Rahul wrote:
> Paul wrote in news:hma831$c9m$1@speranza.aioe.org:
>
>> something to hide. You could always use the free version of HDTune
>> from hdtune.com and check the seek rate and see what RPM speed that
>> result is consistent with.
>
> I'm not sure what is a "good" result but here's what I got from hdtune:
>
> http://tinyurl.com/yg5ynal
> http://tinyurl.com/y9abw95
> http://tinyurl.com/yc5hevt
> http://tinyurl.com/yg227ju
> http://tinyurl.com/y8eobyz
>
> Do any of the stats. look super slow?
>
I use version 2.55 (free version), so haven't really looked at the Pro version.
The "Benchmark" graph (hdtune5.png) appears to go from 0 to 64MB and measures 165MB/sec.
The box labeled "cache" is checked. That test appears to be testing
the cache on the drive controller, rather than the entire drive. Still,
it proves your interface cable is running faster than SATA I rates. Some
storage devices manage 220MB/sec on cable limited tests at SATA II rates.
Not a big deal.
To check your hdtune4 graph result, I'd have to find some "Atto" benchmarks
of some other drives. That would take a while to find some good ones.
OK, this thread has some Atto examples. Perhaps you can change your settings
or something, and rerun the hdtune4 graph.
http://www.planetamd64.com/index.php?showtopic=21897
Your hdtune2.png graph, another "benchmark" type graph, is a bit bumpy.
Drawing a line through the graph, your range is about 90MB/sec down to
about 45MB/sec. My new Seagate 500GB single platter drive does 130MB/sec
down to about 70MB/sec. Some of the "bumpy" behavior in the graph, is
due to sector sparing, where bad blocks are replaced with local spares.
Another quirk I've noticed, is one bump in the graph appears to occur
as the benchmark test passes over the pagefile, if the drive happens
to be your OS drive.
Your average access time from hdtune2.png, is 16.3ms. My 7200RPM 500GB
drive is 14.1ms, and Seagate drives tend to be a bit of a slouch. So you're
a bit slow there. The drive might not be staying fully at 7200RPM for
the whole benchmark.
In your hdtune1.png picture (a test option I don't have in the free version),
you have one suspicious dot at 400 milliseconds. For that to happen, a sector
would need to be retried a number of times. Hard drives are set up to
do many many retries (which is why ordinary disks don't make good RAID
drives). The RE3 versions of drives, cost more than ordinary drives, and
one of the differences is reduction in the timeout value before the
drive stops trying to read a bad sector (which, as far as I know, is just
a simple firmware change). The timeout could be on the
order of 5 seconds on a RE3 type of drive, and longer on
ordinary drives.
Based on my untrained eye, there isn't that much wrong with the drive.
Personally, when I get a new drive, I expect the hdtune2.png to be
relatively smooth. I find after about six months usage of a drive,
it starts to get bumpy like the curve for your brand new drive.
And at that point, there could be lots of spared out sectors.
Even a new drive has spared sectors, but not quite as many.
If you have enough of those "400ms" delay cases, it wouldn't take
too many of those while you're working, to really ruin the
performance. On my random access test result, I have one dot
at 33ms, which is the highest point in my scatter plot and stands
out a bit from the other dots.
So your results aren't "broken", but they're not particular
appetizing either
This is my Seagate ST3500418AS, 500GB 7200RPM drive. It was
selected based on price, rather than being a stellar performer.
The sequential transfer rate is pretty good, compared to some
of my older drives. Some of the older drives only get about half
the sequential transfer rate.
http://img69.imageshack.us/img69/6982/hdtunebenchmark500gbst3.png
Price is currently $55.
http://www.newegg.com/Product/Product.aspx?Item=N82E16822148395
This is an older Seagate. Average access time is a bit slower.
The reason I'm posting this one, is to show the "zoned" behavior
of the disk. The disk is divided into zones, so the transfer
rate is not intended to be a smooth curve, but rather a stair-step
function. This benchmark isn't too bumpy, but that particular
drive is a lot worse now (which is why it got replaced several
months ago).
http://img62.imageshack.us/img62/6494/hdtunebenchmarkst380011.png
Paul
130 Gigs wasn't enough to accomodate all the accounts that needed
