Single user HD benchmarks

[Cruise Director]

Hi, I'm doing strictly single user stuff. My work profiles:
- compiling small to medium sized programs
- hunting and pecking all over the filesystem while building those
programs, and while downloading and installing new stuff from the
internet
- sometimes loading game levels
- sometimes image editing
- sometimes 3D modeling

I've been trying to decide whether to bother to replace my old IDE
drives with some old SCSI drives that I picked up for free. Please see
my other thread rather than ask for all the details. I've decided that
if I actually measure my current system performance, and compare it to
a database of other results, that would give me some perspective on
whether it's worth bothering.

I have been reading about www.storagereview.com 's benchmarks, but
there's a problem: I can't freely download 'em and run 'em myself.
StorageReview has renounced IOMeter in favor of some expensive test
suite that does a better job of measuring a single user's profile. So,
what do you recommend for free single user HD benchmarking? Any
websites like StorageReview that use your recommended benchmark, that
have a database of results?


Cheers,
Brandon Van Every

 

[Rod Speed]

Hi, I'm doing strictly single user stuff. My work profiles:
- compiling small to medium sized programs
- hunting and pecking all over the filesystem while
building those programs, and while downloading
and installing new stuff from the internet
- sometimes loading game levels
- sometimes image editing
- sometimes 3D modeling

Its unlikely that any of that has anything much to do with the
hard drive technology except with rather elderly obsolete drives.

I've been trying to decide whether to bother to replace my old
IDE drives with some old SCSI drives that I picked up for free.
Please see my other thread rather than ask for all the details. I've
decided that if I actually measure my current system performance,
and compare it to a database of other results, that would give
me some perspective on whether it's worth bothering.

You cant really do that, particularly with the motherboard
detail and how its affecting that work you do in that system.

I've generally found that you do get a noticeable improvement
when upgrading the motherboard, mostly due to the FSB speed,
and I usually do that with the same drives. I dont get anything
like as noticeable an effect with a new drive. But then I dont upgrade
the motherboard at a very high rate, so the effect is noticeable. In spades
when I use a dinosaur for stuff that doesnt need anything too flash.

I have been reading about www.storagereview.com 's benchmarks,
but there's a problem: I can't freely download 'em and run 'em myself.

And they dont tell you anything useful about
where the bottlenecks are in your system.

StorageReview has renounced IOMeter in favor of some expensive
test suite that does a better job of measuring a single user's profile.
So, what do you recommend for free single user HD benchmarking?

On the basics, HDTach. But its not useful for actual work you list at
the top. The only real way to quantify that stuff is to actually time it
yourself, with both technologys, and that isnt practical in your situation
unless you can borrow a SCSI controller and try it.

Any websites like StorageReview that use your
recommended benchmark, that have a database of results?

They arent useful in your situation.

 

[Eric Gisin]

If the old SCSI drives are not listed at SR, they are not worth using.
If your old IDE drives are not listed, a new IDE will be much faster.

If you want a simple tool for seeing how bad old drives are, use HD Tach.

 

[Cruise Director]

If the old SCSI drives are not listed at SR, they are not worth using.
If your old IDE drives are not listed, a new IDE will be much faster.

My SCSI drives are listed. I was just thinking empirically. I had
forgotten that my IDE drives might also be listed. Will check.


Cheers,
Brandon Van Every

 

[Cruise Director]

My SCSI drives are listed. I was just thinking empirically. I had
forgotten that my IDE drives might also be listed. Will check.

The drives under consideration:
Fujitsu MAJ3091MP
Seagate Cheetah ST318451LW
Western Digital Caviar 200BB
Western Digital Caviar 300AB

StorageReview benches the Fujitsu and the Seagate. They bench a Caviar
400BB and a 600AB, which may differ only in 2x the number of platters.
Best I can come up with. IOMeter scores do indicate a roughly 50%
performance improvement of the SCSI drives over the IDE drives, "in
some bullshit respect" that I'm not entirely confident about, given
that StorageReview has utterly renounced the IOMeter Workstation
benchmark's validity. The Fileserver and Database benchmarks show a
100% performance improvement. These scores I think are more reliable
and to be expected, but I'm not currently planning to use my box as a
server.

IOMeter - Workstation Access Pattern - Load = Linear *

Western Digital Caviar WD400BB (40.8 GB ATA-100) 88.61 0.69 11.28 ms
0.35 % 253.17
Seagate Cheetah 18XL (18.4 GB Ultra160/m SCSI) 127.60 1.00 7.84 ms
0.47 % 271.49
Western Digital Caviar WD600AB (60.0 GB ATA-100) 78.30 0.61 12.77 ms
0.32 % 244.69
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) 131.03 1.02 7.63 ms 0.45 %
291.18

So, anyways, these StorageReview scores don't help much for trying to
compare single user performance. Maybe the SCSI drives are 50% faster,
maybe they aren't.


Cheers,
Brandon Van Every

 

[Rod Speed]

The drives under consideration:
Fujitsu MAJ3091MP
Seagate Cheetah ST318451LW
Western Digital Caviar 200BB
Western Digital Caviar 300AB

StorageReview benches the Fujitsu and the Seagate. They bench a
Caviar 400BB and a 600AB, which may differ only in 2x the number of
platters. Best I can come up with. IOMeter scores do indicate a
roughly 50% performance improvement of the SCSI drives over the IDE
drives, "in some bullshit respect" that I'm not entirely confident
about, given that StorageReview has utterly renounced the IOMeter
Workstation benchmark's validity. The Fileserver and Database
benchmarks show a 100% performance improvement. These scores I think
are more reliable and to be expected, but I'm not currently planning
to use my box as a server.

IOMeter - Workstation Access Pattern - Load = Linear *

Western Digital Caviar WD400BB (40.8 GB ATA-100) 88.61 0.69 11.28 ms
0.35 % 253.17
Seagate Cheetah 18XL (18.4 GB Ultra160/m SCSI) 127.60 1.00 7.84 ms
0.47 % 271.49
Western Digital Caviar WD600AB (60.0 GB ATA-100) 78.30 0.61 12.77 ms
0.32 % 244.69
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) 131.03 1.02 7.63 ms 0.45 %
291.18

So, anyways, these StorageReview scores don't help much for trying to
compare single user performance. Maybe the SCSI drives are 50%
faster, maybe they aren't.

I doubt you'd be able to pick it in a double blind
trial with the sort of work you do on your PC.

 

[Cruise Director]

I doubt you'd be able to pick it in a double blind
trial with the sort of work you do on your PC.

Yeah, it's looking like I'd need to do an AI genetic algorithm /
evolutionary database project to make it worth it. I have inklings of
ideas for doing such a project, but many other requirements that are
needed first, so it's sort of the cart before the horse.


Cheers,
Brandon Van Every

 

[Odie Ferrous]

Western Digital Caviar 200BB
Western Digital Caviar 300AB

Dangerous territory - these are very, very unreliable drives. Take them
down to your local rubbish tip.


Odie

 

[Peter]

Yeah, it's looking like I'd need to do an AI genetic algorithm /

evolutionary database project to make it worth it. I have inklings of
ideas for doing such a project, but many other requirements that are
needed first, so it's sort of the cart before the horse.

No need to go that far. I'm not sure which OS you plan to use, but for
Windows, just open three command shells and copy a large file (a GB or
more) to a different one in each shell. Simultaneously. Check "%Idle Time"
in "PhysicalDisk" by performance monitor. Higher numbers indicate better
performance.You can also measure "Disk Write Bytes/sec".

 

[Cruise Director]

Dangerous territory - these are very, very unreliable drives. Take them
down to your local rubbish tip.

Well they've been fine for 4..5 years so I don't see that I have bad
ones.


Cheers,
Brandon Van Every

 

[Arno Wagner]

Well they've been fine for 4..5 years so I don't see that I have bad
ones.

Or you have been very lucky. That luck does not need to keep.

Arno

 

[Folkert Rienstra]

Or you have been very lucky.

That luck does not need to keep.

The luck was in that he got very good ones.
They won't suddenly turn into bad ones after 4..5 years.
They may run out of service life but that has nothing got to do with luck.