Time to debate RAID again.
Here’s the results of an earlier test using a 2000+ on an force2 mb, 2x40gig segate hard drives on 768meg ram:
<!–QuoteBegin>
Quote |
DskBench 2.12 © 1998, SESA, J.M.Catena (cat@sesa.es, www.sesa.es) Timer Check = 1040 (should be near 1000) CPU Check = 50.06 % (should be near 50.00 %) CPU index (relative to Pro 200 MHz) = 6.126362 Open = 0 ms Write = 4397 ms, 58.22 MB/s, CPU = 4.85 % Flush = 6 ms Rewin = 0 ms Read = 4109 ms, 62.30 MB/s, CPU = 4.64 % Close = 7 ms BlockSize = 131072, MB/s = 9.09, Tracks = 108.06, CPU = -2.97 % BlockSize = 65536, MB/s = 4.95, Tracks = 58.84, CPU = -4.51 % BlockSize = 32768, MB/s = 3.19, Tracks = 37.89, CPU = -4.45 % BlockSize = 16384, MB/s = 1.84, Tracks = 21.88, CPU = -4.39 % BlockSize = 8192, MB/s = 1.01, Tracks = 11.98, CPU = -4.34 % BlockSize = 4096, MB/s = 0.54, Tracks = 6.46, CPU = -4.29 % |
And here’s one, using the same cpu, overclocked to 2400+ speeds - but using a new segate 160gig (PATA, dunno about cache), on the same motherboard without raid, and 1gig ram.
<!–QuoteBegin>
Quote |
DskBench 2.12 © 1998, SESA, J.M.Catena (cat@sesa.es, www.sesa.es) Timer Check = 1000 (should be near 1000) CPU Check = 49.98 % (should be near 50.00 %) CPU index (relative to Pro 200 MHz) = 11.268870 Open = 0 ms Write = 6390 ms, 40.06 MB/s, CPU = 1.03 % Flush = 0 ms Rewin = 0 ms Read = 5531 ms, 46.28 MB/s, CPU = 1.50 % Close = 0 ms BlockSize = 131072, MB/s = 14.42, Tracks = 171.46, CPU = 0.85 % BlockSize = 65536, MB/s = 7.35, Tracks = 87.35, CPU = 0.87 % BlockSize = 32768, MB/s = 3.72, Tracks = 44.25, CPU = 0.46 % BlockSize = 16384, MB/s = 2.04, Tracks = 24.31, CPU = 0.58 % BlockSize = 8192, MB/s = 9.75, Tracks = 115.94, CPU = 4.21 % BlockSize = 4096, MB/s = 13.36, Tracks = 158.88, CPU = 10.72 % |
Interesting, no? The new drive has 3/4 the throughput (measured roughly the same in HD tach) of the raid drive, but has around 50% more tracks, and actually was pretty quick in the lower cluster sizes. I’ll test one of the single 40’s that used to be in the raid array.
Edit: Tested, with more wacky results, but probably more in line with what you’d expect.<!–QuoteBegin>
Quote |
DskBench 2.12 © 1998, SESA, J.M.Catena (cat@sesa.es, www.sesa.es) Timer Check = 1000 (should be near 1000) CPU Check = 50.08 % (should be near 50.00 %) CPU index (relative to Pro 200 MHz) = 11.274411 Open = 0 ms Write = 4563 ms, 56.10 MB/s, CPU = 1.36 % Flush = 0 ms Rewin = 0 ms Read = 4563 ms, 56.10 MB/s, CPU = 1.57 % Close = 0 ms BlockSize = 131072, MB/s = 10.68, Tracks = 126.97, CPU = 0.47 % BlockSize = 65536, MB/s = 8.50, Tracks = 101.03, CPU = 0.73 % BlockSize = 32768, MB/s = 4.66, Tracks = 55.43, CPU = 0.59 % BlockSize = 16384, MB/s = 2.44, Tracks = 28.98, CPU = 0.73 % BlockSize = 8192, MB/s = 1.55, Tracks = 18.47, CPU = 0.81 % BlockSize = 4096, MB/s = 1.19, Tracks = 14.15, CPU = 1.18 % |
Throughput is huge, up there with the raid. Gonna have to retest some things here.
Willy.