Nothing unique about the -8E; any lsi2008-based HBA (with real/up-to-date LSI fw) is a valid testbed.
If your host is PCIe Gen3, just get a 2308-based HBA; that'll raise your ceiling to ~4.5GB/s.
After all, your own test, while not ideal, still reached 2.8
Best to eliminate extraneous cruft.
Just do:
and add up the speeds. Then multiply by 1.048 (hdparm does MiB).
If your host is PCIe Gen3, just get a 2308-based HBA; that'll raise your ceiling to ~4.5GB/s.
Either you misinterpreted, or their methodologies are flawed.edit: found some youtube benchmarks from artofserver, and a post on STH from Patrick himself...they seemed to max out around 2.2 GB/s
After all, your own test, while not ideal, still reached 2.8
Best to eliminate extraneous cruft.
Just do:
Code:
for i in DrvLtr1 DrvLtr2 ...
do
hdparm -t /dev/sd$i &
done