Age | Commit message (Collapse) | Author | Files | Lines | |
---|---|---|---|---|---|
2016-01-14 | relative dst supportHEADmaster | Rob Clark | 3 | -4/+19 | |
2016-01-14 | raise some limits | Rob Clark | 1 | -4/+4 | |
2015-05-08 | ir3emu: more fixes | Rob Clark | 2 | -4/+28 | |
2015-04-18 | ir3emu: add some more instructions | Rob Clark | 1 | -0/+19 | |
2015-04-18 | ir3emu: initial cat6 support (ldg) | Rob Clark | 4 | -66/+92 | |
2015-04-04 | ir3emu: some fixes | Rob Clark | 1 | -32/+55 | |
2015-04-01 | update (passing) reference shaders w/ @const | Rob Clark | 434 | -45588/+30858 | |
for passing shaders, where reference shader did not contain @const, copy the output shader over. $ir3compiler $args $f >& $out ./ir3test $ref $out res=$? if [ $res = 0 ]; then pass=$((pass + 1)) + grep @const $out > /dev/null + outconst=$? + grep @const $ref > /dev/null + refconst=$? + if [ $outconst = 0 ] && [ $refconst != 0 ]; then + cp $out $ref + fi else | |||||
2015-04-01 | ir3test: add support for @const immediates | Rob Clark | 6 | -5/+69 | |
This lets the compiler tell us where the immediates are and what values they should have, so that we can test two different compiler outputs from same tgsi src, even if there are changes in where/if immediates are placed in const registers. | |||||
2015-04-01 | ir3emu: some fixes | Rob Clark | 1 | -5/+18 | |
2015-01-23 | ir3emu: sorta emulate texture sam instructions better | Rob Clark | 2 | -2/+29 | |
Don't completely ignore all the other src registers. Otherwise we could miss some RA bugs. | |||||
2015-01-23 | add relative dest test | Rob Clark | 2 | -0/+249 | |
2015-01-12 | add firefox OTMC shaders | Rob Clark | 120 | -0/+6749 | |
2015-01-06 | add more 0ad shaders | Rob Clark | 286 | -0/+26362 | |
2015-01-05 | test updates | Rob Clark | 20 | -168/+1448 | |
Drop duplicate chrome shader, add a handful of shaders extracted from piglit tests. | |||||
2015-01-01 | fix shadertoy-frag.tgsi | Rob Clark | 1 | -1612/+0 | |
Some stray junk ended up in the .tgsi file... not noticed so far since we can't actually handle this shader yet. But some day.. | |||||
2015-01-01 | add relative-temp tests | Rob Clark | 6 | -0/+574 | |
2015-01-01 | ir3emu: couple fixes | Rob Clark | 1 | -10/+8 | |
2014-11-01 | add a set of 0ad shaders | Rob Clark | 339 | -0/+22612 | |
2014-11-01 | half-precision for builtin frag shaders | Rob Clark | 2 | -19/+19 | |
2014-11-01 | handle half vs full for inputs/outputs | Rob Clark | 2 | -4/+16 | |
2014-11-01 | few more opcodes | Rob Clark | 1 | -1/+13 | |
2014-11-01 | quick/dirty dsy/dsx | Rob Clark | 1 | -0/+2 | |
2014-11-01 | partially handle rpt's | Rob Clark | 1 | -30/+37 | |
At least enough to handle (rptN)nop.. since compiler actually emits those.. | |||||
2014-11-01 | update reference shaders | Rob Clark | 274 | -15041/+22845 | |
regenerate w/ --nocp since that is easier than visually inspecting larger shaders, and gives correct output in a few cases where the original RA would fail. We don't want shaders that are fixed by new RA to be flagged as fail due to incorrect reference. | |||||
2014-10-25 | remove fake 'kill' outputs | Rob Clark | 6 | -7/+0 | |
ir3_compiler was inadvertantly listing kill's in the outputs | |||||
2014-10-25 | initial commit | Rob Clark | 680 | -0/+39995 | |