Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

[NFS] Noticed your publication.

0 views
Skip to first unread message

Spencer Shepler

unread,
May 10, 2001, 3:27:57 PM5/10/01
to
On Thu, Don Capps wrote:

>
> The proper procedure for run, disclosure, and publication may be found in the SFS 2.0 Users Guide.
> The submission is sent to the SPEC office and reviewed by the SFS Subcommittee before
> the publication is made available on the SPEC web site. The review process also
> helps guarantee that the result is conformant with the run rules and that all information
> that is required in the disclosure is accurate.

I would like to point out that the SFS licensee is not required to
submit published results to SPEC. The only constraint is that the run
and disclosure rules for the SFS benchmark be followed.

--

- Spencer -

_______________________________________________
NFS maillist - N...@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/nfs

Don Capps

unread,
May 10, 2001, 3:21:50 PM5/10/01
to
Dear Sir,
 
    I noticed your publication of SFS97 results:
 
>For example I have obtained the following results using 1 Solaris 8 intel
> client with 2 cpus and Linux nfs server (2.4.4 with latest nfs patches):
>
> Ops       Ops       Response
> requested performed time
>                     msec/op
> 1800      1825      1.7
> 2000      2021      1.9
> 2200      2280      0.6
> 2400      1935      2.4
> 2600      2658      0.8
> 2800      2811      0.8
>
> Any ideas would be greatly appreciated.
 
Since this appears  to be a public forum I would mention that
you may be in violation of your SFS license. In the SFS 2.0
guide you will also find things like:
 
"As a requirement of the license of the benchmark, these run and disclosure rules must be followed. If the user of the SFS 2.0 benchmark suite does not adhere to the rules set forth herein, SPEC may choose to terminate the license with the user. "
 
Please take a moment to refresh your memory on the SFS license agreement
and disclosure rules.
 
Also, support for the SPEC SFS 2.0 benchmark is normally conducted by sending
email to the SFS support mail alias sfs97s...@spec.org
 
The proper procedure for run, disclosure, and publication may be found in the SFS 2.0 Users Guide.
The submission is sent to the SPEC office and reviewed by the SFS Subcommittee before
the publication is made available on the SPEC web site. The review process also
helps guarantee that the result is conformant with the run rules and that all information
that is required in the disclosure is accurate.
 
Thank you,
Don Capps
SPEC SFS Chair.
 
 
 

 

Greg Lindahl

unread,
May 10, 2001, 3:46:08 PM5/10/01
to
On Thu, May 10, 2001 at 02:27:57PM -0500, Spencer Shepler wrote:

> I would like to point out that the SFS licensee is not required to
> submit published results to SPEC. The only constraint is that the run
> and disclosure rules for the SFS benchmark be followed.

This is certainly true for SPEC CPU benchmarks. I always add a
disclaimer to my results: "These results can't be official because of
the following deficiencies: X, Y, Z." That way no one is confused.

For example, for a long time SPEC CPU couldn't be built on AlphaLinux,
and it took the committee 2 years to approve my fairly trivial changes
to the tools. During that time my runs were not able to be official.
But they didn't object to the way I published them.

-- g

Denis Gerasimov

unread,
May 10, 2001, 5:37:54 PM5/10/01
to
Dear Don and fellow NFS'ers:

On Thu, 10 May 2001, Don Capps wrote:
> Since this appears to be a public forum I would mention that
> you may be in violation of your SFS license.

I apologize for this violation. I did not realize that such an informal
message would be counted as a disclosure. I assure you that I and our
company as a SPEC SFS97 licensee will not repeat this mistake again.

> Also, support for the SPEC SFS 2.0 benchmark is normally conducted by sending
> email to the SFS support mail alias sfs97s...@spec.org

I will certainly try to use the mentioned support address again dispite
that in the past it has been less then helpful.

Particularly when I have asked for help with Linux port of SFS97, I have
been told that a version working with Linux is not available. However
there are already officially published results with Linux clients, for
example:
http://www.specbench.org/osg/sfs97/results/res2001q1/sfs97-20010122-00239.html
Which means there IS a version (or a set of changes) that were approved by
by the SFS Subcommittee.

Sincerely,
Dennis Gerasimov

Greg Lindahl

unread,
May 11, 2001, 1:46:05 PM5/11/01
to
On Thu, May 10, 2001 at 05:37:54PM -0400, Denis Gerasimov wrote:

> Particularly when I have asked for help with Linux port of SFS97, I have
> been told that a version working with Linux is not available. However
> there are already officially published results with Linux clients, for
> example:

Funny, that's the same problem I had with the SPEC CPU people.

-- g

0 new messages