Harddisk drivers

Hello all!

I have a problem, possibly with my 120 GB / 8 MB cache Hitachi 7K250. When benchmarking it and comparing it to my 40 GB / 2 MB cache IBM 60GXP, it scores good results and under Windows XP Pro, I have no complaints.
See the link below for benchmarks, the 7K250 is green and the 60GXP is yellow:

http://img33.photobucket.com/albums/v98/Engdahl/AIDA32/7K250_QLR.jpg

However, when running Norton Ghost from a Ghost Boot Disk, I have very disappointing transfer rates. The 60GXP checks an Image at 404 MB/min (6.74 MB/sec) compared to the 7K250 which does the same job at 142 MB/min (2.37 MB/sec).

These rates are just a fraction of the rates shown in the benchmarks but I suppose that working under DOS makes a difference compared to working under XP due to different drivers (?) but according to the ghststat.txt (a text file which can be written to the floppy, holding information on the ghost status) they both operate in UDMA mode.

If anyone has an explanation to why the 7K250 gets more degraded than the 60GXP, then I´m all ears!

Thanks for Your time,
Christer

 

Hi guys,
thanks for Your response!

Yes, I´m convinced of that too but what made me curious is that it affects the 7K250 substantially more than the 60GXP.

Is smartdrive included on a Win98SE or WinME start disk?
I have used both to start the computer and next, have loaded ghost.exe from a separate floppy. The results were as disappointing as with the Ghost Boot Disks.

Ghost reports both drives to work in UDMA mode but maybe it´s only reading a "label ".

Christer

 

Well, I found out that smartdrv.exe is not included by default.

I have found it on a WinME computer and will see what happens.

Christer

 

I added smartdrv.exe to a WinME start disk. I started the computer from it and when running at the A: command promt, I typed smartdrv.exe and hit enter. It hummed away for a few seconds.
Next, I started ghost.exe from a separate floppy.

No change in rates but I believe that it isn´t as simple as this to make smartdrv work properly.

All references I found on the web regarding smartdrv.exe was from the early nineties. It seems to me that Ghost should have developed beyond that and the below indicates that.

On a friends computer (Seagate 7200.7 SATA to 7200.7 plus PATA), Ghost shows rates of 1.500 MB/min. That´s ten times the rates for my 7K250. His processor is a 3.0 GHz P4 which probably makes a difference compared to my 1 GHz Thunderbird.

1.500 / 3 = 500 which still makes me a bit disappointed ...... ......

Christer

 

Hi Matt!

I´ve tried all combinations of hardware, I have taken all out but one HDD.

As the single drive in the system, the 60GXP creates an image from one partition to another (on itself) at some 400 MB/min compared to the 7K250 under the same conditions at some 140 MB/min.

With only the two HDDs connected, creating images from one to the other gives the same results. Slow with the target partition on the 7K250 and quick(er) with the target partition on the 60GXP.

His computer is based on an ASUS P4P800DeLuxe motherboard with a P4 3.0GHz/800MHz processor and 2*512MB Dual Channel RAM.
The Hyper Threading and Dual Channel doesn´t work under DOS, though. The neat little P4/HT logo changes to a P4 logo when booting to DOS.

I think so too. I have heard/read about something called "command queing ". Some drives have it and others don´t.

I don´t know what it is ...... ...... but it may be a factor!

Christer

 

The 7K250 has 8 MB cache and the faster 60GXP has 2 MB cache.

Both target partitions are FAT32, 8 kB clusters on the 60GXP and 16 kB clusters on the 7K250.

All devices jumpered Cable Select.

7K250 Primary Master, 60GXP Primary Slave.

When time permits, I´ll rearrange to have the below setup:
PM = 7K250
PS = DVD-ROM
SM = 60GXP
SS = CD-R/RW

I doubt that it makes a difference since the 7K250 is slow when the target partition is on it.
If the source partition is on the 7K250 or on the 60GXP doesn´t make any difference whatsoever.
With the target partition on the 60GXP, it is always faster, no matter where the source partition is located.

Target partitions are 15 GB (60GXP) and 18 GB (7K250), which give 8 kB and 16 kB clusters respectively. The way I understand it, that would be an advantage for the 7K250.

The time to create and to check the image varies with the size of the source partition but the transfer rates are always 2.5-3 times slower when the 7K250 holds the target partition.

I have timed image creation and image checking but have not had a reason to restore an image ...... ...... just yet!
I would believe that creating an image is the same as writing any other file.
Restoring an image may be different if it isn´t restored to the original source partition (resizing).

Well, I quit counting several weeks ago but I will "leave no stone unturned "!

Christer

 

Well Matt, I have the Ghost Images of the system partition on the 60GXP (and the 7K250 - both belt and suspenders) and have backed up everything on the 7K250 to the 60GXP. It is, as You point out, a risk and it wouldn´t surprise me if Murphy is waiting in the shadows.

In the mean time, I´ve been juggling connectors and checked my smallest Image of 1.026 MB on different hardware setups.

First, a reminder of the hardware setup that I normally use and the rates:

Primary Master - 7K250 - 143 MB/min
Primary Slave - 60GXP - 425 MB/min
Secondary Master - CD-R/RW
Secondary Slave - DVD-ROM

Next, after juggling connectors:

PM - 7K250 - 143 MB/min
PS - DVD-ROM
SM - 60GXP - 135 MB/min
SS - CD-R/RW

The reduced rate for the 60GXP indicates that in DOS environment, it makes a difference if a UDMA device is mixed with a non-UDMA device on the same channel.

PM - 7K250 - 143 MB/min
PS - none
SM - 60GXP - 425 MB/min
SS - none

The restored rate for the 60GXP supports the mixing theory.

The 7K250 alone on Primary, nothing on Secondary - 143 MB/min
The 7K250 alone on Secondary, nothing on Primary - 137 MB/min

There is a slight difference between the channels but I don´t know why. Maybe it´s performance scatter?

Christer