Quantum Atlas 10K2 reported as runnning Burst Speed 32mb/s ??
I have a Dual Xeon 3.2ghz 2gb machine with 3 hard disks, running on a
Tyan Tiger i7505 motherboard (all slots are PCI 32bit/33mhz => 133mb/
s)
I have in one of those slots an Adaptec 19160 SCSI controller and have
attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
for my c: drive.
The 2 other disks are 200gb IDE drives connected directly to the Mobo.
Because I am thinking of going to raid, I started looking at the
current performance of my hard disks (18gb isn't enough for my c:
drive, so that got me thinking to buy new Satas for Raid Data and use
my 2 200gbs for raid bootable - leaving the Scsi maybe as a temp place
for pagefile etc).
Anyway I run some tests and notice that my scsi has a burst rate of
just under 32mb/s ??!!
Surely that can't be right, but when I run the test on my ATA133
(which are running ATA100) 200gb hard disks, I'm getting 55mb/s Burst.
I have used 2 different software packages to test the speeds of my
disks and both softwares are giving similar comparisions.....
What could be the trouble here ??
I'm running a very clean machine, no illegal software, nothing
contentious that I can think of, machine is reliable, but I'm sure
that I was getting good results a few years ago with this scsi as it
being used for Video capture and rates looked healthy back then.
Only thing that changed since that time is mobo but why scsi
underperforming IDE ??
Running XP sp2 plus all the windows updates. Have the latest 19160
drivers.
A_Newsreader <a_newsreader@yahoo.com> wrote:
> I have a Dual Xeon 3.2ghz 2gb machine with 3 hard disks, running on a
> Tyan Tiger i7505 motherboard (all slots are PCI 32bit/33mhz => 133mb/
> s)
>
> I have in one of those slots an Adaptec 19160 SCSI controller and have
> attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
> for my c: drive.
>
> The 2 other disks are 200gb IDE drives connected directly to the Mobo.
>
> Because I am thinking of going to raid, I started looking at the
> current performance of my hard disks (18gb isn't enough for my c:
> drive, so that got me thinking to buy new Satas for Raid Data and use
> my 2 200gbs for raid bootable - leaving the Scsi maybe as a temp place
> for pagefile etc).
>
> Anyway I run some tests and notice that my scsi has a burst rate of
> just under 32mb/s ??!!
>
> Surely that can't be right, but when I run the test on my ATA133
> (which are running ATA100) 200gb hard disks, I'm getting 55mb/s Burst.
>
> I have used 2 different software packages to test the speeds of my
> disks and both softwares are giving similar comparisions.....
>
> What could be the trouble here ??
Maybe the trouble is your SCSI drive is circa 1998. Try one made in the
past several years.
Is the Adaptec BIOS showing 40 or 160 MB/s?
If 40, your terminator or SCA adapter is not LVD.
I get burst of 90/100 MB/s on Fujitsu MAM/MAUs.
"A_Newsreader" <a_newsreader@yahoo.com> wrote in message
news:d10e0fc6-24bf-463b-b4dd-cb6d244a3b81@e6g2000prf.googlegroups.com...
>I have a Dual Xeon 3.2ghz 2gb machine with 3 hard disks, running on a
> Tyan Tiger i7505 motherboard (all slots are PCI 32bit/33mhz => 133mb/
> s)
>
> I have in one of those slots an Adaptec 19160 SCSI controller and have
> attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
> for my c: drive.
>
> Anyway I run some tests and notice that my scsi has a burst rate of
> just under 32mb/s ??!!
>
A_Newsreader <a_newsreader@yahoo.com> wrote:
>I have in one of those slots an Adaptec 19160 SCSI controller and have
>attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
[...]
>Anyway I run some tests and notice that my scsi has a burst rate of
>just under 32mb/s ??!!
As someone mentioned it's an OLD disk.
Also to some extent you're looking at the wrong metric. Seek times are
most of the time far more important than burst rate/media rate. Older
SCSI disks tends to have similar or slightly lower STR compared to a
ATA/SATA disk from the same generation (higher spindle speed BUT lower
bit density, because otherwise it couldn't rotate that fast nor seek
that fast!).
This does mean that there's a small class of applications where the
same generation 7200 RPM ATA/SATA disk will be at least as good, but
it's pretty rare, I can only think of one thing right off the bat
(simple video editing, though if work with enough concurrent streams
throughput collapses and seek rate suddenly becomes important).
Doesn't look like it's easy to find original spec sheets for this
drive (they've been bought up) but hothardware had a copy (and they're
not the only one).
So, "Sustained Throughput": 24-40 MB/s (this depends on where it's
measured on the disk).
I'm going to assume you meant 32 MB/s (mega-byte/second), not the 32
mb/s you wrote (milli-bit/second, off by 8,000,000) :-)
This implies that your 32 MB/s could have been somewhere on middle of
the disk, it's also possible that you've connected it via non-LVD
connector and/or cables, the figure is suspiciously similar to the
real world maximum you'll likely to see with SCSI Ultra-Wide...
You can check this in the BIOS and decent OS'es should also tell you
which speed it's connected at somewhere (not sure if it's possible to
find out in Windows without additional tools).
If this is the case you can increase the STR a little bit at one end
of the disk by connecting it "correctly" but it won't have much
practical impact (as opposed to benchmark figures).
Torbjorn Lindgren wrote in news:3lPlj.9740$ee6.3206@fe10.news.easynews.com
> A_Newsreader <a_newsreader@yahoo.com> wrote:
> > I have in one of those slots an Adaptec 19160 SCSI controller and have
> > attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely [...]
> > Anyway I run some tests and notice that my scsi has a burst rate of
> > just under 32mb/s ??!!
> As someone mentioned it's an OLD disk.
Which should not influence U160 *burst* rate.
(Unless it's caching strategy throws a wrench in the way which bench-
marks do measure burst rate (by reading the same blocks over and over)).
>
> Also to some extent you're looking at the wrong metric. Seek times are
> most of the time far more important than burst rate/media rate. Older
> SCSI disks tends to have similar or slightly lower STR compared to a
> ATA/SATA disk from the same generation (higher spindle speed BUT lower
> bit density, because otherwise it couldn't rotate that fast nor seek
> that fast!).
>
> This does mean that there's a small class of applications where the
> same generation 7200 RPM ATA/SATA disk will be at least as good, but
> it's pretty rare, I can only think of one thing right off the bat
> (simple video editing, though if work with enough concurrent streams
> throughput collapses and seek rate suddenly becomes important).
>
> Doesn't look like it's easy to find original spec sheets for this
> drive (they've been bought up) but hothardware had a copy (and they're
> not the only one).
>
> http://www.hothardware.com/articles/...FSCSI%5FDrive/
>
> So, "Sustained Throughput": 24-40 MB/s (this depends on where it's
> measured on the disk).
>
> I'm going to assume you meant 32 MB/s (mega-byte/second), not the 32
> mb/s you wrote (milli-bit/second, off by 8,000,000) :-)
> This implies that your 32 MB/s could have been somewhere on middle of
> the disk,
Or that it just is what it implied: *burst rate*.
> it's also possible that you've connected it via non-LVD
> connector and/or cables, the figure is suspiciously similar to the
> real world maximum you'll likely to see with SCSI Ultra-Wide...
>
> You can check this in the BIOS and decent OS'es should also tell you
> which speed it's connected at somewhere (not sure if it's possible to
> find out in Windows without additional tools).
> If this is the case you can increase the STR a little bit
He didn't mention low STR.
> at one end of the disk by connecting it "correctly" but it won't
> have much practical impact (as opposed to benchmark figures).
Cydrome Leader wrote in news:fn7mpl$kc$2@reader2.panix.com
> A_Newsreader <a_newsreader@yahoo.com> wrote:
> > I have a Dual Xeon 3.2ghz 2gb machine with 3 hard disks, running on a
> > Tyan Tiger i7505 motherboard (all slots are PCI 32bit/33mhz => 133mb/
> > s)
> >
> > I have in one of those slots an Adaptec 19160 SCSI controller and have
> > attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
> > for my c: drive.
> >
> > The 2 other disks are 200gb IDE drives connected directly to the Mobo.
> >
> > Because I am thinking of going to raid, I started looking at the
> > current performance of my hard disks (18gb isn't enough for my c:
> > drive, so that got me thinking to buy new Satas for Raid Data and use
> > my 2 200gbs for raid bootable - leaving the Scsi maybe as a temp place
> > for pagefile etc).
> >
> > Anyway I run some tests and notice that my scsi has a burst rate of
> > just under 32mb/s ??!!
> >
> > Surely that can't be right, but when I run the test on my ATA133
> > (which are running ATA100) 200gb hard disks,
> I'm getting 55mb/s Burst.
Well, that's not very good for ATA100 as well.
> >
> > I have used 2 different software packages to test the speeds of my
> > disks and both softwares are giving similar comparisions.....
> >
> > What could be the trouble here ??
> Maybe the trouble is your SCSI drive is circa 1998.
And what has that got to do with burst rate.
> Try one made in the past several years.
It's an U160 drive on an U160 controller. It should have way higher
burstrate than 32MB/s (unless the drive is caching behind which can influence some benchmarks negatively).
Folkert Rienstra <see_reply-to@myweb.nl> wrote:
> Cydrome Leader wrote in news:fn7mpl$kc$2@reader2.panix.com
>> A_Newsreader <a_newsreader@yahoo.com> wrote:
>> > I have a Dual Xeon 3.2ghz 2gb machine with 3 hard disks, running on a
>> > Tyan Tiger i7505 motherboard (all slots are PCI 32bit/33mhz => 133mb/
>> > s)
>> >
>> > I have in one of those slots an Adaptec 19160 SCSI controller and have
>> > attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
>> > for my c: drive.
>> >
>> > The 2 other disks are 200gb IDE drives connected directly to the Mobo.
>> >
>> > Because I am thinking of going to raid, I started looking at the
>> > current performance of my hard disks (18gb isn't enough for my c:
>> > drive, so that got me thinking to buy new Satas for Raid Data and use
>> > my 2 200gbs for raid bootable - leaving the Scsi maybe as a temp place
>> > for pagefile etc).
>> >
>> > Anyway I run some tests and notice that my scsi has a burst rate of
>> > just under 32mb/s ??!!
>> >
>> > Surely that can't be right, but when I run the test on my ATA133
>> > (which are running ATA100) 200gb hard disks,
>
>> I'm getting 55mb/s Burst.
>
> Well, that's not very good for ATA100 as well.
>
>> >
>> > I have used 2 different software packages to test the speeds of my
>> > disks and both softwares are giving similar comparisions.....
>> >
>> > What could be the trouble here ??
>
>> Maybe the trouble is your SCSI drive is circa 1998.
>
> And what has that got to do with burst rate.
Unless the "2 different software packages to test the speeds of my disks"
are connected to a logic analyzer measuring I/O on the SCSI bus itself,
they're just measuring plain read/write speeds, not how fast commands can
pass from the controller to the disk and back, most of which are
irrelevant to people benchmarking disks in the first place.
A more interesting test would be something like hdtune that could be setup
to read only 1MB of sectors over and over again, or any other size that
the disk itself will cache and just sent back to the controller while
skipping an actual disk read. That's far more likely to saturate a PCI bus
than reading data off the platters of an atlas II or whatever the original
disk is.
>> Try one made in the past several years.
>
> It's an U160 drive on an U160 controller. It should have way higher
> burstrate than 32MB/s (unless the drive is caching behind which can influence some benchmarks negatively).
>
> What's important is what the STR's are.
>
>
>
"Cydrome Leader" <presence@MUNGEpanix.com> wrote in message news:fnakvv$mh5$1@reader2.panix.com...
>> And what has that got to do with burst rate.
>
> Unless the "2 different software packages to test the speeds of my disks"
> are connected to a logic analyzer measuring I/O on the SCSI bus itself,
> they're just measuring plain read/write speeds, not how fast commands can
> pass from the controller to the disk and back, most of which are
> irrelevant to people benchmarking disks in the first place.
>
Nonsense.
> A more interesting test would be something like hdtune that could be setup
> to read only 1MB of sectors over and over again, or any other size that
> the disk itself will cache and just sent back to the controller while
> skipping an actual disk read. That's far more likely to saturate a PCI bus
> than reading data off the platters of an atlas II or whatever the original
> disk is.
>
That's what they do. What the **** did you think "burst" means?
Cydrome Leader wrote in news:fnakvv$mh5$1@reader2.panix.com
> Folkert Rienstra <see_reply-to@myweb.nl> wrote:
> > Cydrome Leader wrote in news:fn7mpl$kc$2@reader2.panix.com
> > > A_Newsreader <a_newsreader@yahoo.com> wrote:
> > > > I have a Dual Xeon 3.2ghz 2gb machine with 3 hard disks, running on a
> > > > Tyan Tiger i7505 motherboard (all slots are PCI 32bit/33mhz => 133mb/s)
> > > >
> > > > I have in one of those slots an Adaptec 19160 SCSI controller and have
> > > > attached to that 1 Quantum Atlas 10K2 18gb disk which I have purely
> > > > for my c: drive.
> > > >
> > > > The 2 other disks are 200gb IDE drives connected directly to the Mobo.
> > > >
> > > > Because I am thinking of going to raid, I started looking at the
> > > > current performance of my hard disks (18gb isn't enough for my c:
> > > > drive, so that got me thinking to buy new Satas for Raid Data and use
> > > > my 2 200gbs for raid bootable - leaving the Scsi maybe as a temp place
> > > > for pagefile etc).
> > > >
> > > > Anyway I run some tests and notice that my scsi has a burst rate of
> > > > just under 32mb/s ??!!
> > > >
> > > > Surely that can't be right, but when I run the test on my ATA133
> > > > (which are running ATA100) 200gb hard disks,
> >
> > > I'm getting 55mb/s Burst.
> >
> > Well, that's not very good for ATA100 as well.
> >
> > > >
> > > > I have used 2 different software packages to test the speeds of my
> > > > disks and both softwares are giving similar comparisions.....
> > > >
> > > > What could be the trouble here ??
> >
> > > Maybe the trouble is your SCSI drive is circa 1998.
> >
> > And what has that got to do with burst rate.
> Unless the "2 different software packages to test the speeds of my disks"
> are connected to a logic analyzer measuring I/O on the SCSI bus itself,
> they're just measuring plain read/write speeds,
Nonsense.
If they say they measure burst rate then that is what they are measuring.
> not how fast commands can pass from the controller to the disk and back,
> most of which are irrelevant to people benchmarking disks in the first place.
And what has that (passing commands from the controller to the disk and back)
got to do with "burst rate". (Oh, and passing commands is actually deadslow).
>
> A more interesting test would be something like hdtune that could be setup
> to read only 1MB of sectors over and over again, or any other size that
> the disk itself will cache and just sent back to the controller while skipping
> an actual disk read.
Wow, you just defined "burst rate". You must be an expert. *Not*.
That's what the "software packages" are doing to determine "burst rate".
And since this is measured for a drive it is the drive burst rate, not the in-
terface burst rate since that's obviously fixed (ie no point in measuring it).
> That's far more likely to saturate a PCI bus
And what has that got to do with "burst rate".
> than reading data off the platters of an atlas II or whatever the original
> disk is.
Not if the drive firmware handles this type of benchmark behaviour badly.
>
>
> > > Try one made in the past several years.
> >
> > It's an U160 drive on an U160 controller. It should have way higher
> > burstrate than 32MB/s (unless the drive is caching behind which can
> > influence some benchmarks negatively).
> >
> > What's important is what the STR's are.
Eric Gisin <gisin@uniserve.com> wrote:
> "Cydrome Leader" <presence@MUNGEpanix.com> wrote in message news:fnakvv$mh5$1@reader2.panix.com...
>>> And what has that got to do with burst rate.
>>
>> Unless the "2 different software packages to test the speeds of my disks"
>> are connected to a logic analyzer measuring I/O on the SCSI bus itself,
>> they're just measuring plain read/write speeds, not how fast commands can
>> pass from the controller to the disk and back, most of which are
>> irrelevant to people benchmarking disks in the first place.
>>
> Nonsense.
>
>> A more interesting test would be something like hdtune that could be setup
>> to read only 1MB of sectors over and over again, or any other size that
>> the disk itself will cache and just sent back to the controller while
>> skipping an actual disk read. That's far more likely to saturate a PCI bus
>> than reading data off the platters of an atlas II or whatever the original
>> disk is.
>>
> That's what they do. What the **** did you think "burst" means?
Quantum Atlas 10K2 reported as runnning Burst Speed 32mb/s ?? 
Linear Mode
