Bug 11039

Summary: x11perf test run aborting
Product: xorg Reporter: zy <yan.z.zhang>
Component: Server/GeneralAssignee: Keith Packard <keithp>
Status: VERIFIED NOTOURBUG QA Contact: Xorg Project Team <xorg-team>
Severity: major    
Priority: high CC: gordon.jin, nian.wu
Version: git   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
datafile none

Description zy 2007-05-22 18:47:00 UTC 
we run the following command to test Xmark 
$x11perf -rop GXcopy GXxor -all > datafile 
on 915gm and 945gm using both intel i810 and vesa driver
later
$/opt/X11R7.Opensrc/lib/X11/x11perfcomp/Xmark datafile
the result shows that:
WARNING: datafile contains 413, not 441 or 447 'trep' results;
ERROR: sum of weights =2231.0, not equal to 4566.0;
ABORTING!
both on 915gm and 945gm
Comment 1 zy 2007-05-22 18:48:48 UTC 
Created attachment 10074 [details]
datafile
Comment 2 Wang Zhenyu 2007-06-06 19:21:58 UTC 
Why do you use Xmark? you should just use x11perfcomp.
Comment 3 WuNian 2007-06-06 19:35:22 UTC 
x11perfcomp is used to compare several x11perf result. It will skip the failed case.
Xmark can give a performance number against single x11perf result. If x11perf abort, Xmark will report this problem.

Anyway, we will run x11perf again with the latest source.
Comment 4 Wang Zhenyu 2007-06-06 20:33:05 UTC 
x11perfcomp can be used on single data to get speed number, although -r/-ro has no effect then. 

What's your repeat time? your have 413 'trep' result, which Xmark script will refuse as it only agrees on 441 or 447 results. Not sure what's wrong.
Comment 5 WuNian 2007-06-07 20:26:47 UTC 
It's XMark issue. 
Now we are suggested to use x11perfcomp to compare performance data, not 
Xmark. 
So don't care about Xmark.

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.