Skip to contentSkip to conversation

3w-9xxx status in kernel

DISCUSSION Stéphane Jourdois 3 participants 4 messages Apr 18, 2006 – Apr 19, 2006

lore
Share & tools
lore
#0 lore raw

Hi,

I have a question about 3w-9xxx driver versions :

3w-9xxx version in vanilla 2.6.16 is 2.26.02.005
3w-9xxx version in 2.6.17-rc1-mm2 is 2.26.02.006

but :
3w-9xxx version in 3ware.com 9.3.0.3 codebase is 2.26.04.007

The documentation with 9.3.0.3 codebase says it will not works with kernel driver <2.26.04.004. But I observe it works fine with codebase 9.3.0.2 (the documentation says it should not).

What is current status of 3w-9xxx driver in 2.6 ?
Will it works on a 9550SX using 9.3.0.3 firmware ?
Could you update documentation about that somewhere, for exemple in 3w-9xxx.c header ?

Thanks very much.

Signature · 4 lines
 ///  Stephane Jourdois     /"\  ASCII RIBBON CAMPAIGN \\\
(((    Consultant securite  \ /    AGAINST HTML MAIL    )))
 \\\   24 rue Cauchy         X                         ///
  \\\  75015  Paris         / \    +33 6 8643 3085    ///
#1 lore raw

Hi,

same problem over here. Why does the newest kernel contain an old version of the 3w-9xxx driver?

We are having performance problems using a 9550SX controller. Read throughput (measured with hdparm) is worse than on a Desktop system. We are considering trying to replace it with the newest driver from 3ware.com.

Greetz,
Martin

Quoted 22 lines · Stéphane Jourdois

Hi,

I have a question about 3w-9xxx driver versions :

3w-9xxx version in vanilla 2.6.16 is 2.26.02.005
3w-9xxx version in 2.6.17-rc1-mm2 is 2.26.02.006

but :
3w-9xxx version in 3ware.com 9.3.0.3 codebase is 2.26.04.007

The documentation with 9.3.0.3 codebase says it will not works with kernel driver <2.26.04.004. But I observe it works fine with codebase 9.3.0.2 (the documentation says it should not).

What is current status of 3w-9xxx driver in 2.6 ?
Will it works on a 9550SX using 9.3.0.3 firmware ?
Could you update documentation about that somewhere, for exemple in 3w-9xxx.c header ?

Thanks very much.

#2 lore raw
Quoted 3 lines · Martin Honermeyer

We are having performance problems using a 9550SX controller. Read throughput (measured with hdparm) is worse than on a Desktop system. We are considering trying to replace it with the newest driver from 3ware.com.

I saw somewhere in the knowledgebase on there website that 2.26.02 is newer than 2.26.04.

anyway :

Don't test using hdparm, it doesn't works (I don't know why). Test using bonnie++. I have 100MB/s throughput reading and writing on a 9550SX-12. hdparm says 7MB/s.

Hope it helps.

Stéphane Jourdois.

Signature · 4 lines
 ///  Stephane Jourdois     /"\  ASCII RIBBON CAMPAIGN \\\
(((    Consultant securite  \ /    AGAINST HTML MAIL    )))
 \\\   24 rue Cauchy         X                         ///
  \\\  75015  Paris         / \    +33 6 8643 3085    ///
#3 lore raw
In reply to Martin Honermeyer

In article <[email protected]>,

Quoted 8 lines · Martin Honermeyer

Hi,

same problem over here. Why does the newest kernel contain an old version of the 3w-9xxx driver?

We are having performance problems using a 9550SX controller. Read throughput (measured with hdparm) is worse than on a Desktop system. We are considering trying to replace it with the newest driver from 3ware.com.

The default settings for the 3w9xxx cards suck.

You need to make sure that the nr_requests (kernel request queue) is at least twice the size of queue_depth (hardware requests queue). Also the deadline or cfq i/o schedulers work a bit better for database-like workloads.

Try something like this, replacing sda with the device name of your 3ware controller.

# Limit queue depth somewhat
echo 128 > /sys/block/sda/device/queue_depth

# Increase nr_requests
echo 256 > /sys/block/sda/queue/nr_requests

# Don't use as for database-like loads echo deadline > /sys/block/sda/queue/scheduler

CFQ seems to like larger nr_requests, so if you use CFQ, try 254 (maximum hardware size) for queue_depth and 512 or 1024 for nr_requests.

Oh, remember, if you have just created a RAID array on the disks, wait with testing until the whole array has been rebuild..

Mike.