15:42 -0700)]
i965: Allocate the whole URB to the VS and fix calculations for Gen6.
Since we never enable the GS on Sandybridge, there's no need to allocate
it any URB space.
Furthermore, the previous calculation was incorrect: it neglected to
multiply by nr_vs_entries, instead comparing whether twice the size of
a single VS URB entry was bigger than the entire URB space. It also
neglected to take into account that vs_size is in units of 128 byte
blocks, while urb_size is in bytes.
Despite the above problems, the calculations resulted in an acceptable
programming of the URB in most cases, at least on GT2.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Mon, 11 Apr 2011 17:10:30 +0000 (10:10 -0700)]
glsl: Emit a warning when the left-hand operand of a comma has no effect
The expression
x = y, 5, 3;
will generate
0:7(9): warning: left-hand operand of comma expression has no effect
The warning is only emitted for the left-hand operands, becuase the
right-most operand is the result of the expression. This could be
used in an assignment, etc.
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Brian Paul [Mon, 18 Apr 2011 18:59:28 +0000 (12:59 -0600)]
mesa: fix void pointer arithmetic warnings
And fix a couple logic errors in the put_*_generic() functions.
Brian Paul [Mon, 18 Apr 2011 18:43:42 +0000 (12:43 -0600)]
mesa: fix wrong parameter type in set_sampler_max_lod()
Brian Paul [Mon, 18 Apr 2011 14:19:42 +0000 (08:19 -0600)]
nvfx: comment-out unused var
Eric Anholt [Sun, 17 Apr 2011 17:44:46 +0000 (10:44 -0700)]
intel: Add I8 and L8 to intel_mesa_format_to_rb_datatype().
Fixes warnings in fbo-storage-formats.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Sat, 16 Apr 2011 03:00:38 +0000 (20:00 -0700)]
Revert "intel: Add spans code for the ARB_texture_rg support."
This reverts what remains of commit
28bab24e1698843e27d27204a1117066e7ffeabb. It was garbage, trying to
use a MESA_FORMAT enum as a preprocessor token, and I don't know how I
thought it was even tested.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Sat, 16 Apr 2011 02:58:21 +0000 (19:58 -0700)]
intel: Use mesa core's R8, RG88, R16, RG1616 RB accessors.
Fixes:
ARB_texture_rg/fbo-alphatest-formats
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Sat, 16 Apr 2011 02:55:38 +0000 (19:55 -0700)]
swrast: Don't try to adjust_colors for <8bpc when handling R16, RG1616.
The GL_RED and GL_RG were tricking this code into executing, but it's
totally unprepared for a 16-bit channel and just rescaled the values
down to 0. We don't have anything with <8bit channels alongside >8bit
channels, so disabling it should be safe.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Sat, 16 Apr 2011 02:11:53 +0000 (19:11 -0700)]
mesa: Add renderbuffer accessors for R8/RG88/R16/RG1616.
This will replace the current (broken by trying to use an enum in the
preprocessor) spantmp2.h support I wrote for the intel driver.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 20:09:21 +0000 (13:09 -0700)]
mesa: Use _mesa_get_format_bytes to refactor out the RB get_row_*
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 15 Apr 2011 19:58:17 +0000 (12:58 -0700)]
mesa: Use _mesa_get_format_bytes to refactor out the RB get_pointer_*
Reviewed-by: Brian Paul <brianp@vmware.com>