Areca 1880

[Anvil]

Steve,

I'm a bit excited , I expect it could take a while to get to know this thing.
The test was using 16KB stripe size, 1GB cache, WBC
Read ahead -> Disabled

@Nizzen
I've explored the 4K and it seems to max out at 93.000 iops using 4 workers.

Write Through or Write Back doesn't seem to matter for random read, unlike the 9260 where it does matter.

edit:

I'll run a few standard tests just to have something to compare to

ATTO, CDM, AS SSD coming up shortly

Still running 4R0 Intel G1 80GB, SS 4KB

ATTO 256MB (cache testing, right 1GB)

atto_256_4r0_Intel_G1.PNG atto_1GB_4r0_Intel_G1.PNG

AS SSD

(left WT, right WB)
as-ssd-bench Areca ARC-1880-V 31.08.2010 20-35-57_4R0_Intel_G1_WT.png as-ssd-bench Areca ARC-1880-V 31.08.2010 20-35-57_4R0_Intel_G1_WB.png

(left WB, Volume read ahead, - right + HDD read ahead)
as-ssd-bench Areca ARC-1880-V 31.08.2010 20-35-57_4R0_Intel_G1_WB_VOL_RA.png as-ssd-bench Areca ARC-1880-V 31.08.2010 20-35-57_4R0_Intel_G1_WB_VOL_RA_HDD_RA.png

CDM
(WB, Volume read ahead + HDD read ahead)
cdm_4R0_IntelG1_100MB.PNG cdm_4R0_IntelG1_1000MB.PNG cdm_4R0_IntelG1_2000MB.PNG

iometer 1GB testfile, read ahead disabled
4K RR 1W QD1-32 + 4K RR 4W QD4-32
+
same runs using Volume read ahead
max ~100.000 iops at low QD on both
+
last file is 32KB stripe size, 1 and 3GB testfile
comment on the 3GB test, ~53.000 iops is max, a bit disappointing.
Looks like as one steps out of the cache, things slow down.