[Bug 104163] [GEN9+] 2-3% perf drop in GfxBench Manhattan 3.1 from "i965: Disable regular fast-clears (CCS_D) on gen9+"

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=104163

            Bug ID: 104163
           Summary: [GEN9+] 2-3% perf drop in GfxBench Manhattan 3.1 from
                    "i965: Disable regular fast-clears (CCS_D) on gen9+"
           Product: Mesa
           Version: git
          Hardware: Other
                OS: All
            Status: NEW
          Severity: normal
          Priority: medium
         Component: Drivers/DRI/i965
          Assignee: intel-3d-bugs at lists.freedesktop.org
          Reporter: eero.t.tamminen at intel.com
        QA Contact: intel-3d-bugs at lists.freedesktop.org

This commit:
----------------------------------------------------
commit ee57b15ec764736e2d5360beaef9fb2045ed0f68
Author:     Jason Ekstrand <jason.ekstrand at intel.com>
AuthorDate: Wed Nov 29 16:22:42 2017 -0800
Commit:     Jason Ekstrand <jason.ekstrand at intel.com>
CommitDate: Fri Dec 1 10:14:28 2017 -0800

    i965: Disable regular fast-clears (CCS_D) on gen9+

    This partially reverts commit 3e57e9494c2279580ad6a83ab8c065d01e7e634e
    which caused a bunch of GPU hangs on several Source titles.  To date, we
    have no clue why these hangs are actually happening.  This undoes the
    final effect of 3e57e9494c227 and gets us back to not hanging.  Tested
    with Team Fortress 2.

    Reviewed-by: Kenneth Graunke <kenneth at whitecape.org>
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102435
    Fixes: 3e57e9494c2279580ad6a83ab8c065d01e7e634e
    Cc: mesa-stable at lists.freedesktop.org
----------------------------------------------------

Causes following performance drops on GEN9+ (Core) platforms:
* 2-3% in GfxBench Manhattan 3.1
* 1-2% in GfxBench CarChase, SynMark v7 Deferred
* 1% in GfxBench Aztec Ruins

Only other affected test was less relevant / partly CPU bound SynMark DrvState,
it improved by 2% (because it has large number of small viewports which cause
fast-clear handling CPU overhead affect it more than memory bandwidth).

I don't have yet data from BXT, but on GLK the impact was smaller.

-- 
You are receiving this mail because:
You are the assignee for the bug.
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/20171207/4dc740f9/attachment.html>