[Bug 107510] [GEN8+] up to 10% perf drop on several 3D benchmarks

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Mon Feb 18 09:31:39 UTC 2019


--- Comment #22 from Eero Tamminen <eero.t.tamminen at intel.com> ---
(In reply to Kenneth Graunke from comment #20)
> There are too many things being tracked in this bug for me to know whether
> it's fixed or not.

This bug tracks just regression coming from this commit:
commit 8d8222461f9d7f497d657c2c0eff70820986429b
Author: Jason Ekstrand <jason.ekstrand at intel.com>
Date:   Mon Jul 23 22:20:41 2018 -0700
    intel/nir: Enable nir_opt_find_array_copies

Which re-introduces back the same issue that got fixed in Jason's revert of
"intel/nir: Call nir_lower_io_to_scalar_early" commit.

Timothy's patch series fixing the issue (but being explicitly limited to
tess/geom shaders) with following explanation:
   Commit 8d8222461f9d7f49 caused substantially more URB messages in
   geometry and tessellation shaders (due to enabling
   nir_lower_io_to_scalar_early). This combines io again to avoid
   this regression while still allowing link time optimisation of

> Vulkan regressions are likely fixed by
> 39aee57523a02552e7eae7df5da488e535aeb1eb (anv: Put MOCS in the correct
> location).  But, that wouldn't affect any GL programs, so any GL regressions
> may still exist.

If your commit didn't fix them, IMHO this bug is most likely explanation for
tessellation regression.

You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20190218/21a95bf6/attachment.html>

More information about the intel-3d-bugs mailing list