EVGA 680i -- Raid Problems

[IntelOwnz]

A buddy of mine tried running a Raid-0 on his motherboard but is getting horrid results.......4x 74GB 16mb Raptors, but only getting like 230MB/s Throughput, I've seen two drive R-0 get better results.

Do not Raid Flame, just curious what some of you 680i owners thought of this.

 

[unhappy_mage]

How often does he play HDTach? In other words, who cares about synthetic benchmarks, how does it do with his workload?

 

[drizzt81]

I've seen two drive R-0 get better results.

What Drives were these arrays using? I would love to know which drive can sustain more than 115 MiB/s.

How often does he play HDTach? In other words, who cares about synthetic benchmarks, how does it do with his workload?

Good point. If he is really curious, he can try and see what happens if he uses only three drives. Maybe there are diminishing returns to scale in RAID-0?

 

[Madwand]

230 MB/s / 4 drives = 57.5 MB/s. This is hardly what I'd call "horrible". Sure, Raptors can do better at peaks, but this is in the right ball park for some results with these drives and this configuration. Perhaps you'd like to explain better why this is "horrible". Is it just a mistake confusing burst speeds with sustained? Because no 2 conventional drives can sustain 230 MB/s, and the burst speed results are generally better ignored than worried about.

A smaller stripe size, e.g. 16k, derived from the formula 64k / 4 data drives, because HD Tach requests 64k at a time (as do some others), might get better sustained HD Tach numbers. But this will not necessarily be better for real applications.

There can also be some southbridge link bandwidth tuning options that can make a difference in some cases, but this is deeper and implementation-specific stuff; not necessarily a problem here, and besides, 230 MB/s is pretty decent.