[Bug 109216] 4-27% performance drop in Vulkan benchmarks

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Fri Jan 11 18:41:16 UTC 2019


https://bugs.freedesktop.org/show_bug.cgi?id=109216

--- Comment #16 from Ian Romanick <idr at freedesktop.org> ---
(In reply to Eero Tamminen from comment #15)
> (In reply to Ian Romanick from comment #14)
> > Created attachment 143048 [details] [review] [review]
> > intel/compiler: Try nir_opt_if once before nir_opt_peephole_select
> 
> This patch fixes the smaller perf regressions in (nbody & raytracing)
> compute tests, but it doesn't help at all with the largest regression, in
> Deferred Multisampling.

It appears that the inner loop still isn't getting unrolled because of the
problem this patch should have resolved.  I'll keep digging.

-- 
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/20190111/4447cb2a/attachment.html>


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