Some latest test results using USB2.0 external HD

[DZM]

I am able to test USB2.0 external HDs with different HD and interfaces

here are the results
http://www.nuassembly.com/USBHD/HardfactsUSB20.htm
Tests are done with 1 file 1.17GB, and copied around
Obviously, in terms of copying one file, the monster 7K60 NB 2.5HD beats a desktop 60G 3.5HD possibly because of the buffer size?
ALSO, the USB2.0HD with a Myson CS8813 controller IC and a could 8MB HD could beat a HD with much lower RPM and 2MB Notebook 2.5" HD easily.

MB buffer in a HD really affect the USB2.0 external mode performance, right? 7x!!!

 

[Snugglebear]

Don't think I'm trying to rain on your parade, but here's some things you're missing. Make sure your methodology is more detailed than it is. Copying a file from one drive to another is a decent test, but make sure you record the Windows version, that the file is defragged, ensure no other applications are running or accessing the disk during the test, what controller is in the computer(s), which driver revision is being used, what the drive firmware settings are (some may have acoustic management bits set, or have SMART turned off, etc.), what mode the disks are operating in, whether caching is enabled, be sure the computers aren't going to drop into low-power mode, and, if you can get a reliable reading, what the processor usage and disk activity really is. Trust me, you need to watch all the variables otherwise the results could be skewed severely. Also, as a little niggle, that graph on the page isn't very descriptive. Which is the target drive and which is the source?

As for why the 7k60 can beat a 120GXP, well, look at the spec sheets. The 120GXP is almost two years old and the areal density is about 60% of the 7k60. If you disabled write caching the 7k60 would still beat it because of that higher density. Besides, on massive contiguous writes, cache isn't going to do a whole lot for performance.

 

[DZM]

Don't think I'm trying to rain on your parade, but here's some things you're missing. Make sure your methodology is more detailed than it is. Copying a file from one drive to another is a decent test, but make sure you record the Windows version, that the file is defragged, ensure no other applications are running or accessing the disk during the test, what controller is in the computer(s), which driver revision is being used, what the drive firmware settings are (some may have acoustic management bits set, or have SMART turned off, etc.), what mode the disks are operating in, whether caching is enabled, be sure the computers aren't going to drop into low-power mode, and, if you can get a reliable reading, what the processor usage and disk activity really is. Trust me, you need to watch all the variables otherwise the results could be skewed severely. Also, as a little niggle, that graph on the page isn't very descriptive. Which is the target drive and which is the source?

As for why the 7k60 can beat a 120GXP, well, look at the spec sheets. The 120GXP is almost two years old and the areal density is about 60% of the 7k60. If you disabled write caching the 7k60 would still beat it because of that higher density. Besides, on massive contiguous writes, cache isn't going to do a whole lot for performance.

I totally agree, and I will have a more complete report soon. But, the most obvious thing is that the MB buffer really matters in the USB 2.0HD drive performance!

 

[DZM]

I have added more pictures and results.
http://www.nuassembly.com/USBHD/HardfactsUSB20.htm

It is obvious that the buffer size in the HD that connected to USB2.0 controllers really matters. With 8MB cache, the 5.4KRPM HGST USB HD easily beats a 4.2KRPM 2MB cache 80G HD inside a notebook (2'23" v.s. 3'58"). With the 7.2KRPM 8MB external 3.5" HD, the performance is further enhanced. I wonder what is the USB performance if the HD cache is increased to 16MB?

It is also obvious that the Myson CS8813 outperforms the Genesys G811E easily (2'23" v.s. 3'10"), and I believe that with a 3.5" 7.2KRPM 8MB HD the Myson CS8813 could even compete with a desktop computer (2'17").

 

[Ice Czar]

DZM

if youd be so Kind as to incorporate the information posted here by lessthanjakejohn & PassMark

and your other thread
http://www.hardforum.com/showthread.php?s=&threadid=745447
and this one into a new FAQ not just a link

I'll sticky it
Ideally it would be cautions to take for adopting USB & Firewire Externals in comparision to Internals as SATA or ATA
(might include a note on external SATAs)

know Im asking for alot of work...but strong arming FAQs is part of my job

If you do PM me