Comments on: Measure Actual Hard Disk Performance Under Windows

By: Dean Hiller

Dean Hiller — Mon, 31 Jan 2011 18:47:19 +0000

hmmm, I think this tool misreport MB(bytes) and it should be Mb(bits). I ran Iometer and it reports in MB and I got about 5MB which is 40Mb and this tool told me my drive was 40MB which must be Mb actually instead.

ie. you need to divide by 8 if you want Megabytes instead of it reporting in bits I beleive. This makes more sense and is in line with my other program I wrote then.

By: Dean Hiller

Dean Hiller — Mon, 31 Jan 2011 18:15:10 +0000

very odd, I ran the test and wrote a program that wrote out bytes as well. I only wrote a small file out and got a through put of

INFO: throughput=177053.82436260625 bytes/second
INFO: throughput=1416430.59490085 bits/second
177kB/sec or 1.4Mb/sec (bytes vs. bits as I am comparing to network speeds for an architecture I am working on right now)

This is much much slower than the results I am told by the other tool….very odd…any ideas why? (The program is in java so I wonder if it is because I am not using native buffers or something to optimize the speed).

results from tool
1863MB in 44.48s = 41.9MB/s
1863MB in 43.12s = 43.2MB/s
1863MB in 501.81s = 3.7MB/s

later,
Dean

By: Ernie

Ernie — Wed, 16 Jun 2010 20:40:15 +0000

This is an extreme situation, but I think it’s working. I think I have maxxed out the bandwidth of PCI-E x8:

Areca ARC-1231 PCI-E x8 card with 4 GB Buffer and 8x Intel 160 GB 34nm SSD drives in RAID 0 configuaration:

seq write: 1024MB in 1.09s = 936.8MB per second
seq read: 1024MB in 0.23s = 4371.8 MB per second
rdm read: 1024MB in 0.23s = 4371.8 MB per second

By: Don

Don — Thu, 08 Apr 2010 00:07:23 +0000

Don’t get too excited with the results yet.

The laptop 5,400 could be faster than the desktop 7,200 for the following reasons:

- Has larger cache built in
- 7,200 RPM drive is on last legs, or greatly reduced
- Windows set a lower DMA value in Device Manager
- 2 different windows versions used
- Need new controller driver for your Sata drive
- Malware/Spyware/Viruses/Fragmentation affect drive
- Laptop has fewer startup items (so loads faster)
- Drive is nearly full on desktop, causing disk thrashing
- Windows is optimized for speed on laptop

I am sure there are many other reasons to include on why this is interesting, but definitely NOT a final conclusion.

Raymond? Might I suggest?

Get an external drive, update your Sata controllers, & test both drives via high speed USB 3 to see what happens!

By: RMM

RMM — Mon, 15 Mar 2010 17:06:18 +0000

@ravi
there is something wrong with your drive (is an IDE or SATA drive?). If it is IDE check your DMA mode under windows.

or the Safeboot encryption slows your I/O down…

By: RMM

RMM — Mon, 15 Mar 2010 17:03:08 +0000

DiskTT FLUSHES the cache after write thus writing all the cached data to the disk. It does not CLEAR the cache after write. So some or maybe all data written to the disk will be in the cache when the read test start. This explains the extreme high read values.

BTW there is a new version 2 of DiskTT that doesn’t use the cache at all by default. It will still test the thruput of a disk system as experienced by a normal application.
Try it!

By: nFectious

nFectious — Mon, 15 Mar 2010 06:43:26 +0000

It locks up on me at 4% complete on the random access test. I gave it about 10 minutes at 4% and it never went anywhere after that. Had the (not responding) message and I had to close down.