🚀 go-pugleaf

On Sun, 30 Nov 1997 13:55:25 GMT, bob@lintilla.demon.co.uk (Bob Cousins)
wrote:

>It's funny how people with SCSI systems often seek to put down EIDE, 
>and at the same time make ridiculous performance claims for SCSI. 

It is funny how _some_ people with SCSI systems do this. It's also funny
how _some_ people with non-SCSI systems claim that performance is
equivalent to SCSI. 

>Maybe its because they have paid twice as much for their disk storage, 
>and feel a need to justify its cost, even if by inventing performance 
>figures?

That is certainly a possible reason. 

>If the Seagate Cheetah can do 14-21 MB/sec, I doubt you are seeing continuous
>40MB/sec. The fact is in a typical workstation, SCSI offers little or no benefit
>to the overall performance of the system. If you don't believe me, see
>http://www8.zdnet.com/pcmag/pclabs/report/r960702a.htm.

If one were to pay twice as much for a SCSI system you certainly 
would not get components that were roughly equivalent to their IDE
counterparts. PC-Mag, however, purposely sought out drives that had
physically similar data transfer rates. It seems clear to me that SCSI
cannot compete with IDE on a cost/performance evaluation, but typically
SCSI purchasers have performance as the primary consideration. 

I also have reservations about their definition of what a typical
workstation user might be doing for multitasking. Two Autocad windows?
That's it? I have a network in my house, and the NT machine is generally
used as the file server and CD-Rom server, as well as my programming
machine. I think they need to re-examine how someone who decides to buy
a SCSI system actually uses it. 

The physical limit for transfer rate of today's HD is about 15mb/sec. No HD
can sustain a rate higher than that.
33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
Do your home work and don't brag your scsi here.
It's a waste of money if not used as multitasking server.
BTW, scsi cards usually are trouble makers.

Zypher wrote in message <01bcfd55$9845f3e0$900189ca@jake>...
>UltraDMA (the latest standard-forget EIDE also known as Ultra ATA or Ultra
>33) can give a 33mb/sec transfer rate...but in real life its about
>20mb/sec-not bad.
>
>If you want real speed-go with Ultrawide AV SCSI or higher like I have
>(true 40mb/sec contionous, used for video capture, servers and the like)
>

Re disk speeds:

    It's all very well you perfomance-hounds arguing about the last
 few mm of performance.  But for comparison, on my completely
 untuned Cyrix P133+ system with a 1.2Gb Seagate EIDE (fogotten
 which model) on PT5 PCI m/board, I noticed that when I was doing
 a disk-check with Norton 8 Disk Doctor this afternoon, I was
 getting a rock-steady 5 Mb/sec until the last tenth of the disk,
 when it slowed down to about 3 Mb/sec (presumably due to geometry
 on the inner tracks).  Doesn't seem too bad to me.  But then I
 don't play movies.

                                        Chris K.
-- 
Chris Kennington (newshound)                        01491 574065
  You can mail me with suitable modifications to header-addresses,
   or to "chris <at> inf-eden.demon.co.uk", replacing the <at>.

jake@binghamton wrote:

> The physical limit for transfer rate of today's HD is about 15mb/sec. No HD
> can sustain a rate higher than that.
> 33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
> Do your home work and don't brag your scsi here.
> It's a waste of money if not used as multitasking server.

    Are we a little hostile?    I don't have a degree in computer science, but
let me share what my impressions about SCSI are.  I will agree that most drives
(EIDE vs. SCSI) are fairly equal in real world performance.  However, I was
under the impression that since SCSI interfaces only use a fraction of the CPU
cycles that IDE interfaces use, SCSI devices are preferable when doing CPU
intensive operations, as the controller handles most of the work without the
CPU's help.  Me, I just like the fact that I can hook up to 32 devices to one
controller.  I'll never have that many, but, you never know.  8^>

> BTW, scsi cards usually are trouble makers.

  I don't know for sure about this, I have yet to see one fail.  Sure, you may
need to tinker, but that's what separates the men from the boys.  8^>  I don't
think this will be as much as a problem anymore as the clone BIOS are now
actually setup to recognize SCSI devices as being legitimate.  Operating
systems (I use the term lightly) like Windows95 and OS/2 have built in SCSI
drivers.  Other real OS's like Mac/OS and AmigaDOS were designed from the
beginning to use SCSI as the 'main' fixed disk interface.  They have very
little (none) trouble in this respect.

jason

On Sun, 30 Nov 97 21:33:12 GMT, Newz@inf-eden.demon.co.ux (Chris
Kennington - newshound) wrote:
> untuned Cyrix P133+ system with a 1.2Gb Seagate EIDE (fogotten
> which model) on PT5 PCI m/board, I noticed that when I was doing
> a disk-check with Norton 8 Disk Doctor this afternoon, I was
> getting a rock-steady 5 Mb/sec until the last tenth of the disk,
> when it slowed down to about 3 Mb/sec (presumably due to geometry
> on the inner tracks).  Doesn't seem too bad to me.  But then I
> don't play movies.

Just a hint, that slow down you experianced is the same as I
experianced when my Seagate/Conner 1.2 was failing.

So far I have had 4 of this size (1.2 gig - All different model
numbers) drives fail - ALL Seagate/Conner drives, all failed the same
way, last sectors slowed down, then had head seek problems (noisy,
slow and other odd problems), then heads either locked up or sounded
like rocks in a tin can.

Now, I have had 6 Seagate/Conner drives fail in the last 12 months, I
admit that I use my drives harder than most, but this failure rate is
unacceptable, in 18 systems.

Considering that QDI, a major motherboard manufacture has a bulletin
on their web site that specifically state DO NOT use SEAGATE Medialist
drives with the Intel TX  Based Motherboards.

reference : http://www.qdi.ca/products/faq_tx.htm

I used to like Seagate hard drives, but due to constant and
undocumented problems, I am no longer using them at all, in any
system.

I am only using Quantum, and Fujitsu in all my future systems.

In article <34824337.CAA79322@prodigy.net>, Jason <jneus@prodigy.net>
writes

<SNIPS>

> Me, I just like the fact that I can hook up to 32 devices to one
>controller.  I'll never have that many, but, you never know.  8^>
>

And I bet once upon a time you never thought you'd need 16Mb RAM or fill
a 500Mb Hard Disk either <BG>
-- 
Tim Stannard -  Ad Hoc Computing, Addlestone, Surrey, UK 
                Tel/Fax: 01932 830572  Mob: 0589 918184
                e-mail: stannard@adhoc.demon.co.uk

see.my.sig.4.addr@no.damn.spam.com wrote:
> 
> On Sun, 30 Nov 1997 13:55:25 GMT, bob@lintilla.demon.co.uk (Bob
> Cousins) wrote:
> 
> >In demon.tech.pc, Zypher wrote:
> >
> >>UltraDMA (the latest standard-forget EIDE also known as Ultra ATA or Ultra
> >>33) can give a 33mb/sec transfer rate...but in real life its about
> >>20mb/sec-not bad.
> >>
> >>If you want real speed-go with Ultrawide AV SCSI or higher like I have
> >>(true 40mb/sec contionous, used for video capture, servers and the like)
> >
> >It's funny how people with SCSI systems often seek to put down EIDE, and at the
> >same time make ridiculous performance claims for SCSI. Maybe its because they
> >have paid twice as much for their disk storage, and feel a need to justify its
> >cost, even if by inventing performance figures?
> >
> >If the Seagate Cheetah can do 14-21 MB/sec, I doubt you are seeing continuous
> >40MB/sec. The fact is in a typical workstation, SCSI offers little or no benefit
> >to the overall performance of the system. If you don't believe me, see
> >http://www8.zdnet.com/pcmag/pclabs/report/r960702a.htm.
> 
> With a single HDD running a single app, sure EIDE is going to be the
> same.  The difference is that SCSI is multithreading.  Where you see a
> big performance boost is while multitasking, especially with multiple
> drives, or a RAID controller.
> If you only have one HDD you really don't care how wide the data path
> is for anything but that cheetah.  Sustained transfer rates (which is
> the only thing that counts on single drive systems) are only around
> 10mb/sec for most drives.

I don't think mutli-threading will make much of a difference with SCSI
on a single disk. Since, the disk seems to be the limiting factor for
sustained transfer not the protocol.  Although, I guess in a situation
where one request is for something that isn't in the disk buffer and a
request from another thread for something that is in the buffer.  SCSI
will allow the request for data in the buffer to occur even while the
disk is seeking and reading the first request into the buffer. (Just a
guess though).  The biggest advantage of SCSI though is definitely with
multiple devices whether that is a hard disk, raid, CDROM, DVD, or any
other SCSI device.  Another thought though... Would the burst tranfers
rate have a larger affect on Virtual memory operations?

Don

In article <3482d5d5.1402938@news.mindspring.com>
           jervin@mindspring.com "Felix1" writes:

"Just a hint, that slow down you experianced is the same as I
"experianced when my Seagate/Conner 1.2 was failing.
"
"So far I have had 4 of this size (1.2 gig - All different model
"numbers) drives fail - ALL Seagate/Conner drives, all failed the same
"way, last sectors slowed down, then had head seek problems (noisy,
"slow and other odd problems), then heads either locked up or sounded
"like rocks in a tin can.

    Ugh!  Mine is also rattling at times.  Memo:  Buy some more
 TR1 tapes ....

    Thanks for the warning.

                                            Chris K.
-- 
Chris Kennington (newshound)                        01491 574065
  You can mail me with suitable modifications to header-addresses,
   or to "chris <at> inf-eden.demon.co.uk", replacing the <at>.

george marengo <gmarengo@primenet.com> wrote in article
<3481ad5c.1747139@news.primenet.com>...

> I also have reservations about their definition of what a typical
> workstation user might be doing for multitasking. Two Autocad windows?
> That's it? I have a network in my house, and the NT machine is generally
> used as the file server and CD-Rom server, as well as my programming
> machine. I think they need to re-examine how someone who decides to buy
> a SCSI system actually uses it. 

Try CD burner in the near future, server for a LAN, 8 hot swappable drive
bays, Canon slide scanner, fast backup device, 2 CD-ROM's need I go on?

I notice that many IDE owners seek to debunk SCSI in favour of (what the
hell it's called, oh yes,) UltraDMA which supports a MAXimum if 4 devices
which are either hard disks or CD-ROM's.  The occasional ZIP drive may
creep in but IDE CD burners are too crap to mention (breathe and it f***s
the data flow).

TB (who HAS used IDE cd-buners and  who has seen a disc mangled for no real
reason)

In article <65ml7r$c22@mtinsc02.worldnet.att.net>,
Massa <someone@somewhere.net> wrote:
>NO drive can pump out 40 MB/s constantly. SCSI drives are a little faster,
>but from 10-20 MB/s is more realistic.
>

[ lossy compression ]

>>that's the reason people go scsi.  SCSI drives can do constant
>>transfer rates of 5, 10, 20 and 40 mb/s...
>>i think this is right.  i'm all for corrections ;)

The SCSI *bus* can do 5/10/20/40 Mb/sec.  Looking at a very fast hard drive,
the seagate cheetah, a 10,000 RPM drive, it's theoretical maximum xfer rate
is like 21 Mb/sec, but I'll bet in reality even 10 Mb/sec is rare ...

-Eric
--
Eric van Bezooijen     eric@activesw.com     http://www.activesw.com/~eric
"But the meaning of life is a mystery, that we don't understand so far.
 And the music of life is a rhapsody if you're happy the way that you are"
   -- "Freudiana"

In article <3481bb21.0@bingnews.binghamton.edu>,  <jake@binghamton> wrote:
>The physical limit for transfer rate of today's HD is about 15mb/sec. No HD
>can sustain a rate higher than that.
>33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
>Do your home work and don't brag your scsi here.
>It's a waste of money if not used as multitasking server.
>BTW, scsi cards usually are trouble makers.

Not true.  It can be much more convenient.  I can have three hard drives,
a scanner, a tape backup, a zip drive, and a CD-ROM drive all hooked up to
the same high performance SCSI bus, without mucking with the parallel port,
serial port, floppy port, or other not-designed-for-high-data-xfer-rates bus.

-Eric
--
Eric van Bezooijen     eric@activesw.com     http://www.activesw.com/~eric
"But the meaning of life is a mystery, that we don't understand so far.
 And the music of life is a rhapsody if you're happy the way that you are"
   -- "Freudiana"

On Sun, 30 Nov 1997 13:55:25 GMT, bob@lintilla.demon.co.uk (Bob
Cousins) wrote:

>In demon.tech.pc, Zypher wrote:
>
>>UltraDMA (the latest standard-forget EIDE also known as Ultra ATA or Ultra
>>33) can give a 33mb/sec transfer rate...but in real life its about
>>20mb/sec-not bad.
>>
>>If you want real speed-go with Ultrawide AV SCSI or higher like I have
>>(true 40mb/sec contionous, used for video capture, servers and the like)
>
>It's funny how people with SCSI systems often seek to put down EIDE, and at the
>same time make ridiculous performance claims for SCSI. Maybe its because they
>have paid twice as much for their disk storage, and feel a need to justify its
>cost, even if by inventing performance figures?
>
>If the Seagate Cheetah can do 14-21 MB/sec, I doubt you are seeing continuous
>40MB/sec. The fact is in a typical workstation, SCSI offers little or no benefit
>to the overall performance of the system. If you don't believe me, see
>http://www8.zdnet.com/pcmag/pclabs/report/r960702a.htm.
>-- 
>Bob Cousins, Software Engineer.
>Home page at http://www.demon.co.uk/sirius-cybernetics/


Hey! I'm a long time SCSI enthusiast, yes I grin when moving drives
between PC, Macs, and SGIs, yes I grin when I have a fully populated
chain (I'm one of those that know how to actually use 7 devices), yes
I grin when my database apps fly, and yes I have recommended many to
buy IDE when their use did not justify the extra cost.

Personally I'm waiting to 7200rpm UDMA

see.my.sig.4.addr@no.damn.spam.com wrote in article
<34819d72.9334338@news.jps.net>...
> On Sun, 30 Nov 1997 13:55:25 GMT, bob@lintilla.demon.co.uk (Bob
> Cousins) wrote:
> 
> >In demon.tech.pc, Zypher wrote:
> >
> >>UltraDMA (the latest standard-forget EIDE also known as Ultra ATA or
Ultra
> >>33) can give a 33mb/sec transfer rate...but in real life its about
> >>20mb/sec-not bad.
> >>
> >>If you want real speed-go with Ultrawide AV SCSI or higher like I have
> >>(true 40mb/sec contionous, used for video capture, servers and the
like)
> >
> >It's funny how people with SCSI systems often seek to put down EIDE, and
at the
> >same time make ridiculous performance claims for SCSI. Maybe its because
they
> >have paid twice as much for their disk storage, and feel a need to
justify its
> >cost, even if by inventing performance figures?
> >
> >If the Seagate Cheetah can do 14-21 MB/sec, I doubt you are seeing
continuous
> >40MB/sec. The fact is in a typical workstation, SCSI offers little or no
benefit
> >to the overall performance of the system. If you don't believe me, see
> >http://www8.zdnet.com/pcmag/pclabs/report/r960702a.htm.
> 
> With a single HDD running a single app, sure EIDE is going to be the
> same.  The difference is that SCSI is multithreading. 

So is busmastering EIDE so long as only one device per cable is active
simultaneously.

Regards,

Ron Reaugh

Zypher <zypher@link.net.id> wrote in article
<01bcfd55$9845f3e0$900189ca@jake>...
> UltraDMA (the latest standard-forget EIDE also known as Ultra ATA or
Ultra
> 33) can give a 33mb/sec transfer rate...but in real life its about
> 20mb/sec-not bad.
> 
> If you want real speed-go with Ultrawide AV SCSI or higher like I have
> (true 40mb/sec contionous, used for video capture, servers and the like)


A single UW SCSI channel will provide at most 35MB/sec and only then in a
RAID 0 setup.  A single drive can do at most 14MB/sec.

Regards,

Ron Reaugh

Jason <jneus@prodigy.net> wrote in article
<34824337.CAA79322@prodigy.net>...
> 
> 
> jake@binghamton wrote:
> 
> > The physical limit for transfer rate of today's HD is about 15mb/sec.
No HD
> > can sustain a rate higher than that.
> > 33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
> > Do your home work and don't brag your scsi here.
> > It's a waste of money if not used as multitasking server.
> 
>     Are we a little hostile?    I don't have a degree in computer
science, but
> let me share what my impressions about SCSI are.  I will agree that most
drives
> (EIDE vs. SCSI) are fairly equal in real world performance.  However, I
was
> under the impression that since SCSI interfaces only use a fraction of
the CPU
> cycles that IDE interfaces use, SCSI devices are preferable when doing
CPU
> intensive operations, as the controller handles most of the work without
the
> CPU's help.

Busmastering EIDE levels that playing field.

Regards,

Ron Reaugh

Eric van Bezooijen <eric@godzilla.berkeley.edu> wrote in article
<65vmkf$rq5@masters0.InterNex.Net>...
> In article <3481bb21.0@bingnews.binghamton.edu>,  <jake@binghamton>
wrote:
> >The physical limit for transfer rate of today's HD is about 15mb/sec. No
HD
> >can sustain a rate higher than that.
> >33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
> >Do your home work and don't brag your scsi here.
> >It's a waste of money if not used as multitasking server.
> >BTW, scsi cards usually are trouble makers.
> 
> Not true.  It can be much more convenient.  I can have three hard drives,
> a scanner, a tape backup, a zip drive, and a CD-ROM drive all hooked up
to
> the same high performance SCSI bus, without mucking with the parallel
port,
> serial port, floppy port, or other not-designed-for-high-data-xfer-rates
bus.

That's great but many scanners and tape drives are not well behaved SCSI
devices and if you multitask they'll kill your HD performance.  Often the
non-HD SCSI gadgets should be isolated on a different SCSI channel for the
high multitasking performance.

Regards,

Ron Reaugh

In article <66171n$ltq@bgtnsc03.worldnet.att.net>,
Ron Reaugh <Ron-Reaugh@worldnet.att.net> wrote:
} 
} Jason <jneus@prodigy.net> wrote in article
} <34824337.CAA79322@prodigy.net>...
} >     Are we a little hostile?    I don't have a degree in computer
} science, but
} > let me share what my impressions about SCSI are.  I will agree that most
} drives
} > (EIDE vs. SCSI) are fairly equal in real world performance.  However, I
} was
} > under the impression that since SCSI interfaces only use a fraction of
} the CPU
} > cycles that IDE interfaces use, SCSI devices are preferable when doing
} CPU
} > intensive operations, as the controller handles most of the work without
} the
} > CPU's help.
} 
} Busmastering EIDE levels that playing field.

They most certainly do not. Busmastering IDE is a bloody joke--
it causes more problems than it solves for a large number of users,
and they don't address most of the other advantages that SCSI has
over IDE.

The newsgroups are filled with stories of problems related to these
damn drivers, and most people see no real-world improvement with them
anyway.

cheers,

-*-
Charles M. Kozierok (mailto:ixl@PCGuide.com)
Webslave, The PC Guide - http://www.PCGuide.com
Comprehensive PC reference, troubleshooting, optimization and procedures...

Charles M. Kozierok <ixl@PCGuide.com> wrote in article
<661a2l$89q$1@usenet11.supernews.com>...
> In article <66171n$ltq@bgtnsc03.worldnet.att.net>,
> Ron Reaugh <Ron-Reaugh@worldnet.att.net> wrote:
> } 
> } Jason <jneus@prodigy.net> wrote in article
> } <34824337.CAA79322@prodigy.net>...
> } >     Are we a little hostile?    I don't have a degree in computer
> } science, but
> } > let me share what my impressions about SCSI are.  I will agree that
most
> } drives
> } > (EIDE vs. SCSI) are fairly equal in real world performance.  However,
I
> } was
> } > under the impression that since SCSI interfaces only use a fraction
of
> } the CPU
> } > cycles that IDE interfaces use, SCSI devices are preferable when
doing
> } CPU
> } > intensive operations, as the controller handles most of the work
without
> } the
> } > CPU's help.
> } 
> } Busmastering EIDE levels that playing field.
> 
> They most certainly do not. Busmastering IDE is a bloody joke--
> it causes more problems than it solves for a large number of users,
> and they don't address most of the other advantages that SCSI has
> over IDE.
> 
> The newsgroups are filled with stories of problems related to these
> damn drivers, and most people see no real-world improvement with them
> anyway.

Some do have problems with busmastering EIDE but otherwise you're just flat
mistaken.

Regards,

Ron Reaugh

On 2 Dec 1997 14:52:49 GMT, "Ron Reaugh" <Ron-Reaugh@worldnet.att.net>
wrote:

>That's great but many scanners and tape drives are not well behaved SCSI
>devices and if you multitask they'll kill your HD performance.  Often the
>non-HD SCSI gadgets should be isolated on a different SCSI channel for the
>high multitasking performance.
>Regards,
>Ron Reaugh

I would like to avoid these devices. Which brands/models of SCSI
scanners and tape drives are the poorly behaved ones? 

In article <661a2l$89q$1@usenet11.supernews.com>, "Charles M. Kozierok"
<ixl@PCGuide.com> writes

[On bus mastering IDE]

>The newsgroups are filled with stories of problems related to these
>damn drivers, and most people see no real-world improvement with them
>anyway.

Most people probably wouldn't see any real-world improvement with SCSI
either. Stop all being silly and realise that scsi does well for some
people but is no significant advantage for most people.

Such is life.

-- 
Oz

In demon.tech.pc, Tom Bird wrote:

>george marengo <gmarengo@primenet.com> wrote in article
><3481ad5c.1747139@news.primenet.com>...
>
>> I also have reservations about their definition of what a typical
>> workstation user might be doing for multitasking. Two Autocad windows?
>> That's it? I have a network in my house, and the NT machine is generally
>> used as the file server and CD-Rom server, as well as my programming
>> machine. I think they need to re-examine how someone who decides to buy
>> a SCSI system actually uses it. 
>
>Try CD burner in the near future, server for a LAN, 8 hot swappable drive
>bays, Canon slide scanner, fast backup device, 2 CD-ROM's need I go on?
>
>I notice that many IDE owners seek to debunk SCSI in favour of (what the
>hell it's called, oh yes,) UltraDMA which supports a MAXimum if 4 devices
>which are either hard disks or CD-ROM's.  

I said for a *typical*  workstation, bozo, not for people hanging half the
world's peripherals off their PC. Your hardware fetish seems to obscure you from
the fact that 99% of PCs have just 1 hard disk and 1 CDROM, for which case SCSI
is a waste of money.

I agree that SCSI is superior for the mid to high end uses, but for the low end
the only advantage is that it makes your wallet lighter. 
-- 
Bob Cousins, Software Engineer.
Home page at http://www.demon.co.uk/sirius-cybernetics/

In demon.tech.pc, Tom Bird wrote:

>jake@binghamton wrote in article <3481bb21.0@bingnews.binghamton.edu>...
>
>> It's a waste of money if not used as multitasking server.
>
>Nope, the SCSI does bring a notable improvement.

Care to back that up with some independently verified figures?

-- 
Bob Cousins, Software Engineer.
Home page at http://www.demon.co.uk/sirius-cybernetics/

In demon.tech.pc, Eric van Bezooijen wrote:

>In article <3481bb21.0@bingnews.binghamton.edu>,  <jake@binghamton> wrote:
>>The physical limit for transfer rate of today's HD is about 15mb/sec. No HD
>>can sustain a rate higher than that.
>>33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
>>Do your home work and don't brag your scsi here.
>>It's a waste of money if not used as multitasking server.
>>BTW, scsi cards usually are trouble makers.
>
>Not true.  It can be much more convenient.  I can have three hard drives,
>a scanner, a tape backup, a zip drive, and a CD-ROM drive all hooked up to
>the same high performance SCSI bus, without mucking with the parallel port,
>serial port, floppy port, or other not-designed-for-high-data-xfer-rates bus.

Great! So for the typical user with one HD and one CDROM, SCSI is unneeded, i.e.
its a waste of money...
-- 
Bob Cousins, Software Engineer.
Home page at http://www.demon.co.uk/sirius-cybernetics/

In article <34886102.16034927@news.demon.co.uk>, bob@lintilla.demon.co.uk 
says...

>>>Nope, the SCSI does bring a notable improvement.
>>
>>Care to back that up with some independently verified figures?


Here are some figures to ponder.

My threadmark results and system configuration:

Data Transfer Rate (MB/sec) was 9.37
Average CPU Utilization was 26.18 percent (using Adaptecs method of CPU 
usage for every Megabyte/second of data moved this is 2.79%. see 
--> http://www.adaptec.com/threadmark/ntcpubnch.html) 
Shuttle 603 mb with 1 meg L2 cache 66mhz bus
Intel P233mmx
128 megs 60ns edo ram 
Diamond Fireport40UW
Seagate ST34501W Cheetah 4.5gb, Umax scanner, Plextor cd-rom, and Panasonic 
cd-r on the scsi bus
Running Windows NT 4ws with sp3 and IE4
Test was run on the second of three 1.42 gb partitions

Have a great day,
Butch Smith

>Great! So for the typical user with one HD and one CDROM, SCSI is unneeded,
i.e.
>its a waste of money...
>--
>Bob Cousins, Software Engineer.
>Home page at http://www.demon.co.uk/sirius-cybernetics/

Typical users don't have only one hard drive anymore.  Sad but true.

Matt Linden
mattlinden@usa.net

>> The newsgroups are filled with stories of problems related to these
>> damn drivers, and most people see no real-world improvement with them
>> anyway.

>Some do have problems with busmastering EIDE but otherwise you're just flat
>mistaken.

He is?  Based on what?

You've repeatedly claimed that workstations do almost entirely
contiguous block file transfers (totally false), and that busmastering
EIDE eliminates the performance advantage of SCSI in a multitasking
multithreading environment (it does not, in my EXPERIENCE, or in the
experience of most others in this NG).

You have never been able to provide anything other than unsupported
repetition of your statments to prove these assertions, so why are you
surprised that people seem to be ganging up on you?

> >>that's the reason people go scsi.  SCSI drives can do constant
> >>transfer rates of 5, 10, 20 and 40 mb/s...
> >>i think this is right.  i'm all for corrections ;)
> 
> The SCSI *bus* can do 5/10/20/40 Mb/sec.  Looking at a very fast hard
drive,
> the seagate cheetah, a 10,000 RPM drive, it's theoretical maximum xfer
rate
> is like 21 Mb/sec, but I'll bet in reality even 10 Mb/sec is rare ...
> 
> -Eric
> --

Actually, even the older Seagate Barracuda 9GB 7,200RPM drive is capable of
a SUSTAINED data transfer of better than 10 MB/s thru about 75% of the
drive, and it's minimum rate is 7.5 MB/s at the core of the drive.  Yes, I
have tested this many times, I have several of these drives that I use in
video editing.  I have not used any of the Cheetah drives for video storage
yet, so I can't speak about their performance, but from the specs they
should be capable of >10 MB/s all the way to the core.


-- 
Scott Cooper
Seehorse Video
seehorse@earthlink.net

: >Some do have problems with busmastering EIDE but otherwise you're just flat
: >mistaken.

I use busmastering ATA on a wide variety of Linux boxes, and have never
seen a problem.  new/big drives sustain 8-11 MB/s with small CPU overhead
(tested with 6.4G drives from Seagate or IBM, Intel-chipset boards,
bonnie or iozone on .4-6G files, ext2, usually 4K blocks, hdparm -c3d1m8)

: You've repeatedly claimed that workstations do almost entirely
: contiguous block file transfers (totally false), and that busmastering

there are indeed many applications where IO is mainly sequential, and Unixen
go to some length to ensure that it is.  clustering is common, blocksizes
larger than 1 sector are common, filesystem prefetching is common, drive
readahead is common.  there certainly are cases where IO is mainly
punctate fetches of 4-8K, but even those are more than a handful of 
sectors.  besides, those cases are often swapping, which can only
sensibly be considered a nonfatal error in a day when ram is $2/MB.

: EIDE eliminates the performance advantage of SCSI in a multitasking
: multithreading environment (it does not, in my EXPERIENCE, or in the
: experience of most others in this NG).

SCSI bigots will always believe it, no matter what.  it's simply not 
true though: I've measured it, why don't you?

in short:

	SCSI has a higher peak bandwidth, though current disks of either
	variety only consume 10-15 MB/s apiece.

	SCSI encompasses more variety and number of devices, though ATAPI
	offers increased variety to ATA buses.

	SCSI disks often have higher MTBF reliability ratings, though it's
	unclear whether the difference between 500K and 1000K hours is
	meaningful or accurate.

	SCSI disks are currently around 2x the price of roughly comparable
	ATA disks. 

	good SCSI controllers start at about $130 whereas good ATA
	controllers are built into most motherboards.

regards, mark hahn.
--
operator may differ from spokesperson.	hahn@neurocog.lrdc.pitt.edu
					http://neurocog.lrdc.pitt.edu/~hahn/

Michael Wong <nospam_mikwon@ebtech.net> wrote in article
<3484e405.83667086@news.ebtech.net>...
> >> The newsgroups are filled with stories of problems related to these
> >> damn drivers, and most people see no real-world improvement with them
> >> anyway.
> 
> >Some do have problems with busmastering EIDE but otherwise you're just
flat
> >mistaken.
> 
> He is?  Based on what?
> 
> You've repeatedly claimed that workstations do almost entirely
> contiguous block file transfers (totally false), and that busmastering
> EIDE eliminates the performance advantage of SCSI in a multitasking
> multithreading environment (it does not, in my EXPERIENCE, or in the
> experience of most others in this NG).

Are you beginning to catch on about how big your set of "most others in
this NG"
is?

Regards,

Ron Reaugh

In article <3482d5d5.1402938@news.mindspring.com>
           jervin@mindspring.com "Felix1" writes:

"So far I have had 4 of this size (1.2 gig - All different model
"numbers) drives fail - ALL Seagate/Conner drives, all failed the same
"way, last sectors slowed down, then had head seek problems (noisy,
"slow and other odd problems), then heads either locked up or sounded
"like rocks in a tin can.
"
"Now, I have had 6 Seagate/Conner drives fail in the last 12 months, I
"admit that I use my drives harder than most, but this failure rate is
"unacceptable, in 18 systems.

    Fwiw, I've had the following comment from my f.n.d.guru :-

Seagate have been having quality problems, but these have been batch
problems associated with certain models and particular months of
manufacture. If the problem is more widepread than that than maybe we
shuold be worried.

                                        Chris K.
-- 
Chris Kennington (newshound)                        01491 574065
  You can mail me with suitable modifications to header-addresses,
   or to "chris <at> inf-eden.demon.co.uk", replacing the <at>.

Bob Cousins <bob@lintilla.demon.co.uk> wrote in article
<34875e7e.15390149@news.demon.co.uk>...
> In demon.tech.pc, Tom Bird wrote:

> I said for a *typical*  workstation, bozo, not for people hanging half
the
                                       ^^^^ 
My, my aren't we being original tonight - one syllable.

> world's peripherals off their PC. Your hardware fetish seems to obscure
you from
> the fact that 99% of PCs have just 1 hard disk and 1 CDROM, for which
case SCSI
> is a waste of money.
> 
> I agree that SCSI is superior for the mid to high end uses, but for the
low end
> the only advantage is that it makes your wallet lighter. 

Wears out your pockets less :)

TB

Ron Reaugh wrote:
> 
> Jason <jneus@prodigy.net> wrote in article
> <34824337.CAA79322@prodigy.net>...
> >
> >
> > jake@binghamton wrote:
> >
> > > The physical limit for transfer rate of today's HD is about
> > > 15mb/sec. No HD can sustain a rate higher than that.
> > > 33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
> > > Do your home work and don't brag your scsi here.
> > > It's a waste of money if not used as multitasking server.

Actually Jake, the minute you start multitasking across more than 2
devices it isn't a waste of money.



> > Are we a little hostile?    I don't have a degree in computer
> > science, but let me share what my impressions about SCSI are.  I
> > will agree that most drives (EIDE vs. SCSI) are fairly equal in
> > real world performance.  However, I was under the impression that
> > since SCSI interfaces only use a fraction of the CPU cycles that
> > IDE interfaces use, SCSI devices are preferable when doing
> > CPU intensive operations, as the controller handles most of the work
> > without the CPU's help.


Excellent synopsis of the beauty of Busmastering controllers of
any type :-)

 
> Busmastering EIDE levels that playing field.

Ron,

You keep making these claims about how great EIDE is.  Would you
mind pointing us towards some benchmarks which validate your
claims?  As it is, you're statements are just words on the screen
with no backup to establish validity to them.

> 
> Regards,
> 
> Ron Reaugh

-- 
I've munged my address to help foil those
insiduous spammer-type persons.  Remove the
_s to get my real address.

jkotches@_pobox_.com

Why? If you are a performance user, you should never buy many Hd at the same
time, you can always get a bigger better faster one cheaper than many. By
the time you buy your 3rd hd for space, your previous two should be thrown
out because they will only degrade your system performance.
Matt Linden wrote in message <667dtn$l0t$1@news3.microserve.net>...
>>Great! So for the typical user with one HD and one CDROM, SCSI is
unneeded,
>i.e.
>>its a waste of money...
>>--
>>Bob Cousins, Software Engineer.
>>Home page at http://www.demon.co.uk/sirius-cybernetics/
>
>Typical users don't have only one hard drive anymore.  Sad but true.
>
>Matt Linden
>mattlinden@usa.net
>
>
>

Multitasking means there's a signigicant portion of operation swapping
between mutiple devices. Multitasking across 2 or 3 devices once in a while
is not considered multitasking in this sence, thus can't justify the extrs
investment in SCSI, both controller and hard drives.
Ok, I saw there were quite a few replied to my post in this issue, I'll try
to make my points clear once for good.

1. The original question was "Is it possible to get 16MB/sec", the answer is
no, no Hard drive can sustain that. Neither EIDE nor SCSI.
2. My previous poster claims he can " have real 40MB/sec " though his scsi
card. That claim is ignorant bragging.
3. Scsi is slightly better in performance, but agian, the performance lead
doesn't justify the price.
4. UDMA brings down the cpu load close to Scsi level.
5. SCSI's are usually not stable running @ 83MHz bus. Those claim their
UW2940 stable could have a better than average card, or more likely, their
systems are running asynchronize PCI bus, i.e. their pci stays at 33MHz, not
41.5. I won't even risk oc SCSI @ 83MHz bus, cuz the reason to choose scsi
is for reliability and important work in the first place.
6. SCSI has been losing ground to IDE for at least five years, and will
continue if they can't be cost effective.
7. For most home users which have 1 or 2 hd, 1 cd, no cdr, no scanner, no
jaz etc, there is ABSOLUTELY no reason to think about SCSI for now. But if
you already have a scsi card, consider scsi cd-rom instead of hard drive,
cuz fast IDE cd-roms are really cpu sucker.
8. My experience? I am usnig IBM Deskstar3, Quantum 6.4 STA , Seagate 9GB UW
Scsi drive w/Adaptec 2940 UW card. If I am going to buy another hd for home
pc for now, IBM udma for sure.

John Kotches wrote in message <34899966.B65AB640@pobox.com>...
>Ron Reaugh wrote:
>>
>> Jason <jneus@prodigy.net> wrote in article
>> <34824337.CAA79322@prodigy.net>...
>> >
>> >
>> > jake@binghamton wrote:
>> >
>> > > The physical limit for transfer rate of today's HD is about
>> > > 15mb/sec. No HD can sustain a rate higher than that.
>> > > 33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
>> > > Do your home work and don't brag your scsi here.
>> > > It's a waste of money if not used as multitasking server.
>
>Actually Jake, the minute you start multitasking across more than 2
>devices it isn't a waste of money.
>
>
>
>> > Are we a little hostile?    I don't have a degree in computer
>> > science, but let me share what my impressions about SCSI are.  I
>> > will agree that most drives (EIDE vs. SCSI) are fairly equal in
>> > real world performance.  However, I was under the impression that
>> > since SCSI interfaces only use a fraction of the CPU cycles that
>> > IDE interfaces use, SCSI devices are preferable when doing
>> > CPU intensive operations, as the controller handles most of the work
>> > without the CPU's help.
>
>
>Excellent synopsis of the beauty of Busmastering controllers of
>any type :-)
>
>
>> Busmastering EIDE levels that playing field.
>
>Ron,
>
>You keep making these claims about how great EIDE is.  Would you
>mind pointing us towards some benchmarks which validate your
>claims?  As it is, you're statements are just words on the screen
>with no backup to establish validity to them.
>
>>
>> Regards,
>>
>> Ron Reaugh
>
>--
>I've munged my address to help foil those
>insiduous spammer-type persons.  Remove the
>_s to get my real address.
>
>jkotches@_pobox_.com

In demon.tech.pc, Matt Linden wrote:

>>Great! So for the typical user with one HD and one CDROM, SCSI is unneeded,
>i.e.
>>its a waste of money...
>
>Typical users don't have only one hard drive anymore.  Sad but true.

Proof?

At work we have around 1000 PCs. Every one apart from 4 servers has 1 HD.
-- 
Bob Cousins, Software Engineer.
Home page at http://www.demon.co.uk/sirius-cybernetics/

John Kotches <jkotches@pobox.com> wrote in article
<34899966.B65AB640@pobox.com>...
> Ron Reaugh wrote:
> > 
> > Jason <jneus@prodigy.net> wrote in article
> > <34824337.CAA79322@prodigy.net>...
> > >
> > >
> > > jake@binghamton wrote:
> > >
> > > > The physical limit for transfer rate of today's HD is about
> > > > 15mb/sec. No HD can sustain a rate higher than that.
> > > > 33MB/sec of UDMA and 40MB/sec of UWSCSI are theoretical limits.
> > > > Do your home work and don't brag your scsi here.
> > > > It's a waste of money if not used as multitasking server.
> 
> Actually Jake, the minute you start multitasking across more than 2
> devices it isn't a waste of money.
> 
> 
> 
> > > Are we a little hostile?    I don't have a degree in computer
> > > science, but let me share what my impressions about SCSI are.  I
> > > will agree that most drives (EIDE vs. SCSI) are fairly equal in
> > > real world performance.  However, I was under the impression that
> > > since SCSI interfaces only use a fraction of the CPU cycles that
> > > IDE interfaces use, SCSI devices are preferable when doing
> > > CPU intensive operations, as the controller handles most of the work
> > > without the CPU's help.
> 
> 
> Excellent synopsis of the beauty of Busmastering controllers of
> any type :-)
> 
>  
> > Busmastering EIDE levels that playing field.
> 
> Ron,
> 
> You keep making these claims about how great EIDE is.  Would you
> mind pointing us towards some benchmarks which validate your
> claims?  As it is, you're statements are just words on the screen
> with no backup to establish validity to them.

See:  http://www.geocities.com/SiliconValley/Peaks/9892/nt4.htm


> > 
> > Regards,
> > 
> > Ron Reaugh
> 
> -- 
> I've munged my address to help foil those
> insiduous spammer-type persons.  Remove the
> _s to get my real address.
> 
> jkotches@_pobox_.com
> 

jake@binghamton wrote in article <3489f02a.0@bingnews.binghamton.edu>...
> Multitasking means there's a signigicant portion of operation swapping
> between mutiple devices. Multitasking across 2 or 3 devices once in a
while
> is not considered multitasking in this sence, thus can't justify the
extrs
> investment in SCSI, both controller and hard drives.
> Ok, I saw there were quite a few replied to my post in this issue, I'll
try
> to make my points clear once for good.
> 
> 1. The original question was "Is it possible to get 16MB/sec", the answer
is
> no, no Hard drive can sustain that. Neither EIDE nor SCSI.

Actually the Ultrastar 9ZX has started shipping and it sustains about
20MB/sec.

Regards,

Ron Reaugh

In article <3489f019.3886808@news.demon.co.uk>, Bob Cousins
<bob@lintilla.demon.co.uk> writes

>Proof?
>
>At work we have around 1000 PCs. Every one apart from 4 servers has 1 HD.

Ah, there are replies and there are beautiful replies. 
This is one of the latter.

-- 
Oz

In demon.tech.pc, Butch Smith wrote:

>In article <34886102.16034927@news.demon.co.uk>, bob@lintilla.demon.co.uk 
>says...
>
>>>>Nope, the SCSI does bring a notable improvement.
>>>
>>>Care to back that up with some independently verified figures?
>
>
>Here are some figures to ponder.
>
>My threadmark results and system configuration:
>
>Data Transfer Rate (MB/sec) was 9.37
>Average CPU Utilization was 26.18 percent (using Adaptecs method of CPU 
>usage for every Megabyte/second of data moved this is 2.79%. see 
>--> http://www.adaptec.com/threadmark/ntcpubnch.html) 
>Shuttle 603 mb with 1 meg L2 cache 66mhz bus
>Intel P233mmx
>128 megs 60ns edo ram 
>Diamond Fireport40UW
>Seagate ST34501W Cheetah 4.5gb, Umax scanner, Plextor cd-rom, and Panasonic 
>cd-r on the scsi bus
>Running Windows NT 4ws with sp3 and IE4
>Test was run on the second of three 1.42 gb partitions

Thanks for providing some hard figures, though your system is towards the top
end rather than typical. For comparative purposes its not very useful as there
are no 10,000 RPM EIDE drives (marketing reasons, rather than technical?). Any
performance benefit you see will be almost entirely due to spin speed rather
than a better interface.

Threadmark results don't mean much when running applications, try the same thing
doing an Access query for example.

However, your figures are interesting in themselves. On paper the ST34501 should
achieve 11 to 17 MB/sec, average say 14 MB/sec. So you are getting 70% of the
expected rate, which probably is quite good depending on transfer size. 

In practice you can only get the paper transfer rates if you use large block
sizes with sequential reads. In typical applications doing mixed read/write
random access, small transfers, the max transfer rate disappears quickly. The
limiting factors are seek time and latency.

Regards,
-- 
Bob Cousins, Software Engineer.
Home page at http://www.demon.co.uk/sirius-cybernetics/

Ron Reaugh wrote:
 
> > Ron,
> >
> > You keep making these claims about how great EIDE is.  Would you
> > mind pointing us towards some benchmarks which validate your
> > claims?  As it is, you're statements are just words on the screen
> > with no backup to establish validity to them.
> 
> See:  http://www.geocities.com/SiliconValley/Peaks/9892/nt4.htm
 
Ok, I went there.  It talks about 2 EIDE drives in a RAID 0
configuration.  Quite frankly, the results were unimpressive.
Besides that, This discussion is about single drive performance
not RAID 0.

Then I see that CPU utilizations to get these "marvelous" transfer
rates (8 and I just shake my head.....

37.96% to achieve 8.09 MB/sec.  That means almost 2 out of
every 5 cycles are dedicated to disk I/O.

By the time you add in the 1-2 cycles for NT to do its housekeeping
you aren't left with a lot of CPU horsepower to do any actual work.

Nice try.


-- 
I've munged my address to help foil those
insiduous spammer-type persons.  Remove the
_s to get my real address.

jkotches@_pobox_.com

John Kotches <jkotches@pobox.com> wrote in article
<348AD346.D2E0CB00@pobox.com>...
> Ron Reaugh wrote:
>  
> > > Ron,
> > >
> > > You keep making these claims about how great EIDE is.  Would you
> > > mind pointing us towards some benchmarks which validate your
> > > claims?  As it is, you're statements are just words on the screen
> > > with no backup to establish validity to them.
> > 
> > See:  http://www.geocities.com/SiliconValley/Peaks/9892/nt4.htm
>  
> Ok, I went there.  It talks about 2 EIDE drives in a RAID 0
> configuration.  Quite frankly, the results were unimpressive.
> Besides that, This discussion is about single drive performance
> not RAID 0.
> 
> Then I see that CPU utilizations to get these "marvelous" transfer
> rates (8 and I just shake my head.....
> 
> 37.96% to achieve 8.09 MB/sec.  That means almost 2 out of
> every 5 cycles are dedicated to disk I/O.
> 
> By the time you add in the 1-2 cycles for NT to do its housekeeping
> you aren't left with a lot of CPU horsepower to do any actual work.
> 
> Nice try.


Go read it again.  There are single drive results as well as RAID 0 results
or did you actually read the page.  What kind of CPU utilizations do you
think busmastering SCSI gets?  If you'll read Adaptec's Threadmark
documentation then you'll find that all CPU utilization figures are
meaningful as % per MB/sec or didn't you know that.  The results at
http://www.geocities.com/SiliconValley/Peaks/9892/nt4.htm
demonstrate the busmastering EIDE provides similar low CPU utilization to
SCSI.

Regards,

Ron Reaugh

Bob Cousins <bob@lintilla.demon.co.uk> wrote in article
<348abb65.10457046@news.demon.co.uk>...
> In demon.tech.pc, Butch Smith wrote:
> 
> >In article <34886102.16034927@news.demon.co.uk>,
bob@lintilla.demon.co.uk 
> >says...
> >
> >>>>Nope, the SCSI does bring a notable improvement.
> >>>
> >>>Care to back that up with some independently verified figures?
> >
> >
> >Here are some figures to ponder.
> >
> >My threadmark results and system configuration:
> >
> >Data Transfer Rate (MB/sec) was 9.37
> >Average CPU Utilization was 26.18 percent (using Adaptecs method of CPU 
> >usage for every Megabyte/second of data moved this is 2.79%. see 
> >--> http://www.adaptec.com/threadmark/ntcpubnch.html) 
> >Shuttle 603 mb with 1 meg L2 cache 66mhz bus
> >Intel P233mmx
> >128 megs 60ns edo ram 
> >Diamond Fireport40UW
> >Seagate ST34501W Cheetah 4.5gb, Umax scanner, Plextor cd-rom, and
Panasonic 
> >cd-r on the scsi bus
> >Running Windows NT 4ws with sp3 and IE4
> >Test was run on the second of three 1.42 gb partitions
> 
> Thanks for providing some hard figures, though your system is towards the
top
> end rather than typical. For comparative purposes its not very useful as
there
> are no 10,000 RPM EIDE drives (marketing reasons, rather than
technical?). Any
> performance benefit you see will be almost entirely due to spin speed
rather
> than a better interface.
> 
> Threadmark results don't mean much when running applications, try the
same thing
> doing an Access query for example.
> 
> However, your figures are interesting in themselves. On paper the ST34501
should
> achieve 11 to 17 MB/sec, average say 14 MB/sec. So you are getting 70% of
the


The sustained transfer rate of the Cheetah on the OD is 14MB/sec. 
Threadmark incorrectly attaches units of MB/sec to it's results; 
Threadmark's units are some kind of weighted average and are not comparable
to mfg's spec's sustained transfer rate figures.  Treat Threadmark results
as scalars that are comparable to other Threadmark figures and nothing
else.  The 9.37 figure for a Cheetah is the highest you'll find for any
single HD at least until in the next couple of days someone posts the
Ultrastar 9ZX figures.


> expected rate, which probably is quite good depending on transfer size. 
> 
> In practice you can only get the paper transfer rates if you use large
block
> sizes with sequential reads. In typical applications doing mixed
read/write
> random access, small transfers, the max transfer rate disappears quickly.
The
> limiting factors are seek time and latency.


The Cheetah does have superior access time to all other drives. 
Workstation performance is dominated by sustained transfer rate and not
access time.

Regards,

Ron Reaugh

Thanks for the news. And I know a technology breakthrough is at the moment.
But sustain rate boost from 10MB/s to 20MB/s still is not significant. I was
always thinking if people really have important application which needs
super fast speed, why not just build a ram pool? e.g. a nice mother board
supports 512 MB is very common, if you can load 256 MB as ramdrive for your
app, imagine how fast it will be. 256MB costs about $500 today. It's just
day dreaming, I don't even bother to check how win95 handles ramdrive.

Ron Reaugh wrote in message <66dmjf$oj3@mtinsc02.worldnet.att.net>...
>
>
>jake@binghamton wrote in article <3489f02a.0@bingnews.binghamton.edu>...
>> Multitasking means there's a signigicant portion of operation swapping
>> between mutiple devices. Multitasking across 2 or 3 devices once in a
>while
>> is not considered multitasking in this sence, thus can't justify the
>extrs
>> investment in SCSI, both controller and hard drives.
>> Ok, I saw there were quite a few replied to my post in this issue, I'll
>try
>> to make my points clear once for good.
>>
>> 1. The original question was "Is it possible to get 16MB/sec", the answer
>is
>> no, no Hard drive can sustain that. Neither EIDE nor SCSI.
>
>Actually the Ultrastar 9ZX has started shipping and it sustains about
>20MB/sec.
>
>Regards,
>
>Ron Reaugh
>

In demon.tech.pc, Ron Reaugh wrote:

>Workstation performance is dominated by sustained transfer rate and not
>access time.

I think that's completely untrue. Only in very few situations do you get
conditions where sustained transfer rate would be dominant. i..e where you are
doing sequential accesses with large transfer sizes. Typical workstation access
is random access with relatively small transfers. It only takes a few seeks to
some other part of the disk to screw sustained rate.

Your theory is contrary to my own work on disk performance and all the
"established" thinking I have ever seen. So unless you have a convincing
argument or some hard figures, your theory sounds like an idle assertion.
-- 
Bob Cousins, Software Engineer.
Home page at http://www.demon.co.uk/sirius-cybernetics/

In article <66dmbn$oj3@mtinsc02.worldnet.att.net>,
Ron Reaugh <Ron-Reaugh@worldnet.att.net> wrote:
>
>
>John Kotches <jkotches@pobox.com> wrote in article
><34899966.B65AB640@pobox.com>...
>> Ron Reaugh wrote:
>> > 
>> > Jason <jneus@prodigy.net> wrote in article
>> > <34824337.CAA79322@prodigy.net>...

[ lossy compression ]

>> You keep making these claims about how great EIDE is.  Would you
>> mind pointing us towards some benchmarks which validate your
>> claims?  As it is, you're statements are just words on the screen
>> with no backup to establish validity to them.
>
>See:  http://www.geocities.com/SiliconValley/Peaks/9892/nt4.htm

Hmm, I read that, and my interpretation of those benchmarks is that
IDE drives use twice as much of your CPU during a disk operation.  I'm
glad I have SCSI in my system!

-Eric
--
Eric van Bezooijen     eric@activesw.com     http://www.activesw.com/~eric
"But the meaning of life is a mystery, that we don't understand so far.
 And the music of life is a rhapsody if you're happy the way that you are"
   -- "Freudiana"

Bob Cousins <bob@lintilla.demon.co.uk> wrote in article
<348c422b.31666353@news.demon.co.uk>...
> In demon.tech.pc, Ron Reaugh wrote:
> 
> >Workstation performance is dominated by sustained transfer rate and not
> >access time.
> 
> I think that's completely untrue. Only in very few situations do you get
> conditions where sustained transfer rate would be dominant. i..e where
you are
> doing sequential accesses with large transfer sizes. Typical workstation
access
> is random access with relatively small transfers. It only takes a few
seeks to
> some other part of the disk to screw sustained rate.
> 
> Your theory is contrary to my own work on disk performance and all the
> "established" thinking I have ever seen. So unless you have a convincing
> argument or some hard figures, your theory sounds like an idle assertion.

No, it's true.  Remember all that's required for STR to dominate is that
the average transfer size to be 100K-200K which is about the size of an
average W95 DLL and smaller than the average W95 program.

> -- 
> Bob Cousins, Software Engineer.
> Home page at http://www.demon.co.uk/sirius-cybernetics/
> 

Eric van Bezooijen <eric@godzilla.berkeley.edu> wrote in article
<66hsvc$i7p@masters0.InterNex.Net>...
> In article <66dmbn$oj3@mtinsc02.worldnet.att.net>,
> Ron Reaugh <Ron-Reaugh@worldnet.att.net> wrote:
> >
> >
> >John Kotches <jkotches@pobox.com> wrote in article
> ><34899966.B65AB640@pobox.com>...
> >> Ron Reaugh wrote:
> >> > 
> >> > Jason <jneus@prodigy.net> wrote in article
> >> > <34824337.CAA79322@prodigy.net>...
> 
> [ lossy compression ]
> 
> >> You keep making these claims about how great EIDE is.  Would you
> >> mind pointing us towards some benchmarks which validate your
> >> claims?  As it is, you're statements are just words on the screen
> >> with no backup to establish validity to them.
> >
> >See:  http://www.geocities.com/SiliconValley/Peaks/9892/nt4.htm
> 
> Hmm, I read that, and my interpretation of those benchmarks is that
> IDE drives use twice as much of your CPU during a disk operation.  I'm
> glad I have SCSI in my system!

Go read it again.  The CPU utilization for EIDE was low about 4% MB/sec.
which is equivalent to what SCSI would do on an >equivalent system<.  CPU
utilization is higher on a slow CPU than on a fast CPU for either SCSI or
EIDE.

Regards,

Ron Reaugh

>> 1. The original question was "Is it possible to get 16MB/sec", the answer
>is
>> no, no Hard drive can sustain that. Neither EIDE nor SCSI.

>Actually the Ultrastar 9ZX has started shipping and it sustains about
>20MB/sec.

Not under real-world conditions.  You're talking about
contiguous-block transfers on a nearly empty hard drive.  In the real
world, drives are partially full and fragmented.

>Thanks for the news. And I know a technology breakthrough is at the moment.
>But sustain rate boost from 10MB/s to 20MB/s still is not significant. I was
>always thinking if people really have important application which needs
>super fast speed, why not just build a ram pool? e.g. a nice mother board
>supports 512 MB is very common, if you can load 256 MB as ramdrive for your
>app, imagine how fast it will be. 256MB costs about $500 today. It's just
>day dreaming, I don't even bother to check how win95 handles ramdrive.

There wouldn't really be any point to that.  Win95 uses most of the
available RAM as a file cache anyway, so it would serve the same
function as your ramdrive.

Michael Wong <nospam_mikwon@ebtech.net> wrote in article
<348f680a.259410342@news.ebtech.net>...
> >> 1. The original question was "Is it possible to get 16MB/sec", the
answer
> >is
> >> no, no Hard drive can sustain that. Neither EIDE nor SCSI.
> 
> >Actually the Ultrastar 9ZX has started shipping and it sustains about
> >20MB/sec.
> 
> Not under real-world conditions.  You're talking about
> contiguous-block transfers on a nearly empty hard drive.  In the real
> world, drives are partially full and fragmented.

Contiguous blocks, yes.  Empty/full has nothing to do with transfer rate
except the drives transfer files located near the OD faster than those near
the ID.  Most workstation hard drives are not very fragmented.

Regards,

Ron Reaugh