[Bug 90895] [IVB/HSW/BSW Bisected] SynMark2_v6_0_0_OglBatch0 to OglBatch4 performance reduce by ~23%

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=90895

--- Comment #5 from Eero Tamminen <eero.t.tamminen at intel.com> ---
Looking at intel_debug.h, the second "if" in perf_debug() macro is lacking
"unlikely" annotation and I wonder could either of the ifs contain the other as
perf_debug is not set in many places:
$ grep 'perf_debug *=' $(find -type f)
./src/mesa/drivers/dri/i915/intel_context.c:      intel->perf_debug = true;
./src/mesa/drivers/dri/i965/brw_context.c:      brw->perf_debug = true;
./src/mesa/drivers/dri/i965/intel_debug.c:      brw->perf_debug = true;
?

Looking further into perf_debug member and INTEL_DEBUG usage, the unlikely()
annotation usage is inconsistent:
$ grep '> *perf_debug.*)' $(find -type f)
...
$ grep '(.*INTEL_DEBUG.*)' $(find -type f) | grep -v unlikely

Maybe missing annotations could be added everywhere at the same time?


Seems weird that the indicated change causes such a huge performance impact
though.  Performance change itself is true, we're seeing it also in addition to
QA.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20150609/1f6f0821/attachment-0001.html>