Adaptec 29160 + IBM DMSV18N write performance problem

[Mika Kuoppala]

Hello to all of you scsi experts!

I seem to have somekind of problem with Adaptec 29160 + IBM DMSV18N 
combination. First some benchmark results:

Adaptec 29160 + IBM DMSV18N   (160m LVD):

Size is MB, BlkSz is Bytes, Read, Write, and Seeks are MB/sec

         File   Block  Num  Seq Read    Rand Read   Seq Write  Rand Write
  Dir    Size   Size   Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
/mnt/sd  500    4096    1  26.38 13.8% 0.884 1.18% 6.490 5.53% 1.065 1.90%

						   *****

Adaptec 29160 + IBM DNES31850  (80m LVD):

         File   Block  Num  Seq Read    Rand Read   Seq Write  Rand Write
  Dir    Size   Size   Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
/mnt/sd  500    4096    1  18.23 9.88% 0.669 1.07% 18.00 16.8% 0.867 1.49%


Tekram DC-390U2W + IBM DMSV18N (80m LVD):

         File   Block  Num  Seq Read    Rand Read   Seq Write  Rand Write
  Dir    Size   Size   Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
/mnt/sd  500    4096    1  25.61 14.4% 0.907 1.62% 24.67 22.3% 1.661 3.19%

Tekram DC-390U2W + IBM DNES3185 (80m LVD):

         File   Block  Num  Seq Read    Rand Read   Seq Write  Rand Write
  Dir    Size   Size   Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
/mnt/sd  500    4096    1  18.39 10.4% 0.647 0.74% 17.87 15.4% 1.024 2.03%

When you look at the benchmarks you see that with Adaptec 29160 with
IBM DMSV18N drive attached, sequential write performance is lost.
No data corruption has yet occured so the problem is just poor
write performance.

This behaviour has been tested on 2 separate machines, and it also
was shown on another adaptec controller in my home machine. I have
to check the exact type of it when i get home.

The Linux kernel was 2.2.15 with aic7xxx.c version 5.1.29.
The benchmark program was http://www.iki.fi/miku/tiobench

In my point of view this seems to be problem in interaction with this 
particular IBM drive model and adaptec controller(s).

Are there others who have this similar drive and controller and
could verify that there is a problem ? I will iterate this further
as time permits and as long i have the equipment, cause they are not mine :)

Cheers,
-- Mika Kuoppala <miku at iki.fi>




To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message