Sign in with
Sign up | Sign in
Your question

IBM 75GXP weird performance - write cahce problem?

Tags:
  • Performance
  • IBM
  • Storage
Last response: in Storage
Share
Anonymous
a b G Storage
December 8, 2004 7:07:15 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Hello all,
I noticed one type of operation where the performance
is very low, it occurs when reading and writing lots of data
simultaneously.
For example, I had a RAR archive of a 235MB avi file.
Extracting the archive from and to this drive took
1 minute 46.
I copied the RAR files to a second drive, a Maxtor 53073h4
and repeated the test, it took 60 seconds.

I used the IBM feature tool utility and disabled the
write cache. Repeating the test above in this
condition, it still took 1 minute 46.

Write cache non-operative?


Dave

More about : ibm 75gxp weird performance write cahce problem

Anonymous
a b G Storage
December 9, 2004 12:27:58 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"davexnet02" <davexnetzerotwo@hooya!.com> wrote in message news:v0vcr0debdd8cst84bsneoset51he1evd2@4ax.com
> Hello all,
> I noticed one type of operation where the performance
> is very low, it occurs when reading and writing lots of data
> simultaneously.
> For example, I had a RAR archive of a 235MB avi file.
> Extracting the archive from and to this drive took
> 1 minute 46.
> I copied the RAR files to a second drive, a Maxtor 53073h4
> and repeated the test, it took 60 seconds.
>
> I used the IBM feature tool utility and disabled the
> write cache. Repeating the test above in this
> condition, it still took 1 minute 46.
>
> Write cache non-operative?

2MB cache, 40MB/s transfer rate, or write cache filled in 1/40 second.
Gee, on a 1 minute 46 sec operation how much difference do you think that
will make?

>
>
> Dave
Anonymous
a b G Storage
December 9, 2004 6:09:10 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

On Wed, 8 Dec 2004 21:27:58 +0100, "Folkert Rienstra"
<see_reply-to@myweb.nl> wrote:

>"davexnet02" <davexnetzerotwo@hooya!.com> wrote in message news:v0vcr0debdd8cst84bsneoset51he1evd2@4ax.com
>> Hello all,
>> I noticed one type of operation where the performance
>> is very low, it occurs when reading and writing lots of data
>> simultaneously.
>> For example, I had a RAR archive of a 235MB avi file.
>> Extracting the archive from and to this drive took
>> 1 minute 46.
>> I copied the RAR files to a second drive, a Maxtor 53073h4
>> and repeated the test, it took 60 seconds.
>>
>> I used the IBM feature tool utility and disabled the
>> write cache. Repeating the test above in this
>> condition, it still took 1 minute 46.
>>
>> Write cache non-operative?
>
>2MB cache, 40MB/s transfer rate, or write cache filled in 1/40 second.
>Gee, on a 1 minute 46 sec operation how much difference do you think that
>will make?
>
Obviously I had no idea. If it really works as you describe,
I can see not much at all.
Since most HD accesses in a typical desktop scenario are pretty
small reads and writes, then what's the point of the cache at all?

However, if some of the problem is to do with timing, and the
cache having an effect of smoothing out the operation,
the benefit could be greater than the sum of the parts.
OR so I thought.


Dave
Anonymous
a b G Storage
December 9, 2004 8:29:56 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"davexnet02" <davexnetzerotwo@hooya!.com> wrote in message news:q3gfr0trsb24lmb96sdmcm7ku6oguj3f4b@4ax.com
> On Wed, 8 Dec 2004 21:27:58 +0100, "Folkert Rienstra" <see_reply-to@myweb.nl> wrote:
> > "davexnet02" <davexnetzerotwo@hooya!.com> wrote in message news:v0vcr0debdd8cst84bsneoset51he1evd2@4ax.com
> > > Hello all,
> > > I noticed one type of operation where the performance
> > > is very low, it occurs when reading and writing lots of data
> > > simultaneously.
> > > For example, I had a RAR archive of a 235MB avi file.
> > > Extracting the archive from and to this drive took 1 minute 46.
> > > I copied the RAR files to a second drive, a Maxtor 53073h4
> > > and repeated the test, it took 60 seconds.
> > >
> > > I used the IBM feature tool utility and disabled the
> > > write cache. Repeating the test above in this
> > > condition, it still took 1 minute 46.
> > >
> > > Write cache non-operative?
> >
> > 2MB cache, 40MB/s transfer rate, or write cache filled in 1/40 second.
> > Gee, on a 1 minute 46 sec operation how much difference do you think that
> > will make?
> >
> Obviously I had no idea.
> If it really works as you describe, I can see not much at all.

> Since most HD accesses in a typical desktop scenario are pretty
> small reads and writes, then what's the point of the cache at all?

1. Being just that: cache.
2. Being an extension to the buffer so that several commands can be queued.

>
> However, if some of the problem is to do with timing, and the
> cache having an effect of smoothing out the operation,

It probably does somewhat do that but it may be only noticable on pure
sequential operations.
And do note that the buffer is already 128 kB and may take more than one
command of less than 128 kB transfer size.

> the benefit could be greater than the sum of the parts.
> OR so I thought.
>
>
> Dave
!