SATAI or SATAII

  • Thread starter Thread starter Reaper
  • Start date Start date
DevilsPGD said:
In message <[email protected]> Odie Ferrous


I'd guess closer to 55MB/s, based on the fact that I moved 16.1GB in
just under 5 minutes (290-something seconds)

Real world "test", I was moving some disk images from one drive to
another...

*shrugs*



Can you please state the hardware, number of disks, RAID and controller
types ?

The best performance I've ever seen was four WD 74GB Raptors in RAID0 on a
3Ware 9500 series controller card.

At ~$1450CND, it's not an inexpensive option.
 
Can you please state the hardware, number of disks, RAID and controller
types ?

The best performance I've ever seen was four WD 74GB Raptors in RAID0 on a
3Ware 9500 series controller card.

At ~$1450CND, it's not an inexpensive option.


55MB/s doesn't seem all that far off.

I had a system open that I'd been working on and out of
curiosity did a random test (not even trying to maximize
performance).

KM266 board, Southbridge VT8233 controller, each drive below
as master on opposing IDE channels.

Source - Maxtor Plus 8 40GB, 7K2 RPM, 2MB cache

3.67GB data as two files on the last (slowest) 1/3 of the
drive. I'd just defragged it's FAT32 partition prior to
making an image of the OS partition. In other words, this
old 40GB HDD was reading from 28GB through ~32GB region on
the platters.

Destination - Maxtor Plus 9 120GB, 7K2 RPM, 8MB cache New
drive, FAT32 empty primary partition. All FAT32 partitions,
all 32K clusters. Note "new drive" in preceeding sentence,
at one point Maxtor had their drives set to write-verify for
X number of power cycles and I have no idea they still do so
with current drive models, if this drive is currently doing
so, but it has only been power cycled about 4 times thus
far.

89 seconds
42 MB/s

40GB drive's primary partition held the OS but AFAIK there
was no OS or app related drive access during the transfer.

Subsequent test with HDTack showed the 40GB drive had
roughly 45MB/s over that region of the drive. It appeared
to clearly be the bottleneck.

Next I copied same two files to an extended partition on the
120GB drive, but unfortunately got distracted and lost count
of how long that took. 42:30 45:13
Even copying from the secondary to the primary partition on
the same drive which "might" be doing write-verify the whole
time, total time was only 163 seconds, 23MB/s.

Next I copied the same two files, 3.67GB of data from the
primary partition of the 120GB new HDD to the primary
parititon of the old 40GB HDD. 40GB drive had also been
recently defragged but since it had OS on it too, file was
presumably written to about 7GB through 11GB portion of
space on the platter. 76 seconds, 49 MB/s

I was a little surprised this older 40GB drive could do
this, so I rebooted just to be sure nothing was cached.
Subsequent trial it took 77 seconds, close enough, still
rounds off to 49 MB/s

Above figures don't guarantee 55MB/s but it seems easily
possible with large files, even with low-end drives and
ATA100 or better controller.
 
In message <b4X1f.121094$tl2.18946@pd7tw3no> "S.Heenan"
Can you please state the hardware, number of disks, RAID and controller
types ?

Sure :)

The system is as follows:

CPU: AMD Athlon64 3000+
Motherboard: Gigabyte GA-K8NXP-SLI
RAM: 4x512MB PC3200 DDR
Onboard SATA+RAID controller: NVIDIA chipset

I also have an additional SATA RAID controller, which is not being used.
The NVIDIA controller has RAID disabled (in the BIOS)

The NVIDIA SATA controller has 4 ports, ports #1 and #2 have WD Raptor
36.7GB drives, #4 has a Seagate 400GB/7200rpm/8MB/NCQ drive. NCQ is
enabled.

Port #2 had a drive connected, but there aren't any partitions created
yet. I'm planning to RAID it with #1, but at the time when I performed
the installation I had some files which I needed to recover and I
haven't had time to set up RAID.

I was simply copying files from one drive to the other using a JPSoft
TCMD command line. Time was measured by virtue of the fact that my
command prompt includes a timestamp (and yes, that means it could have
sat at a command prompt longer then I thought. I actually used the time
from when the previous command finished until the copy finished, I
assumed that I took no time to type the command. This is likely roughly
correct since I had just finished deleting a bunch of files I no longer
needed and was copying the rest.

While anything is possible, there is no logical reason that any of the
data would have been cached before the process started, the files had
not been opened at all since Windows was installed (I was restoring
files to their new placements after reinstalling)
The best performance I've ever seen was four WD 74GB Raptors in RAID0 on a
3Ware 9500 series controller card.

At ~$1450CND, it's not an inexpensive option.

$200 for the motherboard, $200 for one WD Raptor drive over a year ago,
$315 for the Seagate 400GB drive.
 
Back
Top