[Mesa-dev] [PATCH] i965 : optimized bucket index calculation

[Matt Turner]

On Tue, Sep 12, 2017 at 10:19 AM, Ian Romanick <idr at freedesktop.org> wrote:
> On 09/12/2017 02:40 AM, Marathe, Yogesh wrote:
>> Hi Jason,
>>
>>
>>
>> On the asserts you’ve mentioned below, I assume we need to add them
>> after ‘bufmgr->num_buckets++’ in add_bucket() as num_buckets could be 0
>> initially. Another clarification on ~1%, we meant approx. 1% there,
>> that’s an improvement we saw in 3Dmark total not a degradation, we’ll
>> correct it in commit msg.
>
> I think the problem is that there is insufficient information about your
> data.  What we want to see in a commit message is something like:
>
> commit 5ae2de81c8350272c122ea38e6bb4c0a41d58921
> Author: Kenneth Graunke <kenneth at whitecape.org>
> Date:   Mon Aug 28 16:08:32 2017 -0700
>
>     i965: Use BLORP for buffer object stall avoidance blits instead of BLT.
>
>     Improves performance of GFXBench4 tests at 1024x768 on a Kabylake GT2:
>     - Manhattan 3.1 by 1.32134% +/- 0.322734% (n=8).
>     - Car Chase by 1.25607% +/- 0.291262% (n=5).
>
>     Reviewed-by: Jason Ekstrand <jason at jlekstrand.net>
>
> The important bits are:
>
> - average improvement
> - statistical deviation
> - number of runs

And for generating such data, we often use http://anholt.net/compare-perf/