Summary: | Increased difference from reference image in shaders | ||
---|---|---|---|
Product: | Mesa | Reporter: | Ruslan Kabatsayev <b7.10110111> |
Component: | Drivers/DRI/i965 | Assignee: | Matt Turner <mattst88> |
Status: | RESOLVED FIXED | QA Contact: | Intel 3D Bugs Mailing List <intel-3d-bugs> |
Severity: | normal | ||
Priority: | medium | CC: | mattst88 |
Version: | 17.2 | ||
Hardware: | Other | ||
OS: | All | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: | patch |
Description
Ruslan Kabatsayev
2017-11-08 08:00:53 UTC
CCing author of the commit. Thanks for the bug report. I tried building precomputed_atmospheric_scattering, but I get: % make integration_test mkdir -p output/Release/atmosphere/reference g++ -Wall -Wmain -pedantic -pedantic-errors -std=c++11 -I. -Iexternal -Iexternal/dimensional_types -Iexternal/glad/include -Iexternal/progress_bar -DNDEBUG -O3 -fexpensive-optimizations -c atmosphere/reference/functions.cc -o output/Release/atmosphere/reference/functions.o In file included from ./atmosphere/reference/functions.h:42:0, from atmosphere/reference/functions.cc:39: ./atmosphere/reference/definitions.h:53:24: fatal error: math/angle.h: No such file or directory #include "math/angle.h" ^ compilation terminated. make: *** [Makefile:113: output/Release/atmosphere/reference/functions.o] Error 1 Searching for other include files under math/, like "math/ternary_function.h" on Google I only see references to your project. Help? (In reply to Matt Turner from comment #2) > ./atmosphere/reference/definitions.h:53:24: fatal error: math/angle.h: No > such file or directory > #include "math/angle.h" Please do `git submodule update --init` (or clone the repo with --recursive option) to fetch the required files. After that, to build only integration test, you can say `make integration_test`. After building you'll have to wait several minutes while the CPU model generates data, but then after this run you can re-run the test as needed without delays. Created attachment 135615 [details] [review] patch Thanks. I can build it and run the integration tests now. (In reply to Ruslan Kabatsayev from comment #0) > I tested on Xeon E3-1226 v3 built-in GPU, as well as on Core i7-4765T Both of those are Haswells, FWIW. Testing on a Skylake, all 13 tests fail regardless of the fix. I'll get my Haswell out and see what it looks like there. The patch works great on my Core i7-4765T (all tests pass). (In reply to Ruslan Kabatsayev from comment #5) > The patch works great on my Core i7-4765T (all tests pass). Great! Thank you. I've sent that patch to the mailing list for review along with some new unit test cases for the case that was broken. Thanks again for the bug report and bisection! Fixed by commit a05af1f7b8f82a38513bba31f9573cd62d82f18d Author: Matt Turner <mattst88@gmail.com> Date: Mon Nov 20 14:24:57 2017 -0800 i965/fs: Handle negating immediates on MADs when propagating saturates MADs don't take immediate sources, but we allow them in the IR since it simplifies a lot of things. I neglected to consider that case. Fixes: 4009a9ead490 ("i965/fs: Allow saturate propagation to propagate negations into MADs.") Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=103616 Reported-and-Tested-by: Ruslan Kabatsayev <b7.10110111@gmail.com> Reviewed-by: Ian Romanick <ian.d.romanick@intel.com> and tested by commit beaea7abfa9f25460284f739fa50050a889ce6cd Author: Matt Turner <mattst88@gmail.com> Date: Mon Nov 20 14:21:43 2017 -0800 i965/fs: Check ADD/MAD with immediates in satprop unit test The gen had to be changed from 4 to 6 so that we could test MAD, which is new on Gen6. mad_imm_float_neg_mov_sat tests the case fixed by the previous commit. Reviewed-by: Ian Romanick <ian.d.romanick@intel.com> |
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.