summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri
AgeCommit message (Collapse)Author
2010-03-06dri: drop MINIGLX_SOURCES (2)George Sapountzis
2010-03-06dri: drop MINIGLX_SOURCESGeorge Sapountzis
2010-03-05i965: Only use the effective subset of the WM sampler key.Eric Anholt
The whole structure is 836 bytes, but if only the first one or two samplers are used (as is common), the part that matters is only 56 or 108 bytes. By using just that subset as the key (since the key size is also part of the key), we improve firefox-talos-gfx performance by 1%.
2010-03-05intel: Remove support for the DRI1 TFP extension.Eric Anholt
2010-03-05i965: Remove the paths for making a const surface object for a NULL BO.Eric Anholt
2010-03-05i965: Remove the unused SDC BO pointer from the sampler state key.Eric Anholt
Since it's a BO pointer, it already lives outside the key in brw->wm.sdc_bo[] which is used for sampler state lookup and setup.
2010-03-05i965: Rename SNB brw_*_key structs to gen6_*_key structs.Eric Anholt
2010-03-05i965: Pack gen6 blend_state_key structure.Eric Anholt
2010-03-05intel: Only do batchbuffer debug if --enable-debug is used.Eric Anholt
This saves 6.6KB on the 965 driver, and appears to speed firefox-talos-gfx up by 1-2%. Unlike many other asserts in the driver, when we make a mistake that would trigger one of these it generally shows up all the time for developers, so turning it off for release seems fine.
2010-03-05intel: Replace batch macro contents with function calls.Eric Anholt
This manages to cut down another 3800 bytes.
2010-03-05intel: Move the assertions about reloc delta from the macros to the function.Eric Anholt
Cuts another 1800 bytes from the driver.
2010-03-05intel: Check that the batch is mapped per BEGIN, not each OUT.Eric Anholt
Shaves 800 bytes off the driver.
2010-03-05i965: Try to assign SNB SF inputs to outputs.Eric Anholt
2010-03-05i965: Fix the SNB SF input size.Eric Anholt
2010-03-05Merge branch 'object-purgeable'Chris Wilson
Acked-by: Brian Paul <brianp@vmware.com>
2010-03-05APPLE_object_purgeable: intelChris Wilson
Implement support for purgeable objects by using the GEM madvise ioctl. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2010-03-04intel: Remove non-kernel-exec-fencing support.Eric Anholt
Shaves 60k off the driver from removing the broken spans code. This means we now require 2.6.29, which seems fair given that it's a year old and we've removed support for non-KMS already in the last release of 2D.
2010-03-04intel: Remove the unused s8 spans code. Not hit during no_rast piglit.Eric Anholt
Shaves 5.5k off of the driver.
2010-03-04dri/nouveau: Remove some CI remnants.Francisco Jerez
2010-03-04dri/nouveau: Trivially add GL_ARB_texture_mirrored_repeat.Andrew Randrianasulu
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
2010-03-04dri/nouveau: Enable GL_EXT_stencil_wrap.Andrew Randrianasulu
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
2010-03-04dri/nouveau: Add GL_EXT_stencil_wrap for nv04.Andrew Randrianasulu
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
2010-03-04dri/nouveau: Fix rb->DataType for z24s8 renderbuffers.Francisco Jerez
2010-03-04dri/nouveau: Try to keep client buffers smaller than the scratch VBO length.Francisco Jerez
2010-03-04dri/nouveau: Pack client arrays as they're copied to the real BO.Francisco Jerez
2010-03-04i915: Enable texture tiling by default.Eric Anholt
This now passes piglit testcases, and shows performance improvements on openarena. Blit-heavy apps may see degradation, but the expectation is that the common case of 3D rendering will see significant wins. This regresses gen-teximage on my 855, and no piglit tests on my 945.
2010-03-04i915: Fix up i830 for tiled drawing offsets.Eric Anholt
Corresponds to b87406e55f029d29594ae76a4b39a4fe1007fe4f.
2010-03-04DRI: Put back missing break-statementIan Romanick
The break in the __DRI_ATTRIB_RENDER_TYPE case was accidentally removed in commit 5cf2c5851bcd29c2d53bb04ab692b4b156f5a74d. This puts it back. Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
2010-03-04r300: make blit code work on swtcl machines.Dave Airlie
Only tested this on my rv530 with R300_NO_TCL=1, but it works there at least. It would be nice to fix the tex_offset by chaning the vertex program on tcl to read from offset 6 but this isn't really necessary. Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-04r300_blit: initialise whole fp structure to 0Dave Airlie
at least on the r500 we get a lockup unless I do this, the compiler seems to fail to compile and we just get a null fp. Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-03radeon/r200/r300/r600: add check_blit vtbl functionAlex Deucher
Check if the native blit formats are supported, if not, attempt to use an alternate format. Skip 3, >4 bpp as per comments from mcencora on irc. Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
2010-03-03Remove stray defines of HAVE_RGBAIan Romanick
Now that color-index support is removed from t_dd_tritmp.h and t_dd_unfilled.h, drivers no longer need define HAVE_RGBA. Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
2010-03-03mesa: Remove ClearIndex and IndexMask from device-driver interfaceIan Romanick
These are used to inform the driver of the clear value for color-index buffers and to control write-masking of bits in color-index buffers. No driver use or need (not even Nouveau) these interfaces. Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
2010-03-03DRI: Remove support for color-index visualsIan Romanick
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
2010-03-03DRI/swrast: Remove color-index support from DRI swrast driverIan Romanick
It appears that color-index rendering wasn't actually supported anyway. swrastFillInModes did not previously create an color-index configs, so it doesn't seem like there would be any way to get a color-index visual. Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
2010-03-03i915: Use x,y drawing offsets instead of changing buffer offsets.Eric Anholt
This should fix rendering into mipmaps of tiled textures.
2010-03-03i915: Tell the kernel when we actually need fence registers on our BOs.Eric Anholt
This improves tiled texture performance of OA on my 945 from 25.3fps to 29.0fps, whereas untiled is 28.2fps, by avoiding stalls for fence register changes.
2010-03-03intel: Use drm_intel_bo_alloc_tiled for region allocs.Eric Anholt
This moves the logic for how to align pitches, heights, and sizes of objects to one central location. Fixes rendering with texture tiling on i915. Note that current libdrm is required for the change for I915_TILING_NONE pitch alignment.
2010-03-03i915: Don't do the pitch expansion for tiled buffers.Eric Anholt
The weirdness that led to the bumping of pitch for those 512/1024 pixels is that in taking a 2x2 subspan or bilinear filtering we'd end up hitting the same channel in 2 different pages, leading to lower performance. With tiling, that doesn't occur, so we don't need to waste the memory.
2010-03-03intel: Use GTT mapping in the blit fallback path in case there's tiling.Eric Anholt
2010-03-03i965: Fix up Sandybridge VS sizing.Eric Anholt
2010-03-03r600: add new rs880 pci idAlex Deucher
2010-03-03intel: Set InternalFormat for renderbuffers created from an EGLImageKristian Høgsberg
2010-03-03radeon/r200/r300/r600: fix blit BO buffer validationDave Airlie
there is no need for these to be persistent since we flush the CS at the end of the blit, and we don't reuse these buffers at all. on my r600 the Wine DX9 test suite doesn't crash at least anymore. Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-02intel: passs correct context type to intel_prepare_render() callBrian Paul
2010-03-02i965: include macros.h to get CLAMP macroBrian Paul
2010-03-02r600: fix blender picking.Dave Airlie
This fixes the sw fallback for GL_SELECT picking modes. Fixes object picking blender + depthpick test http://bugs.freedesktop.org/show_bug.cgi?id=26419 Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-01radeon: Reduce number of radeon_bo_map calls that require kernel entry.Pauli Nieminen
DMA buffers are often remapped many times for singel cs. To reduce number of mapping calls ending to kernel dma buffers are mapped when allocated and unmapped when full or in cs flush. This makes indifual mapping calls in other parts of code simple increment/decrement reference count which is faster than entering kernel. This improves Anholt's openarena benchmark from 36 fps to 44 fps.
2010-03-01dri/nouveau: Combine the dummy texture more deterministically on nv0x.Francisco Jerez
2010-03-01dri/nouveau: Fix up the ADD texture environment on nv0x.Francisco Jerez
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-06 09:14:24 +0000