CORSAIR 8GB (2 x 4GB) Dominator-GT DHX Pro PC3-16000 DDR3 2000MHz
Single ATI Radeon HD 6850 1GB 256-bit GDDR5 PCI Express 2.0 x 16
(2) WD1002FAEX SATA III 1TB drives connected in RAID 1 through the Intel P-67 connectors
Windows 7 64bit Professional (boots to the RAID 1)
Everything setup quickly and without problems until I started to benchmark the system. The drives show terrible performance on several benches, including the “Windows experience” which shows a 7.8 to 7.9 on all systems but the drives, which show at a 5.9.
Sandra reports very low numbers and even shows the drives spinning at 5400rpm when in raid configuration,(although they show 7200rpm when setup up in AHCI.
I’ve checked for and verified all the latest drivers and Bios versions.
Could we see some read/write benchmark graphs in both RAID and AHCI mode, eg using HD Tune?
You can determine the actual RPM from HD Tune’s access time graph. The drive can report its rated RPM via the ATA Identify Device command, but this command may not be available to utilities such as Sandra when the drive is behind a RAID controller.
note you are using RAID 1 so that is mirroring so not sure what to expect what speeds you would get from it (personally I would expect what you got as its mirroring not RAID 0 so the disk would perform the same if not Worse as you have 50% overhead on Writes + overhead on reads from the driver, Quite sure that chipset/software raid does not support 2 reads at the same time, only hardware raid cards do in RAID 1 I thnk)
From the tests you have done looks about right for RAID 1 as you picked redundancy (RAID 1) over performance (RAID 0) the random reads look about right, your getting about the same as my WD1001FALS drive (click on disk monitor when you do the test you see its a lot smoother then the benchmark)
The burst speed is high due to the driver that’s normal (norm its the Write back cache is enabled, try disabling the Write back cache in the Intel software Right click on the raid set norm in there as an tick box)
When using RAID you do not have direct access to the HDDs (things like SMART and other disk monitoring tools norm will not work or correctly report info as it gets the info from the RAID driver not the disk)
And HDDs can never get higher then 5.9 score due to random access, still good that you still got 5.9 in RAID 1 (RAID 0 mite be able to get into 6-7 with Write back cache enabled but I never used RAID 0 under windows 7, sure the random access speed of the HDD stops that) only SSDs can get 6.0-7.9 (mine is 7.3 but its needs low level formatting to get it back up to full speed as I dropped it into an degraded state {I filled it up so it ran out of space} )
Other Tricks apart from using RAID 0 you can do is short stroke the disk to get the best performance out of the drive (300gb/700gb or 400 600) OS programs games C: other stuff D: (this is how i would setup any High end HDD system same for RAID 0 as well below)
Best way is really for speed is RAID 0 + 500GB/1.35TB partition for an HDD {as raid 0 would give you 1.86TB of disk space} or buy an C300 or alike SSD (Not raided as that’s Pointless for an SSD as they are 10-50x faster then an HDD on random data rate and you lose TRIM support in SSD raid)
The only thing I can add to what has already been said is that the two large dips in the read benchmark graph could be due to “difficult” sectors that require several retries. Otherwise they could be the result of some background Windows process interfering with the test. If you can reproduce them on one drive but not the other, then you may have reason to be wary of it. The drive’s rated maximum sustained data transfer rate of 126 MB/s is close to what you see on the graph.
As for the RPM, the spread of data points in the access time graph spans about 8msec, which is the latency of one complete rotation for a 7200 RPM drive. A 5400 RPM drive would have a latency of 11ms. Therefore Sandra is wrong.
