summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2009-10-25Fix behavior of wait_reg_time() in case of system time wrap.HEADmasterMatthias Hopf1-0/+4
2009-05-21extend the size of array for complex shaderCooper Yuan1-1/+1
2009-03-12Fix OFFSET_[XYZ] macro for TEX_DWORD2 to accept floatsYang Zhao1-3/+3
Values for OFFSET_[XYZ] are 5-bits two's-complement fixed-point with one-bit after decimal point. Values in [-8.0, 7.5] are valid. Inputs that do not exactly land on 0.5 increments are rounded towards 0 to the nearest increment.
2009-02-27convert_shader: add --r7xx optionAlex Deucher1-10/+17
Specify the native type of the shader being decoded
2009-02-27convert_shader: read shader from a fileCooper Yuan1-52/+24
2009-02-24Add wait_for_vsync().Matthias Hopf3-1/+19
2009-02-24Less verbosity with buffer renewal.Matthias Hopf1-2/+3
2009-02-24Use actually adapt->free* for object upload.Matthias Hopf1-1/+1
2009-02-24Add pointer for free memory to adapter_t.Matthias Hopf2-8/+5
Also nuke unused depth* entries.
2009-02-24Only print out full buffer contents on verbose >= 2.Matthias Hopf1-6/+10
2009-02-24Try to better detect display width.Matthias Hopf2-6/+10
Still doesn't work as intended.
2009-02-17Get new DRM command buffer if old one is filled up at least half.Matthias Hopf1-5/+41
2009-02-17Add wait for engine idle '/'.Matthias Hopf1-0/+4
2009-02-17Fix 'P' test - DRM buffer is too small to hold 2MB vertex list.Matthias Hopf1-2/+4
2009-02-17Split upload() into gart and gpu variant.Matthias Hopf2-18/+26
2009-02-16Add *really* waiting wait_3d_full_idle_clean() call, use for timing.Matthias Hopf2-2/+27
2009-02-16Fetch and save framebuffer size.Matthias Hopf2-2/+6
2009-02-16Move temporary test to r600_tmp.cMatthias Hopf3-8/+43
2009-02-16Workaround for bug in R600.Matthias Hopf1-2/+4
2009-02-16Allow more fine granular cache flushes.Matthias Hopf2-12/+13
2009-02-13Add M74, M82 chipsets (no pci ids yet).Matthias Hopf3-3/+11
2009-02-05We need the r6xx-r7xx-support branch of drm.Matthias Hopf1-1/+1
2009-02-03Fix op2 macro arguments for rv770 pm4 tri testAlex Deucher1-88/+88
2009-02-03Allow manual specification of pciid.Matthias Hopf1-2/+10
2009-02-03Add missing interpolator setup to ALU performance test.Matthias Hopf1-0/+9
2009-02-03Add continuously running full speed test 'F'.Matthias Hopf3-13/+23
2009-01-29Add M72 as known good chipset, correct num_qd_pipes calculation.Matthias Hopf3-5/+14
2009-01-29Start performance test with 3 ALU instructions per clause.Matthias Hopf1-1/+1
1 ALU instruction (5 entries) is never saturating the ALU units, and on RV770 you need a clause pretty full (approx. 50 of 128 entries) to saturate them, while on RV670 10 entries was enough. When starting with 15 entries, we can fill up 120 entries (24 ALU instructions), while when starting with 5 entries, we end up with a maximum of 80.
2009-01-29Use multiply-and-add MULADD instead of MUL for performance tests.Matthias Hopf1-46/+27
Getting actually close to marketing values.
2009-01-29Use trans ALU component as well during performance test.Matthias Hopf1-5/+30
2009-01-29Add ALU performance test.Matthias Hopf3-8/+413
2009-01-29Proper flush_gpu_source_cache().Matthias Hopf6-36/+58
Also called on upload(), no need for extra syncs. Remove cp_set_surface_sync(), except for _pm4.c
2009-01-27Use non-uniform background for clear.Matthias Hopf1-4/+11
2009-01-27Add textured triangle performance test 'P'.Matthias Hopf5-11/+464
2009-01-27Move create_sample_texture() to _lib.c. Don't swap components on textures.Matthias Hopf3-18/+20
2009-01-27Add comments to register file.Matthias Hopf1-2176/+2303
Helps with developing, especially when used with tags target.
2009-01-27Adapt to actually released register specs.Matthias Hopf2-79/+8
2009-01-06Add RS780 chip idsAlex Deucher1-0/+7
2009-01-02Add 0x944C as RV770.Matthias Hopf1-1/+1
Reported by Michael Larabel.
2008-12-29Initial public commit.Matthias Hopf25-0/+18732