To me data safety is more important than outright speed, and as I'm looking to build a new system I'm considering using onboard fake raid 5 (Raid 5 on bios with 3 hd's). Various articles I've read says about the cpu doing all the work etc. but with Quad cores and 4GB of ram, how much of an impact will this actually be? And if these raid 5 bios stuff is SOOOO bad (Which seems to be the normal comment) then why do they put it on?
Raid 5 fake setup
- Thread starter Mat-d-Rat
- Start date
My machine runs what you would call "fake" RAID5, 6 disks connected to the intel ICH9R chipset. Even under strained continuous disk access, my Q6600 only takes 4-5% load.
It's nothing.
I don't believe all the hype that onboard RAID5 is as bad as people make out, I've been running my array for 3 years now and it's performance still kicks the ass out of a more costly Intel X25 RAID0 setup.
It's nothing.
I don't believe all the hype that onboard RAID5 is as bad as people make out, I've been running my array for 3 years now and it's performance still kicks the ass out of a more costly Intel X25 RAID0 setup.
Excuse the random peaks and troughs, my machine is in use doing a whole host of things whilst I ran this test, but it gives you an idea (also why my CPU load is higher, but still, not much):
Take the bar chart with a pinch of salt, as that will have all been off the cache on the drives, not the spindles themselves. It can usually sustain over 300MB/s read without problem.
Take the bar chart with a pinch of salt, as that will have all been off the cache on the drives, not the spindles themselves. It can usually sustain over 300MB/s read without problem.
Not bad at all 
What about write though, and I did read that teh Intel raid system was good. I'm looking at an AM3 system so no intel raid. Thanks for taking the time to show that, may I enquire as to what disks? I was looking at samsung f1's
What about write though, and I did read that teh Intel raid system was good. I'm looking at an AM3 system so no intel raid. Thanks for taking the time to show that, may I enquire as to what disks? I was looking at samsung f1'sThese are all Hitachi Deskstar 320GB's (best £/performance I could get at the time, again, 3-year old array!).
I've never benched the write, never found a benchmark that will deal with it, but I've seen well over 70MB/s write speeds between my server (also on pseudo RAID5, on an AM3 board, 3 drives) and this machine.
I've never benched the write, never found a benchmark that will deal with it, but I've seen well over 70MB/s write speeds between my server (also on pseudo RAID5, on an AM3 board, 3 drives) and this machine.
For comparison, here is my 3-disk array on the server. Again, take the figures with a pinch of salt due to the fact that the machine itself is host to 5 other Hyper-V servers all of which will have activity at present.
Soldato
- Joined
- 2 Nov 2007
- Posts
- 4,185
ATTO does write speeds.
Well after building the system with raid-5 the performance was abysmal, peeked at around 60mb/s. As I had 4 drives I have switched to raid 10 and now get around 210mb/s average (Tad better than raid 5!). Again thanks for your input though, but this Gigabyte 7980XTA-UD4 is rubbish for raid-5.