* Performance issue with SATA port multiplier @ 2011-02-24 14:32 Sylvain LEVEQUE 2011-02-24 15:21 ` Tejun Heo 0 siblings, 1 reply; 4+ messages in thread From: Sylvain LEVEQUE @ 2011-02-24 14:32 UTC (permalink / raw) To: linux-ide Hello I own a Venus T5C mini box based on a JMicron JM393 controller, used in pure port multiplier mode (no JBOD, no RAID). It is plugged in eSATA to a QNAP TS-119 NAS which runs a 2.6.37 kernel on a Marvell Kirkwood 88F6281. When I first started copying data on any disk of the external box, I realized the write speed was low (read speed is OK). Testing with dd and various block sizes shows that the larger the block size, the higher the throughput is. On my NAS, I determined the work buffer size of the cp command to be approximately 8KB. With a 8KB blocksize, when writing a 16MB file with dd on one of the disks, I average at 6,5 MB/s write speed. It is already a great improvement compared to the 393 KB/s write speed I obtained on a 2.6.36 kernel, but it is low compared to results I obtain with higher blocksizes (52,4 MB/s with 4MB blocksize) or using other hardware (laptop with Intel ICH9M-E/M controller reaches 286 MB/s with 8KB blocksize using a 2.6.31 kernel). When I take one of the disks out and connect it in eSATA with a basic external case, I reach 37,6 MB/s with 8KB blocksize. This result is pretty constant whatever the blocksize value I tested (from 1KB to 16MB, doubling each time). Please let me know what information is needed to have a better understanding of this issue. Thanks in advance for your help. -- Sylvain ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Performance issue with SATA port multiplier 2011-02-24 14:32 Performance issue with SATA port multiplier Sylvain LEVEQUE @ 2011-02-24 15:21 ` Tejun Heo 2011-02-24 16:30 ` Sylvain LEVEQUE 2011-03-10 12:30 ` Sylvain LEVEQUE 0 siblings, 2 replies; 4+ messages in thread From: Tejun Heo @ 2011-02-24 15:21 UTC (permalink / raw) To: Sylvain LEVEQUE; +Cc: linux-ide, justin (cc'ing Justin) Hello, On Thu, Feb 24, 2011 at 03:32:50PM +0100, Sylvain LEVEQUE wrote: > I own a Venus T5C mini box based on a JMicron JM393 controller, used > in pure port multiplier mode (no JBOD, no RAID). It is plugged in > eSATA to a QNAP TS-119 NAS which runs a 2.6.37 kernel on a Marvell > Kirkwood 88F6281. > > When I first started copying data on any disk of the external box, I > realized the write speed was low (read speed is OK). Testing with dd > and various block sizes shows that the larger the block size, the > higher the throughput is. On my NAS, I determined the work buffer size > of the cp command to be approximately 8KB. > > With a 8KB blocksize, when writing a 16MB file with dd on one of the > disks, I average at 6,5 MB/s write speed. It is already a great > improvement compared to the 393 KB/s write speed I obtained on a > 2.6.36 kernel, but it is low compared to results I obtain with higher > blocksizes (52,4 MB/s with 4MB blocksize) or using other hardware > (laptop with Intel ICH9M-E/M controller reaches 286 MB/s with 8KB > blocksize using a 2.6.31 kernel). > > When I take one of the disks out and connect it in eSATA with a basic > external case, I reach 37,6 MB/s with 8KB blocksize. This result is > pretty constant whatever the blocksize value I tested (from 1KB to > 16MB, doubling each time). Sounds like a latency related issue. Is NCQ being used? -- tejun ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Performance issue with SATA port multiplier 2011-02-24 15:21 ` Tejun Heo @ 2011-02-24 16:30 ` Sylvain LEVEQUE 2011-03-10 12:30 ` Sylvain LEVEQUE 1 sibling, 0 replies; 4+ messages in thread From: Sylvain LEVEQUE @ 2011-02-24 16:30 UTC (permalink / raw) To: Tejun Heo; +Cc: linux-ide, justin Hello 2011/2/24 Tejun Heo <tj@kernel.org>: > > (cc'ing Justin) > > Hello, > > On Thu, Feb 24, 2011 at 03:32:50PM +0100, Sylvain LEVEQUE wrote: >> I own a Venus T5C mini box based on a JMicron JM393 controller, used >> in pure port multiplier mode (no JBOD, no RAID). It is plugged in >> eSATA to a QNAP TS-119 NAS which runs a 2.6.37 kernel on a Marvell >> Kirkwood 88F6281. >> >> When I first started copying data on any disk of the external box, I >> realized the write speed was low (read speed is OK). Testing with dd >> and various block sizes shows that the larger the block size, the >> higher the throughput is. On my NAS, I determined the work buffer size >> of the cp command to be approximately 8KB. >> >> With a 8KB blocksize, when writing a 16MB file with dd on one of the >> disks, I average at 6,5 MB/s write speed. It is already a great >> improvement compared to the 393 KB/s write speed I obtained on a >> 2.6.36 kernel, but it is low compared to results I obtain with higher >> blocksizes (52,4 MB/s with 4MB blocksize) or using other hardware >> (laptop with Intel ICH9M-E/M controller reaches 286 MB/s with 8KB >> blocksize using a 2.6.31 kernel). >> >> When I take one of the disks out and connect it in eSATA with a basic >> external case, I reach 37,6 MB/s with 8KB blocksize. This result is >> pretty constant whatever the blocksize value I tested (from 1KB to >> 16MB, doubling each time). > > Sounds like a latency related issue. Is NCQ being used? I actually already tried disabling it but I didn't see much change so moved on to other tests. This is what I get for the three drives: $ cat /sys/block/sd[c-e]/device/queue_depth 1 31 31 $ ls -l /dev/disk/by-label/ lrwxrwxrwx 1 root root 10 2011-02-21 20:36 NASBACKUP -> ../../sdc1 lrwxrwxrwx 1 root root 10 2011-02-21 20:36 HDD2_SAMSUNG640 -> ../../sdd1 lrwxrwxrwx 1 root root 10 2011-02-21 20:36 HDD4_SAMSUNG640 -> ../../sde1 $ dd if=testfile of=/media/NASBACKUP/testfile bs=8K 2048+0 records in 2048+0 records out 16777216 bytes (17 MB) copied, 7,34586 s, 2,3 MB/s $ dd if=testfile of=/media/HDD2_SAMSUNG640/testfile bs=8K 2048+0 records in 2048+0 records out 16777216 bytes (17 MB) copied, 2,77398 s, 6,0 MB/s $ dd if=testfile of=/media/HDD4_SAMSUNG640/testfile bs=8K 2048+0 records in 2048+0 records out 16777216 bytes (17 MB) copied, 2,62296 s, 6,4 MB/s Read throughput is ca 30MB/s for all disks. Now, without NCQ: # echo 1 > /sys/block/sdd/device/queue_depth # echo 1 > /sys/block/sde/device/queue_depth $ cat /sys/block/sd[c-e]/device/queue_depth 1 1 1 $ dd if=testfile of=/media/HDD2_SAMSUNG640/testfile bs=8K 2048+0 records in 2048+0 records out 16777216 bytes (17 MB) copied, 2,31012 s, 7,3 MB/s I confirm the observed improvement on the second disk as well. Thanks -- Sylvain ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Performance issue with SATA port multiplier 2011-02-24 15:21 ` Tejun Heo 2011-02-24 16:30 ` Sylvain LEVEQUE @ 2011-03-10 12:30 ` Sylvain LEVEQUE 1 sibling, 0 replies; 4+ messages in thread From: Sylvain LEVEQUE @ 2011-03-10 12:30 UTC (permalink / raw) To: Tejun Heo; +Cc: linux-ide, justin Hello > Sounds like a latency related issue. Is NCQ being used? Is there any other way I could try to work around the suspected latency issue? Thanks -- Sylvain ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2011-03-10 12:31 UTC | newest] Thread overview: 4+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2011-02-24 14:32 Performance issue with SATA port multiplier Sylvain LEVEQUE 2011-02-24 15:21 ` Tejun Heo 2011-02-24 16:30 ` Sylvain LEVEQUE 2011-03-10 12:30 ` Sylvain LEVEQUE
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox