cvs commit: src/sys/dev/amr amr.c amr_cam.c amr_disk.c amr_pci.c amrio.h amrreg.h amrvar.h

[Scott Long]

Brian Fundakowski Feldman wrote:
> On Fri, Dec 16, 2005 at 12:31:22AM -0700, Scott Long wrote:
> 
>>Joao Barros wrote:
>>
>>>On 12/14/05, Scott Long <scottl at freebsd.org> wrote:
>>>
>>>
>>>>scottl      2005-12-14 03:26:49 UTC
>>>>
>>>>FreeBSD src repository
>>>>
>>>>Modified files:
>>>>  sys/dev/amr          amr.c amr_cam.c amr_disk.c amr_pci.c
>>>>                       amrio.h amrreg.h amrvar.h
>>>>Log:
>>>>Mega update to the LSI MegaRAID driver:
>>>>
>>>>Items 3 and 4 significantly increase the performance of the driver.  On 
>>>>an
>>>>LSI 320-2X card, transactions per second went from 13,000 to 31,000 in my
>>>>testing with these changes.  However, these changes are still fairly
>>>>experimental and shouldn't be merged to 6.x until there is more testing.
>>>
>>>
>>>A 238% improvement is at the very least impressive!
>>
>>Yeah, I was shocked at first too.  The PCIe controllers can likely do
>>even better than that, but my only PCIe test system had other hardware
>>related instabilities, so I didn't do much testing with it.
>>
>>
>>>Any specific testing in mind or just stability?
>>>
>>
>>I'm a bit worried that old controllers (especially ones that predate the
>>move to i960 CPUs) might be broken with this.  But yeah, stability
>>testing is always welcome.  It got beat up for about 2 weeks straight,
>>so I'm pretty confident that it's solid.
> 
> 
> Can you define "beat up" in any detail?  Since I've been messing with
> the GEOM RAID-3 driver recently, I've been looking for various stress
> tests to gain confidence in changes I've made locally.  Best I've had
> going so far is some churn-and-burn on several PostgreSQL databases.
> 

Kris Kenneway spent a couple of weeks doing parallel tarball extractions
to an AMR array under 6.0+mega_amr.diff in order to characterize SMP
contention and thoroughput.

Scott