Re: [ata] 0568e61225: stress-ng.copy-file.ops_per_sec -15.0% regression

[Damien Le Moal <damien.lemoal@opensource.wdc.com>]

On 2022/08/18 23:24, Oliver Sang wrote:
> On Thu, Aug 18, 2022 at 10:28:30AM +0100, John Garry wrote:
>> On 18/08/2022 03:06, Oliver Sang wrote:
>>
>> Hi Oliver,
>>
>>>>> did I do the right thing?
>>>> Sorry but I was not really interested in 4cbfca5f77 and I see where that
>>>> build error is coming, but don't be concerned with it. However, for
>>>> avoidance of doubt, if you have results for vanilla v6.0-rc1 then that would
>>>> be appreciated.
>>> for v6.0-rc1, it's still 512/512
>>>
>>>> I will also send a separate patch for testing if you don't mind.
>>> sure! we are very glad that we could help.
>>
>> As you probably saw, I sent "[RFT PATCH] ata: libata: Set __ATA_BASE_SHT
>> max_sectors" for testing on top of v6.0-rc1, and I hope that then we can get
>> same performance as v5.19
> 
> yeah, our test confirmed your expectation:
> 
> stress-ng.copy-file.ops_per_sec
> v5.19                 - 26.85
> v6.0-rc1              - 23.03
> v6.0-rc1 + your patch - 26.94

Thanks for testing Oliver. I pushed the fix and sent it out as an rc2 fix.

> 
>>
>> Thanks,
>> John


-- 
Damien Le Moal
Western Digital Research