I can not get my raid 0 of two 30gig OCZ vertex ssd's to go above a sequential read time of 250?
I have the latest firmware update. I have all the tweaks applied. I have a fresh install with format and Sanitary erase. (multiple times) What ever I do I can not get it into the 400 mb/s little alone a 300 mb/s.
Can someone give me some insight as to what the problem could be?
They are on ports 0 and 1.
Before I put the drives in RAID 0. I tested one of the drives it was getting 200mb/s read time.
All the benchmarks are based off of "CrystalDiskMark" on HD tune bench mark it would only get 120mb/s
I did a fresh install current space being used is 8.02 gig
Expensive hardware RAID for a simple RAID0 array? LOL! Dont waste your money please. Even the most expensive hardware RAID are slower than software RAID when it comes to random IOps benchmarks; Areca ARC-12xx seems limited to 70.000 IOps.
There can be several reasons your RAID0 is not functioning properly:
- you have a 'bad' partition (misaligned) starting at 31.5KiB offset
- you have multiple partitions on the RAID; and the second partition is not properly aligned
- the selected stripesize is too low (< 128KiB)
- you defragmented your SSD at least once during its lifetime (you need a full zero write to regain performance)
- you performed inadequate benchmarking (low queue depth)
If you want some serious advice and really solve this issue, you must do some work yourself:
- explain your setup in more detail (enabled the 'write caching' option in Intel drivers? How did you create the partitions? Can you check for the partition offset, etc)
- post proper benchmarks performed at the filesystem level, for example HDTune Pro's "File" benchmark. The normal HDTune benchmark only tests raw disk access; without using read-ahead which your filesystem does use to allow parallel I/O processing.
If you can do that, i might be able to help you out with some specific advice.
Please download HDTune Pro evaluation and run the File benchmark (its on one of the tabs). This will test on the filesystem level and will give you 'real' scores other programs would get also. The raw read results HDTune produces are only usable for plain HDDs; not for RAID-arrays and such.
So if you test with HDTune Pro and run the Filesystem benchmark, you should get proper scores. If you enable write caching, then also the normal raw read benchmark from HDTune will get a boost. Some of this boost is bogus and does not translate to actual higher performance. So be careful when judging!
HDTune is a very basic benchmark; not really suited for RAID-arrays for numerous reasons. So when enabling 'write caching' (try device manager properties of the Intel RAID controller) - it only appears to increase sequential performance significantly, while right now you're seeing incorrect/meaningless low scores because the benchmark is insufficient.
Hello! It looks like you got some higher results with the File benchmark, which is real performance as it uses the same API as other applications will use.
Still, results as somewhat disappointing. With RAID0, there is no reason a good setup would not get twice the sequential transfer speeds, and at least a decent boost to (multiqueue) random IOps.
Could you run the File benchmark with some different settings? Increase the file size to 2048MB and set the delay to a high value. I'm assuming its in seconds, then 10 second delay between tests would be appropriate. If its in msec, it should be 10000.
Could you also explain what 'tweaks' you applied? It could be one of these will actually hamper performance.
If you're running FW 1.5 it'll keep itself in decent shape. Otherwise you'll want to look into Tony Trim.
Yeah I have the 1.5 firmware on both drives. Do the defrag programs that are suppose to be for SSDs work with out messing up your ssd; because I have come across a couple of defrag programs where they have an ssd mode. Should I stay away from these?