A 4 disk raid5 on the mobo will probably be around 150-180% faster than a single disk and with 1 TB Constellation ES disks, that translates to around 225-270 MB/s sustained transfer rate. Mind you this is a guess, since I never tried a mobo parity raid, but is based on what I heard from users who did use on-board raid. On a dedicated controller you would see figures around 360 MB/s for the same setup.
Great thanks hmm I guess that is usable and saves me ditching the ssd when I go 8 disks...thanks
1 person found this helpful
If this is a temp raid array, I would suggest a Raid 10 instead of a Raid 5. Onboard Raid controllers are really a bad choice for parity raids. However they are fine with Raid 0,1, or 10.
I did not know that...thanks. I suppose the write speed is better as well?
1 person found this helpful
A 4 Drive Raid 5 and Raid 10 would be close. I couldn't say which one would be faster because they would be so close. However a raid 10 would have far less chance of having issues so speed would really be a secondary concern on the onboard.
Ok thanks. I'll still use 2 Samsung 840 pros for cache and os disks they look mean! But platter in raid 10 looks best for project disk thanks.
Ok thanks ... different mobo tho alex is it? Il be on the sata 2 connections as well as I will have 2 ssd's on the sata 3's all on the intel controller I guess. Anyway raid 10 it is...not sure about that software AJA video systems (not seen that) - I use crystalmark but not done much of this sort of thing...
different mobo tho alex is it? Il be on the sata 2 connections as well as I will have 2 ssd's on the sata 3's all on the intel controller I guess.
I think yours is a faster one - Z800 uses older generation 5520 chipset mobo with SATA 2 (3Gbs) ports.
not sure about that software AJA video systems (not seen that) - I use crystalmark but not done much of this sort of thing...
It's a very basic "quickie" test - but fairly accurate.
Confirming the consensus, RAID 10 is definitely how I would go if I had to use my Intel on-board ports. And Alex, your referenced quick test which only considers read speed, is pretty misleading IMHO for Adobe Premiere users.
I completed some actual test results of RAID 0 vs. RAID 5 vs. RAID 10 using 4 x 1TB Hitachi 7200rpm drives and find them quite interesting - see below. I use dedicated Areca controllers for all my RAIDs now, but have always been curious about how RAID 0, 5, and 10 would stack up using an Intel on-board controller.
Build time: < 1 minute
Avg. read / write: 422 / 423 MB/s
PPBM5 composite score: 171 seconds
Build time: around 14 hours
Avg. read / write: 317 / 76 MB/s [POOR write speed]
PPBM5 composite score: 266 seconds [way slower than other 2 options]
Build time: around 36 hours [clearly a downside for on-board RAID 10]
Avg read / write: 207 / 213 MB/s
PPBM5 composite score: 174 seconds
Gigabyte X58 w/ i7-970 6-core at stock speed
Hitachi 7K1000.C 1TB 7200rpm drives
Avg. read/write testing performed with HD Tune Pro 5.0 and using 2MB block size setting
Ah yes but his test does say write no? - I did wonder about the write speeds - read stuff about raid 5 being low writing but didn't think it would go that low.
Whoa 36 hour build :O - sorry I am new to raid : so if one drive goes down its 36 hours back just for a 4 disk 1 tb build :O ! Even if its blank disks? Eeeek! I assume you can work on with 3 till end of project then rebuild? I also assume the Areca builds the raid faster? Would be nice to see the same set up with the areca as well if its out there?
Yes the rebuild time is extremely long on the onboard and that does mean rebuilding after replacing a failed drive in a redundant raid. You can still use the array while it's rebuilding but it runs much slower and is not recommended with a parity raid on the onboard controller. An actual raid card will rebuild the arrays much faster. That is one of the reasons that Parity raids are far better on a raid card versus onboard raid. Also you have less chance of corrupting the array while it's rebuild if your using it during rebuild with a raid card.
And Alex, your referenced quick test which only considers read speed, is pretty misleading IMHO for Adobe Premiere users.
1. Every time I look at the screenshot, there's the write speed. It's in the same place every time, the same number: 238.5MB/s. Is it me or...?
2. Misleading to who? Did you follow the thread?
Speaking of read speeds, those are way more important for most workflows, vs. write speeds.