What is Sisoftware Sandra trying to tell me?

[George Adams]

I'm trying to figure out why my PC (a homebuilt 2.8Ghz P4, 1Gig RAM) is
going so slow these days when it used to be fairly nimble. Even the
nuclear option - reformat and reinstall Windows XP - hasn't really
helped that much.

I suspect the hard drive (Western Digital 200G, 7200RPM, 8M cache). It
seems to thrash an awful lot (even after using Windows' defragging and
error checking tools), and when things get slow or jerky, the hard drive
is always running even if the CPU is idling and there's plenty of RAM
(and virtual RAM) available.

So I fired up Sisoftware Sandra and ran the hard drive benchmark.
Here's what I got:

Drive Index: 52MB/s
Random Access Time: 15ms

Not too bad there. But then there's this:

Speed at position 0% : 50MB/s (96%)
Speed at position 3% : 49MB/s (95%)
Speed at position 6% : 52MB/s (100%)
Speed at position 10% : 51MB/s (99%)
Speed at position 13% : 50MB/s (97%)
Speed at position 16% : 17MB/s (32%)
Speed at position 20% : 50MB/s (97%)
Speed at position 23% : 37MB/s (71%)
Speed at position 26% : 45MB/s (88%)
Speed at position 30% : 48MB/s (93%)
Speed at position 33% : 4441kB/s (8%)
Speed at position 36% : 39MB/s (76%)
Speed at position 40% : 45MB/s (87%)
Speed at position 43% : 46MB/s (89%)
Speed at position 46% : 45MB/s (86%)
Speed at position 50% : 6841kB/s (13%)
Speed at position 53% : 43MB/s (83%)
Speed at position 56% : 44MB/s (85%)
Speed at position 60% : 43MB/s (83%)
Speed at position 63% : 3734kB/s (7%)
Speed at position 66% : 40MB/s (78%)
Speed at position 70% : 40MB/s (77%)
Speed at position 73% : 38MB/s (74%)
Speed at position 76% : 6769kB/s (13%)
Speed at position 80% : 36MB/s (70%)
Speed at position 83% : 5156kB/s (10%)
Speed at position 86% : 11MB/s (22%)
Speed at position 90% : 33MB/s (63%)
Speed at position 93% : 31MB/s (61%)
Speed at position 96% : 30MB/s (57%)
Speed at position 100% : 5421kB/s (10%)
Random Access Time : 15 ms (estimated)
Full Stroke Access Time : 15 ms (estimated)

If you plot that graph, especially noting the values at position
16%,33%,50%,63%,76%,83%, you get a gently downward-sloping graph,
punctuated with some plummeting V-shaped holes.

So what is Sandra is actually TELLING me here? Why do I have sudden
sharp dropoffs in speed every few positions? Is Sandra pointing to a
physical location on my hard drive platter(s) that's physically bad? Is
my drive about to croak?

Should I run some other diagnostic on it? Or is this all normal somehow?

Thanks to anyone who can help!

 

[Arno Wagner]

I'm trying to figure out why my PC (a homebuilt 2.8Ghz P4, 1Gig RAM) is
going so slow these days when it used to be fairly nimble. Even the
nuclear option - reformat and reinstall Windows XP - hasn't really
helped that much.

I suspect the hard drive (Western Digital 200G, 7200RPM, 8M cache). It
seems to thrash an awful lot (even after using Windows' defragging and
error checking tools), and when things get slow or jerky, the hard drive
is always running even if the CPU is idling and there's plenty of RAM
(and virtual RAM) available.

So I fired up Sisoftware Sandra and ran the hard drive benchmark.
Here's what I got:

Drive Index: 52MB/s
Random Access Time: 15ms

Not too bad there. But then there's this:

Speed at position 0% : 50MB/s (96%)
Speed at position 3% : 49MB/s (95%)
Speed at position 6% : 52MB/s (100%)
Speed at position 10% : 51MB/s (99%)
Speed at position 13% : 50MB/s (97%)
Speed at position 16% : 17MB/s (32%)
Speed at position 20% : 50MB/s (97%)
Speed at position 23% : 37MB/s (71%)
Speed at position 26% : 45MB/s (88%)
Speed at position 30% : 48MB/s (93%)
Speed at position 33% : 4441kB/s (8%)
Speed at position 36% : 39MB/s (76%)
Speed at position 40% : 45MB/s (87%)
Speed at position 43% : 46MB/s (89%)
Speed at position 46% : 45MB/s (86%)
Speed at position 50% : 6841kB/s (13%)
Speed at position 53% : 43MB/s (83%)
Speed at position 56% : 44MB/s (85%)
Speed at position 60% : 43MB/s (83%)
Speed at position 63% : 3734kB/s (7%)
Speed at position 66% : 40MB/s (78%)
Speed at position 70% : 40MB/s (77%)
Speed at position 73% : 38MB/s (74%)
Speed at position 76% : 6769kB/s (13%)
Speed at position 80% : 36MB/s (70%)
Speed at position 83% : 5156kB/s (10%)
Speed at position 86% : 11MB/s (22%)
Speed at position 90% : 33MB/s (63%)
Speed at position 93% : 31MB/s (61%)
Speed at position 96% : 30MB/s (57%)
Speed at position 100% : 5421kB/s (10%)
Random Access Time : 15 ms (estimated)
Full Stroke Access Time : 15 ms (estimated)

If you plot that graph, especially noting the values at position
16%,33%,50%,63%,76%,83%, you get a gently downward-sloping graph,
punctuated with some plummeting V-shaped holes.

So what is Sandra is actually TELLING me here? Why do I have sudden
sharp dropoffs in speed every few positions? Is Sandra pointing to a
physical location on my hard drive platter(s) that's physically bad? Is
my drive about to croak?

I think it is areas with marginal or bad sectors.

Should I run some other diagnostic on it? Or is this all normal somehow?

You can run a full SMART selftest and check the SMART attributes.
Personally I would say this is suspicuous enough to replace the
drive now, especially since the problem is all over the disk,
suggesting a systematic problem, rather then a localized surface
problem.

Drives are cheap, data is not.

Arno

 

[Trent W.]

If you plot that graph, especially noting the values at position
16%,33%,50%,63%,76%,83%, you get a gently downward-sloping graph,
punctuated with some plummeting V-shaped holes.

It's possible that those locations are across track boundaries. If that is
the case, the head would need to step to read all the requested data.

 

[Arno Wagner]

It's possible that those locations are across track boundaries. If that is
the case, the head would need to step to read all the requested data.

Hmm. Not track boundaries, but speed zones. Hmmm. In an older disk
this may cause slowdown on reading. The regularity of these steps
suggests this is actually the case. Ok, I take my original assessment
back, the speed data may actually not indicate any disk problem.
It is just that current disks do not do this anymore. Sorry for
my misjudgement.

Arno

 

[Folkert Rienstra]

I think it is areas with marginal or bad sectors.


You can run a full SMART selftest and check the SMART attributes.

Personally I would say this is suspicuous enough to replace the
drive now, especially since the problem is all over the disk,

Yeah, why find out what is causing them, right, babblebot?

suggesting a systematic problem,

Which should tell you something,
But hey, you are babblebot, thinking hurts your brain as it starts to hum.

rather then a localized surface problem.

Drives are cheap, data is not.

So find out what kills it, you babblebot moron.

 

[Folkert Rienstra]

It's possible that those locations are across track boundaries. If that is
the case, the head would need to step to read all the requested data.

Bwahahah.

 

[Folkert Rienstra]

Hmm. Not track boundaries, but speed zones. Hmmm.

You need to take care of that humming in your head babblebot.
And the correct name for that "speedzone" of yours is "notch".

In an older disk this may cause slowdown on reading.

Nonsense.
That's how harddrives and other fixed speed spinning media work.

The regularity of these steps suggests this is actually the case.

Ok, I take my original assessment back, the speed data may actually not indicate
any disk problem.

With speeds as low as 4 MB/s, of course there is, you moronic babblebot.
Just not as simple as to what you think the cause is/was.

It is just that current disks do not do this anymore.

Of course they do, babblebot.

Sorry for my misjudgement.

Story of your life, babblebot.

 

[Jesco Lincke]

Bwahahah.

I am really, really thankful for your constructive contibutions.
At least they make me laugh... wait... it seems you're laughing as
well... - been reading your own posts, have you?

 

[Folkert Rienstra]

I am really, really thankful for your constructive contibutions.

Do I hear some deep seated jealousy here?
Would you like to have this constructive "contibution" applied to your posts as well?

At least they make me laugh... wait... it seems you're laughing as well... -

been reading your own posts, have you?

You mean, do I proofread my messages, unlike the brainfar(c)ted babblebot?
Yes, I do. Thanks for asking.

And oh, uh .... you might try that as well ..... if you're up to it.

 

[George Adams]

Thanks for the help guys. 2 thinks make me still suspicious that it's
something physically bad with the drive (well, 3 if you count the
general slowness and thrashiness of the machine).

- The drive is fairly modern - a 3-year-old Western Digital WD2000JD
(http://tinyurl.com/2onz7a - 200Gig, 7200RPM, 8M Cache, Ultra ATA/100),
and if I understand you guys correctly, a dramatic drop in speed might
be explainable in an older drive, but probably not something this new.
(right?)

- Sisoftware Sandra allows you to see the graphlines of other reference
drives on the same graph as the drive you're testing. All of the drives
show a gentle slope downward from left to right, but none show any hint
of the deep valleys that my drive's graph was showing.

I'll do some SMART testing on the drive (once I figure out how!) but
Folkert, you may be right that the prudent thing to do is just buy a new
drive.

Thanks again.

 

[Folkert Rienstra]

Thanks for the help guys.

What help. The babblebot had a brainfarct, as usual.

2 thinks make me still suspicious that it's something physically bad with the
drive (well, 3 if you count the general slowness and thrashiness of the machine).

- The drive is fairly modern - a 3-year-old Western Digital WD2000JD
(http://tinyurl.com/2onz7a - 200Gig, 7200RPM, 8M Cache, Ultra ATA/100),

and if I understand you guys correctly,

Obviously not.

a dramatic drop in speed might be explainable in an older drive, but probably
not something this new.

Uhuh, and where exactly did I say that.

(right?)
Wrong.


- Sisoftware Sandra allows you to see the graphlines of other reference
drives on the same graph as the drive you're testing. All of the drives
show a gentle slope downward from left to right, but none show any hint
of the deep valleys that my drive's graph was showing.

Now why would that be.

I'll do some SMART testing on the drive (once I figure out how!)

That doesn't change anything. That's just babblebot's favourite stopword.

but Folkert,

you may be right that the prudent thing to do is just buy a new drive.

Nope, that was the babblebot. I actually said the opposite.
Maybe you should respond to the posts in question (like most sen-
sible people do) so your short memory doesn't play tricks on you.

 

[Jesco Lincke]

Do I hear some deep seated jealousy here?
Would you like to have this constructive "contibution" applied to your posts as well?

Oh, please do! I like a good laugh as much as the next person...

Doesn't have to be my posts specifically, though. ANY post will do since
you don't ever seem to touch the OP's content anyway...