Re: Resets on sil3124 & sil3726 PMP

[Petr Vandrovec <vandrove@vc.cvut.cz>]

Petr Vandrovec wrote:
> Tejun Heo wrote:

Hello,

>>
>> Petr Vandrovec wrote:
>>> I have recompiled kernel with all debugging disabled, and it brought me
>>> 1.5MBps, so it is still consistently 1MBps slower than on sil. Disabling
>>> NCQ seems to improve concurrent access a bit (for which I have no
>>> explanation), while slows down single drive scenario:
>>>
>>> With NCQ:
>>>
>>> 1TB alone: 81.22, 79.86
>>> 1TB+1TB:   56.28+56.70, 53.51+56.11
>>>
>>> Without NCQ:
>>>
>>> 1TB alone: 79.78, 80.82
>>> 1TB+1TB:   57.99+58.12, 56.50+56.46
> 
> 3124-2 (norco 4618):
> 
> NCQ:
> 
> 1TB alone:   82.30, 82.43
> 1TB+1TB:     68.36+68.25
> 
> noNCQ:
> 
> 1TB alone:   82.39, 82.29
> 1TB+1TB:     70.33+70.32, 69.47+70.01
> 
> Unfortunately that enclosure has only two slots used.  I'll try to not 
> forget bring two more disks tomorrow to get it fully populated.

In case you are still follwing this thread, 3124-2 with 4 1TB drives 
delivers 58.5MBps for each drive with NCQ enabled, and 60MBps with NCQ 
disabled - total 235-240MBps read speed.  I would like to see 320MBps, 
but apparently today is not my lucky day...

>>> 3512 sil, no NCQ:
>>>
>>> 1TB alone: 82.28, 82.18
>>> 1TB+1TB:   47.20+47.54    # Here apparently command based switching or
>>> 1.5Gbps link between device and PMP becomes bottleneck
>>
>> Hmmmm.... Weird.  Is the different still there if you take PMP out of
>> the picture?
> 
> Will do tomorrow.  I need physical access to the box to do that.

Yes, no difference.  3512 is consistently about 1MBps faster than 3132 
when talking to single Hitachi 1TB drive.
							Petr