interesting findings about USB HD

D

DZM

Weaksauce
Joined
Mar 17, 2004
Messages
125
I also did some test with USB external HD. I would like to share with you on the results.

This is the test I did and the results:
1. I prepared some 12 files each with size of 150~160M, total 1.77G
2. I copied them from one directory to another with "copy all" and
then "paste all"
Results A: on a Notebook (Compaq 2100Z 5400RPM 80G, 15% full) HD, it takes 405 seconds
(i.e. reads and then writes 1.77G in one HD).

Results B: on a desktop HD (IBM IC35L060AVV60G, 7200RPM, 2MB, 30% full), it takes 230 seconds

Results C: with a USB2.0 HD and a IBM 15G (7200RPM, 2MB) 3.5" HD, I copy the files from
a desktop internal 3.5" HD (7200RPM, 2MB 60GHD) to the USB2.0HD and it takes 174 seconds

Results D: with a USB2.0 HD and a IBM 15G (7200RPM, 2MB) 3.5" HD, I copy the files from
the USB2.0HD to a desktop internal 3.5" HD (7200RPM, 2MB 60GHD) and it takes 140 seconds


Results E: with a USB2.0 HD and a IBM 15G (7200RPM, 2MB) 3.5" HD, I copy the files from
the USB2.0HD to the same USB2.0HD (another directory) and it takes 18minutes!!! (1080 seconds)!!!

I believe that if we do single direction read and write with USB2.0 HD, the speed is good,
maybe better than a Notebook internal HD. But, obviously the Windows driver is not well tuned
for USBHD, so the bi-directional read/write is very bad for USB2.0HD. I was told that similar
problem does not exist for Firewire HD. So, if the USB buffer problem is fixed, USB/FW HD could be replacing notebook HD if one
gets back home and have the access to wall-plug power and a jumbo 3.5" HD --- these days notebook
computers are getting as powerful as desktop computers and the only weak thing is the HD.

Some day, all the PDAs could have a powerful CPU, its clock could be changed from a few 100 MHz (on the road) to a few GHz (in home). Its features could also be changed from power saving functions (on road) to graphic intensive (in home). The storage could be all solid state (on road) to jumbo 3.5"HD (in office).
 
Just a brief note- Hitachi has come out with a 7200rpm 8mb cache 60gb laptop drive, and I'm betting with it's popularity, speed, lack of heat and noise, and about-the-same power consumption, it's not going to be the last "high performance" laptop drive from them. Although bloddy expensive, I'm considering getting one to put in my Powerbook- apparently they make a hell of a difference.
 
Originally posted by mpeg4v3
Just a brief note- Hitachi has come out with a 7200rpm 8mb cache 60gb laptop drive, and I'm betting with it's popularity, speed, lack of heat and noise, and about-the-same power consumption, it's not going to be the last "high performance" laptop drive from them. Although bloddy expensive, I'm considering getting one to put in my Powerbook- apparently they make a hell of a difference.
Click to expand...

I am sure it is a performance monster --- I wish to get one too. I noticed that it is now only $200.

But, since it is 2.5", its Sustained transfer rate (STR) is limited and could not compete with 3.5".
 
Originally posted by kleptophobiac
We have beaten this dead horse enough in Mobile Computing.
Click to expand...

I am doing quantitative analysis and test, if you could give solid explanation to the results above that will show your expertise.
 
Originally posted by DZM
I am doing quantitative analysis and test, if you could give solid explanation to the results above that will show your expertise.
Click to expand...

How did you time the transfers? Stopwatch?
 
Originally posted by kleptophobiac
How did you time the transfers? Stopwatch?
Click to expand...

I hope that you could test your USB HD again with a more modern 3.5" HD.

Because the 60GXP is a 5 year old technology, 7K60 is 2 year old, you should really compare 7K60 with 7K250.

In terms of max media transfer rate for these 3 drives, 60GXP (494Mbits/sec), 7K60 (515Mbits/sec) and 7K250 (757Mbits/sec).
 
I copied almost 40gigs from various drives on my network to a USB2.0 drive using USB1.1 (that is the only port my machines have) and it took over 16 hours.
I was copying files from 5 different drives to the USB drive.
1 drive was local to the machine the USB drive was hooked up to and the other 4 drives were on another machines.
 
Well, if you're willing to live with data from a linux machine rather than a windows machine, I will happily write a program to generate x bytes of data and write it.

The reason I wish to do it on a linux machine is that the WD1200JB drive that I have is reiserfs formatted, and has a ton of data on it. I'm not willing to format it so windows can read it.

Also, that 23MB/s I got with Sandra with the 60GXP drive is probably not going to change much with a different drive. Why? USB 2.0 is the bottleneck.

I'll go ahead and re-test for you tonight.
 
Originally posted by kleptophobiac
Well, if you're willing to live with data from a linux machine rather than a windows machine, I will happily write a program to generate x bytes of data and write it.

The reason I wish to do it on a linux machine is that the WD1200JB drive that I have is reiserfs formatted, and has a ton of data on it. I'm not willing to format it so windows can read it.

Also, that 23MB/s I got with Sandra with the 60GXP drive is probably not going to change much with a different drive. Why? USB 2.0 is the bottleneck.

I'll go ahead and re-test for you tonight.
Click to expand...

Thanks! And if possible, please test copy from the same USB HD to the same USB HD, and see if you have similar problem as I had. THG claims that USB2.0 could achieve 32MB/sec. In you Sandra test results please post the detailed results for each items.
 
You must log in or register to reply here.
Back
Top