Bug 44523 - nexuiz perf regression since u_vbuf: implement another upload codepath which unrolls indices
Summary: nexuiz perf regression since u_vbuf: implement another upload codepath which ...
Status: RESOLVED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/Gallium/r600 (show other bugs)
Version: git
Hardware: x86 (IA32) Linux (All)
: medium normal
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-01-06 03:12 UTC by Andy Furniss
Modified: 2012-01-07 12:31 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments

Description Andy Furniss 2012-01-06 03:12:45 UTC
d-r-t kernel, HD4890.

Since -

commit ce44bae366ade59fb2dbdfbfe5a1ab8d24518a57
Author: Marek Olšák <maraeo@gmail.com>
Date:   Tue Jan 3 22:01:03 2012 +0100

    u_vbuf: implement another upload codepath which unrolls indices
    
    Improves performance from cca 1 fps to 23 fps in Cogs.
    This new codepath is not always used, instead, there is a heuristic which
    determines whether to use it. Using translate for uploads is generally
    slower than what we have had already, it's a win only in a few cases.

I get quite a noticeable perf regression running demo1 in nexuiz.

Other games (openarena,ut2004 demo, etqw) seem unaffected 


91.2740132 fps, one-second fps min/avg/max: 50 99 231 (90 seconds)

to

55.6802612 fps, one-second fps min/avg/max: 19 69 231 (90 seconds)

Sometimes I saw a couple of short (1/4 sec) stalls as well, which gave worse results, above was without stalls.
Comment 1 Marek Olšák 2012-01-07 12:31:30 UTC
Fixed by 2bdf93449a0b00145fabdd3f7866acbafa18e734. Closing.


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.