Testing
Before I get straight into the Benchmarks I want to expand on a point I touched upon earlier. This RocketRAID 1542 is dependant on the speed of the PCI Bus, and therefore its performance is going to be limited by the PCI Bus and what ever else may be trying to use the Bus at the same time, as well as the limitation of ATA133 speeds thanks to its use of the HPT374 controller. What this means is you are not going to see performance figures anywhere near their theoretical limits. We are also going to be limited by the drives themselves, since we are using ATA100 drives with RocketHEAD 100 adaptors.
The Test system will be:-
AthlonXP 2500+ @ 2.24gig (11.5x195) (), MSI K7N2 Delta-ILSR, Corsair XMS1024 TWINX PC4000, WindowsXP (SP1 and all other updates), 1x 40gig IBM Deskstar 60GXP, 1x 60gig IBM Deskstar 60GXP (using RocketRAID 100 Adaptors), RocketMATE 1100 e.SATA enclosure
(NB; No other cards were used in the PCI slots, so as to allow the maximum amount of bandwidth available to the 1542 card.)
Test software will be:-
File copy of a 724mb Disk Image
File copy of 26 mp3's, 120mb
Timed System Bootup
For the RAID 0 array a block size of 64k was used and in all tests, the same cables and drives were used. For the single disk tests the 60gig Deskstar was used, connected to the onboard Fastrack controller as well as the 1542. In the case of the File copy tests, the files were copied to the drive/array and from the array copied to another location on the same partition. All tests were done right after a clean format, and in all test cases 3 dry runs were performed before a 4th run to use as a final result. In the case of the Timed System Bootup test, the system was powered on and the timer started as the POST for the motherboard had finished, and was stopped when the Windows login screen was presented. For all of the timed tests, bare in mind that this was done rather unscientifically with a stopwatch, so allow plus or minus .5 seconds for human error. For the RocketMATE 1100 we used the 40 gig drive.
HD Tach
provides insight into the read capabilities of your drives presenting the information in a nice graphical format.
From left to right, Onboard 1 disk, 1542 1 Disk, 1542 RAID 0
From left to right, 1542 RAID 1, RocketMATE 1100
There really isn't much to see out of the norm here. Everything performs as is expected, with RAID 0 results showing off a nice boost. However, whilst it shouldn't have been a surprise, I was pleased to see that the RocketMATE kept up with the other drives in the system as though it was just a normal internal drive.
SiSoft Sandra File System Benchmark
File System Benchmark provides a drive index score after testing the speed across the platters of your drive. Like HD Tach it is another synthetic benchmark but it does provide a nice way of performing easy comparisons between systems.
Higher is better
Again we see the boost afforded us by the RAID 0 configuration, the slowdown of the RAID 1 scores, and once again the RocketMATE has no trouble at all in keeping up with the rest of the pack.
File Copy 1 - 742MB Disk Image
I imaged the second disk of Soldier Of Fortune 2, copied the image to the drive/s attached to the controller and proceeded to time how long it would take to copy that image from one location to another on the same partition after clearing the cache.
Lower is better
File Copy 2 - 120MB, 26 mp3's
Copying lots of smaller files can also give as an idea of the writing capabilities when smaller files are concerned.
Lower is better
File Copy times show once again the nice drop in time it takes to transfer files under RAID 0. It also clearly shows the extra few seconds taken for RAID 1 to write to both drives. Once again, the RocketMATE has no problems keeping up.
Timed System Bootup
With a clean copy of Windows XP installed onto the drives, I proceeded to time how long it would take from the end of POST to get to the Windows Login screen. This kind of test shows us a bit of the reading capabilities.
Lower is better
Like the 1520 card, the 1542 adds 24 seconds to your bootup time which almost doubles the time it takes the RAID 0 to bootup anyway. Very much broken record comments here, everything performs as it should.
NEXT
