Skip to contentSkip to conversation

i/o benchmarks on 2.5.70* kernels

DISCUSSION [email protected] 1 participant 1 message Jun 17, 2003

lore
Share & tools
lore
#0 lore raw
Quoted 4 lines

tiobench on SMP results are not very good, lots of fragmentation, the random IO throughput drops is probably due to AS strangling TCQ though. You are using SMP and TCQ, right?

SMP yes. The non-benchmarked disks (which don't matter say this at boot time:

scsi0:A:1:0: Tagged Queuing enabled. Depth 4
(aic7xxx)

These config options on the non-benchmarked disks:
CONFIG_AIC7XXX_CMDS_PER_DEVICE=4 CONFIG_AIC7XXX_RESET_DELAY_MS=15000 CONFIG_AIC7XXX_DEBUG_ENABLE=y CONFIG_AIC7XXX_DEBUG_MASK=0 CONFIG_AIC7XXX_REG_PRETTY_PRINT=y

----

The benchmarked disks are on scsi3 and scsi4. The benchmarked disks do not tell what TCQ they have.

CONFIG_SCSI_QLOGIC_ISP=y
CONFIG_SCSI_QLOGIC_FC=y
CONFIG_SCSI_QLOGIC_FC_FIRMWARE=y

TCQ on the benchmarked disks may be 1024...
+#define MAXISPREQUEST(isp) ((IS_FC(isp) || IS_ULTRA2(isp))? 1024 : 256)

but that's just a guess.

Signature · 2 lines
Randy Hron
http://home.earthlink.net/~rwhron/kernel/bigbox.html