Strange hard drive performance?

[Video Watcher]

I recently purchased a second hard drive (ntfs) for my laptop. I ran
some tests using HD Tach and was surprised by the results:

Disk #1: HTS548040M9AT00 (Hitachi Travelstar)
40GB, 5400 RPM, "Ultra ATA 100", 12 ms seek time
3.0 MB/s seq. read, 3.1 MB/s burst speed, 100% cpu utilization
Connected internally, 50% free space, 3% fragmented

Disk #2: ST3250623A-RK (Seagate Barracuda)
250GB, 7200 RPM, "Ultra ATA 100", 16MB cache
23.9 MB/s seq read, 26.5 MB/s burst speed, 4% cpu utilization
Connected via 1394a (50MB/s), 98% free space, 0% fragmented

1. Why is Disk #1 only getting 3.0 MB/s?
2. Why is Disk #1 using 100% of my CPU?
3. Why is Disk #2 not getting the 49 MB/s read speed others report?
http://www.gamepc.com/labs/view_content.asp?id=sg7200plus&page=4

Any ideas or help appreciated!

 

[Video Watcher]

3. Why is Disk #2 not getting the 49 MB/s read speed others report?
http://www.gamepc.com/labs/view_content.asp?id=sg7200plus&page=4

I replaced Firewire400 with USB 2.0 and now I'm getting 32.5 MB/s seq
read, 33.7 MB/s burst, and 18% cpu utilization.

I thought 1394a (400Mbps) was often faster in practice than USB 2.0
(480Mbps), and either should be much faster than this?!

23.9 x 8 = 191 Mbps. 32.5 x 8 = 260 Mbps.

 

[Rod Speed]

I recently purchased a second hard drive (ntfs) for my laptop.
I ran some tests using HD Tach and was surprised by the results:

Disk #1: HTS548040M9AT00 (Hitachi Travelstar)
40GB, 5400 RPM, "Ultra ATA 100", 12 ms seek time
3.0 MB/s seq. read, 3.1 MB/s burst speed, 100% cpu utilization
Connected internally, 50% free space, 3% fragmented

Disk #2: ST3250623A-RK (Seagate Barracuda)
250GB, 7200 RPM, "Ultra ATA 100", 16MB cache
23.9 MB/s seq read, 26.5 MB/s burst speed, 4% cpu utilization
Connected via 1394a (50MB/s), 98% free space, 0% fragmented

1. Why is Disk #1 only getting 3.0 MB/s?
2. Why is Disk #1 using 100% of my CPU?

Likely DMA isnt being used for that drive. The OS can turn
DMA off if its seeing a high level of errors on the connection
to the drive, and it does that for the safety of your data.

3. Why is Disk #2 not getting the 49 MB/s read speed others report?
http://www.gamepc.com/labs/view_content.asp?id=sg7200plus&page=4

Those numbers are with an internal drive. Internal drives always get
significantly better thruput numbers than firewire or USB2 connected
drives.

I replaced Firewire400 with USB 2.0 and now I'm getting
32.5 MB/s seq read, 33.7 MB/s burst, and 18% cpu utilization.

I thought 1394a (400Mbps) was often faster
in practice than USB 2.0 (480Mbps),

Yes. Dunno if thats seen with HD Tach tho.

and either should be much faster than this?!
Nope.

23.9 x 8 = 191 Mbps. 32.5 x 8 = 260 Mbps.

There's a lot more involved than the physical layer speed.

You cant do better than the drive can do. Its significantly
worse than the drive can do due to protocol overhead etc.

 

[Video Watcher]

Likely DMA isnt being used for that drive. The OS can turn
DMA off if its seeing a high level of errors on the connection
to the drive, and it does that for the safety of your data.

Thanks! In Device Manager, the primary IDE channel's mode selector was
set to DMA but the current mode was PIO. I uninstalled / re-installed
the channel and now I'm seeing 93.5 BM/s burst times and 27.8 MB/s
sequential read times--a huge improvement! Disk #1 is ok now.

Those numbers are with an internal drive. Internal drives always get
significantly better thruput numbers than firewire or USB2 connected
drives.



Yes. Dunno if thats seen with HD Tach tho.

It's fishy that USB 2.0 is so much faster. I'll try other benchmarks.

There's a lot more involved than the physical layer speed.

You cant do better than the drive can do. Its significantly
worse than the drive can do due to protocol overhead etc.

The drive can handle 394 Mbps. Protocol overhead doesn't explain only
hitting 191 Mbps because LaCie external firewire drives have clocked
280 Mbps and Seagate external firewire drives have clocked 328 Mbps. I
paid a pretty penny for this disk so I want to maximize performance.
Any ideas that I could try? I have a well-regarded Adaptec enclosure.

http://millimeter.com/news/video_lacie_breaks_hd/
http://www.everythingusb.com/hardware/index/Seagate_200GB_Push_Button_External_Hard_Drive.htm

 

[Rod Speed]

Thanks! In Device Manager, the primary IDE channel's mode selector
was set to DMA but the current mode was PIO. I uninstalled / re-installed
the channel and now I'm seeing 93.5 BM/s burst times and 27.8 MB/s
sequential read times--a huge improvement! Disk #1 is ok now.

It's fishy that USB 2.0 is so much faster.

Yeah, specially when all the other tests show
the reverse, with firewire being a little faster.

I'll try other benchmarks.

Yeah, I would, there might be some problem with
the particular firewire implementation or something.

The drive can handle 394 Mbps.

Where are you getting that number from ?

Protocol overhead doesn't explain only hitting 191 Mbps because
LaCie external firewire drives have clocked 280 Mbps and Seagate
external firewire drives have clocked 328 Mbps.

True if those are real numbers. That last is harder to be sure of tho.

I paid a pretty penny for this disk so I want to
maximize performance. Any ideas that I could try?

I'd first ensure that those numbers are real.