PDA

View Full Version : Hitachi sata 3gb/s bench

The Saint
10-20-2005, 10:33 AM
...guys I benchmarked my drive (Hitachi Deskstar T7K250 Sata 3Gb/s) and this is what I got....

very strange no?

http://img467.imageshack.us/img467/2884/untitled4vk.jpg
ScYcS
10-20-2005, 11:16 AM
Strange indeed.

Did you download the Hitachi tool to enable SATAII? I heard from the factory, they disabled the SATAII feature.
The Saint
10-20-2005, 11:39 AM
yes I did....

I'm going to try again though...
ScYcS
10-20-2005, 11:41 AM
Which mobo is it in and which connectors are you using? Maybe it's something strange as the DFI nforce3 board had, where you HAD to put it on the 3rd or 4th sata connector to get speed.
The Saint
10-20-2005, 11:46 AM
it is an Asus A8N SLI Premium...and I'm connected to SATA 1 port....

:(
davidlem
10-20-2005, 01:07 PM
I'm trying to understand - you're comparing a single 160GB 7K250 against 3 RAID-0 arrays and getting lesser results...OK so the numbers aren't excellent but I wouldn't expect to match those other tests either.

Take a look here for example results.
http://www.storagereview.com/articles/200310/20031007HDS722525VLSA80_2.html
The Saint
10-20-2005, 01:19 PM
NO, I'm not even looking at the comparison table, just wondering if 52MB/s is a realistic number for a 3GB/s Hdd.....
DougLite
10-20-2005, 01:51 PM
Sandra is an inconsistent benchmark that lacks repeatability. Also, it does not properly account for locality, which is what your drive's firmware is tuned for - the Sandra tests are either entirely equential or entirely random.

However, The synthetic results look pretty good, and what I'd expect from your drive.
The Saint
10-20-2005, 01:55 PM
oh ..thanks so very much for your input and help DougLite...
you have just calmed me a lot :p
defakto
10-20-2005, 02:01 PM
You have to remember 3gbps is not what the drive will transfer at, it's only the theoretical maximum rate of the interface. Closest performing drive right to 3gbps are the nice, fast 15k scsi drives which put out (very roughly) 95 MB/s which is about .760 gbps
xonik
10-20-2005, 06:19 PM
NO, I'm not even looking at the comparison table, just wondering if 52MB/s is a realistic number for a 3GB/s Hdd.....Note that it's not 3 GB/s but 3 Gb/s (gigabits per second). Converting this into gigabytes per second and excluding command overhead, the theoretical maximum transfer rates comes to 300 MB/s (megabytes per second).
davidlem
10-21-2005, 08:50 AM
Note that it's not 3 GB/s but 3 Gb/s (gigabits per second). Converting this into gigabytes per second and excluding command overhead, the theoretical maximum transfer rates comes to 300 MB/s (megabytes per second).

Or another way of doing the math would be SATA = 150MB/s, SATA2 = SATA*2 (theoretical throughput)
ScYcS
10-21-2005, 10:06 AM
Don't you guys think that his random write scores are a little on the slow side for this drive? I thought that's what the OP was pointing at?!
DougLite
10-21-2005, 10:20 AM
^ Once again, not really. Writes are always slower than reads, as sophisticated error detection mechanisms must be applied to verify that the correct data is written to the disk. Also, many drives have write caching schemes that are designed to preserve read performance, often until the load of requests for the drive is 100% writes.

EDIT: I just noticed the fixed IO queue depth of four on the Sandra tests. Once again, that does not properly reflect a desktop environment. Typical queue depths during desktop application work are around 1-2.