On a GTX 650, the following 6 pow tests fail with the nouveau drivers but pass with the nvidia proprietary driver: https://www.khronos.org/registry/webgl/conformance-suites/1.0.2/conformance/ogles/GL/pow/pow_009_to_016.html test: pow_float_frag_xconsthalf_yvary.test.html FAIL pixel @ (33, 33 was [45,0,0,255] expected [16,0,0,255] test: pow_vec2_frag_xconsthalf_yvary.test.html FAIL pixel @ (33, 33 was [45,90,0,255] expected [16,255,0,255] test: pow_vec3_vert_xconsthalf_yvary.test.html FAIL pixel @ (33, 33 was [45,90,90,255] expected [16,255,255,255] https://www.khronos.org/registry/webgl/conformance-suites/1.0.2/conformance/ogles/GL/pow/pow_017_to_024.html test: pow_float_vert_xconsthalf_yvary.test.html FAIL pixel @ (33, 33 was [45,0,0,255] expected [16,0,0,255] test: pow_vec2_vert_xconsthalf_yvary.test.html FAIL pixel @ (33, 33 was [45,90,0,255] expected [16,255,0,255] test: pow_vec3_vert_xconsthalf_yvary.test.html FAIL pixel @ (33, 33 was [45,90,90,255] expected [16,255,255,255]
Confirmed on a GF108. Need to piglit this so that we can look at what's going on without running a web browser...
Thanks for the report. As I suspected, nothing actually to do with pow, just that the test, as written, happens to tickle a nouveau optimization bug. Should be fixed by: http://patchwork.freedesktop.org/patch/45657/ Still need to run it through piglit.
Fix pushed in 49b86007aa. Thanks for reporting!
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.