HD Tach reliable for benchmarking disk throughput?

I'm a bit confused with some results I got from HD Tach regarding HDD throughput #s. On one computer I have:

Athlon 64 4000+
300GB WD3000JB
Average read: 53.4 MB/s
Burst read: 93.5 MB/s

On another:
Dual Xeon 2.8 GHz
6 x ST3146854LC (146GB, 15,000 RPM U320) in RAID 5
Average read: 65.7 MB/s
Burst read: 99.6 MB/s

These #s came from the "quick bench" option in HD Tach 3.01

I don't claim to be a HDD expert, but I expected a much bigger difference between a run of the mill ATA disc and a 15k U320 array. The PERC card has the latest firmware and drivers. Any ideas why the performance is so lackluster?
6 answers Last reply
More about tach reliable benchmarking disk throughput
  1. Hit storagereview.com for some great comparisons between desktop and server drives.

    Bottom line is your server is tuned for mutilple access from many users, database access patterns etc.

    You will also find other benchmarks to run there if you want.
  2. Quote:
    Not useful in your case. You need to measure IO/s rather than MB/s of STR as pointed out by tphillips63.

    Download Intel's IOMeter (now maintained by sourceforge) and run 'File Server' benchmark.

    Thanks for the tip, I'll give it a shot.
  3. Its reliable for testing throughput, but for many uses throughput is not the most important metric.
  4. 10K and 15K drives do have fast performance, but they really shine with server I/0 requests with databases and web servers. I use HDTach, but I would say a better comparison would be to use Sisoft Sandra. This will give you a better idea between the two drives. There should be a night and day difference.

    My System:
  5. A related question.
    What is the best metric to use in comparing SOHO desktop HDs? Is HD Tach appropriate for this venue?
Ask a new question

Read More

Hard Drives HD Storage