Expected performance of external USB 2 hard drive

[David Dyer-Bennet]

The drive in the external box is a WD Raptor 7200 RPM 80GB drive. The
USB2 port is on an MSI motherboard, and the system is running Windows
2000. It seems, to me, to be running *extremely* slowly; but I have
little experience with external drives newer than SCSI. So, is there
some rough rule-of-thumb expectation for what kind of read and write
throughput I should get?

(My performance test was backing up a big chunk with NTI Backup Now
and then "verifying" it, so straight sequential write and later read
from an unfragmented newly-formatted disk. Oh, I formatted it NTFS.)

 

[CWatters]

The drive in the external box is a WD Raptor 7200 RPM 80GB drive. The
USB2 port is on an MSI motherboard, and the system is running Windows
2000. It seems, to me, to be running *extremely* slowly; but I have
little experience with external drives newer than SCSI. So, is there
some rough rule-of-thumb expectation for what kind of read and write
throughput I should get?

When I last ran tests on my three year old PC the USB 2.0 hard drive was
about half the speed of the internal drive (same WD drive in both). Try
running the free version of SiSoft Sandras on it and report what thats says.
Perhaps you have it connected to a USB 1.1 port or the port isn't set up
correctly.

http://www.sisoftware.net/

 

[Folkert Rienstra]

The drive in the external box is a WD Raptor 7200 RPM 80GB drive.

You should be reported for torturing a second generation Raptor.

The USB2 port is on an MSI motherboard, and the system is running
Windows 2000. It seems, to me, to be running *extremely* slowly;

It should be able to do at least half it's maximum STR.

 

[John .]

You should be reported for torturing a second generation Raptor.


It should be able to do at least half it's maximum STR.

I just tested my hard drives (2 Western Digital w 8mb caches) with
SiSoftware. I got 27MB/sec and 26MB/sec (Windows xp).

I also have an external WD 250gb (8mb cache) on USB2.0 port. It
tested at 17 MB/sec.

Does that sound about right?

 

[CWatters]

I just tested my hard drives (2 Western Digital w 8mb caches) with
SiSoftware. I got 27MB/sec and 26MB/sec (Windows xp).

I also have an external WD 250gb (8mb cache) on USB2.0 port. It
tested at 17 MB/sec.

Does that sound about right?

Yes. I got similar results (if I remember correctly) and same here with
Maxtor drives...

http://compreviews.about.com/od/cases/l/aaBlknUSB2Enc.htm

MethodDrive Index
Internal IDE 21,132 KB/s
USB 2.017,406 KB/s

 

[Folkert Rienstra]

Yes.

For Sandra, maybe. The results are obviously rubbish.

I got similar results (if I remember correctly) and same here with
Maxtor drives...

http://compreviews.about.com/od/cases/l/aaBlknUSB2Enc.htm

MethodDrive Index
Internal IDE 21,132 KB/s

USB 2.017,406 KB/s

No kidding.

 

[Folkert Rienstra]

I just tested my hard drives (2 Western Digital w 8mb caches) with
SiSoftware. I got 27MB/sec and 26MB/sec (Windows xp).

I also have an external WD 250gb (8mb cache) on USB2.0 port. It
tested at 17 MB/sec.

Does that sound about right?

Nope. That's awful.

 

[CWatters]

No kidding.

What results do you get?

http://www.pcstats.com/articleview.cfm?articleID=1015
SiSoft Sandra 2001 Benchmark Results
40GB 7200 RPM Samsung HDD
HDD BenchmarkScore
1.USB 2.0 11462
2.USB 1.1 974
3.Hard Drive 26544

 

[Folkert Rienstra]

Presumably that was to read USB 2.0 17,406 KB/s then.

What results do you get?

And what has that got to do with my comment?
And what's with the snipping, there were hardly any lines in that post.

It's well known that Sandra is a broken benchmark.
What is it to you what (broken) results *I* might get?

http://www.pcstats.com/articleview.cfm?articleID=1015
SiSoft Sandra 2001 Benchmark Results
40GB 7200 RPM Samsung HDD
HDD BenchmarkScore
1.USB 2.0 11462
2.USB 1.1 974
3.Hard Drive 26544

Another wildly differing result. Faster drive yet USB is even worse.
Obviously these USB values are totally bogus.

 

[CWatters]

It's well known that Sandra is a broken benchmark.
What is it to you what (broken) results *I* might get?

Sorry, what I mean is... What should the results be using something other
than SiSoft Sandra?

The results might be "obviously rubbish" to you but not everyone!

 

[Folkert Rienstra]

Sorry, what I mean is... What should the results be using something other
than SiSoft Sandra?

What the drives can do or, if that's faster than USB can deliver, limited
to say 2/3(*) the clockrate (480Mbit/s), so let's say 32 MB/s or so.
Probably depends with transfer size whether you can quench out a little
bit more.

The results might be "obviously rubbish" to you but not everyone!

They should be, knowing that USB 2.0 is 480Mbit/sec.

(*) Read somewhere that USB (12Mbit/sec) was designed/intentioned
for devices upto 8Mbit/sec, hence 33% suspected overhead.
USB 2 may actually do better but I can't find more current data.