Hi,
These patches were enabled eMMC4.41 standard new feature: background
operations and HPI.
Since these two features are opertional, driver only enable for those
eMMC card which supports them.
HPI can interrupt both background operations and low priority foreground
operations. In these patches, only implemented using HPI to interrupt
background operations.
patch1: enable background operations feature for supported eMMC card.
patch2: let eMMC card do background operations when user reqeust queue
is idle. In this patch, also considered when driver added runtime power
management.
patch3: enable HPI feature for supported eMMC card.
patch4: use HPI to interrupt background operations if there is user
request(foreground operations).
Thanks
Chuanxiao
Hi,
Basic question. Do you have any performance gain or improvement?
In my test environment. no performance gain. Yes I know it's not for
performance improvement.
How to measure or check it has some improvement?
Any ideas?
Thank you,
Kyungmin Park
On Tue, Nov 23, 2010 at 5:43 PM, Chuanxiao Dong
<[email protected]> wrote:
> Hi,
> ? ? ? ?These patches were enabled eMMC4.41 standard new feature: background
> ? ? ? ?operations and HPI.
>
> ? ? ? ?Since these two features are opertional, driver only enable for those
> ? ? ? ?eMMC card which supports them.
>
> ? ? ? ?HPI can interrupt both background operations and low priority foreground
> ? ? ? ?operations. In these patches, only implemented using HPI to interrupt
> ? ? ? ?background operations.
>
> ? ? ? ?patch1: enable background operations feature for supported eMMC card.
> ? ? ? ?patch2: let eMMC card do background operations when user reqeust queue
> ? ? ? ?is idle. In this patch, also considered when driver added runtime power
> ? ? ? ?management.
> ? ? ? ?patch3: enable HPI feature for supported eMMC card.
> ? ? ? ?patch4: use HPI to interrupt background operations if there is user
> ? ? ? ?request(foreground operations).
>
> Thanks
> Chuanxiao
> --
> To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
> the body of a message to [email protected]
> More majordomo info at ?http://vger.kernel.org/majordomo-info.html
>
>
> Basic question. Do you have any performance gain or improvement?
> In my test environment. no performance gain. Yes I know it's not for
> performance improvement.
My understanding:
If BKOPS feature is not enabled, the background operations will be started when card handles some read or write command. This will impact read/write performance. If BKOPS is enabled, driver can check whether need to do a BKOPS each time after card handled a read/write command. If need, driver can start BKOPS as soon as possible when user request is empty. This will can improve read/write performance.
So I think, anyhow background operations can improve the performance, but only can be seen when your test card need to do a background operations. If your card needn't, there will be no performance gain I think. Driver will works as normal.
BTW, the BKOPS is an operational features for eMMC4.41 card, to ensure whether the card support this feature, check ext_csd[502].
>
> How to measure or check it has some improvement?
> Any ideas?
>
Right now I didn't know how to do a precise measurement. Maybe keep on transferring large data for a long time with a card which enabled/or not the BKOPS feature can help to get some measurement data.
Thanks
Chuanxiao
????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay? ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m???? ????????I?