Re: [RFC PATCH v4 0/9] slub: Delay freezing of CPU partial slabs

[Chengming Zhou <chengming.zhou@linux.dev>]

On 2023/11/1 21:59, Vlastimil Babka wrote:
>> 3. Testing
>> ==========
>> We just did some simple testing on a server with 128 CPUs (2 nodes) to
>> compare performance for now.
>>
>>  - perf bench sched messaging -g 5 -t -l 100000
>>    baseline	RFC
>>    7.042s	6.966s
>>    7.022s	7.045s
>>    7.054s	6.985s
>>
>>  - stress-ng --rawpkt 128 --rawpkt-ops 100000000
>>    baseline	RFC
>>    2.42s	2.15s
>>    2.45s	2.16s
>>    2.44s	2.17s
> 
> Looks like these numbers are carried over from the first RFC. Could you
> please retest with v4 as there were some bigger changes (i.e. getting
> rid of acquire_slab()).
> 
> Otherwise I think v5 can drop "RFC" and will add it to slab tree after
> the merge window and 6.7-rc1. Thanks!

Ah, yes, I will retest v5 and update the numbers today.

Thanks!

> 
>> It shows above there is about 10% improvement on stress-ng rawpkt
>> testcase, although no much improvement on perf sched bench testcase.
>>
>> Thanks for any comment and code review!
>>
>> Chengming Zhou (9):
>>   slub: Reflow ___slab_alloc()
>>   slub: Change get_partial() interfaces to return slab
>>   slub: Keep track of whether slub is on the per-node partial list
>>   slub: Prepare __slab_free() for unfrozen partial slab out of node
>>     partial list
>>   slub: Introduce freeze_slab()
>>   slub: Delay freezing of partial slabs
>>   slub: Optimize deactivate_slab()
>>   slub: Rename all *unfreeze_partials* functions to *put_partials*
>>   slub: Update frozen slabs documentations in the source
>>
>>  mm/slub.c | 381 ++++++++++++++++++++++++++----------------------------
>>  1 file changed, 180 insertions(+), 201 deletions(-)
>>