Xeon Gold 6138's Running FBSD 11.1 and TrueOS Stable [GCC 5.4]

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Xeon Gold 6138's Running FBSD 11.1 and TrueOS Stable [GCC 5.4]

grarpamp
http://www.phoronix.com/vr.php?view=25148

While we have tested a number of Linux distributions on Intel's new
Xeon Scalable platform, here are some initial BSD tests using two Xeon
Gold 6138 processors with the Tyan GT24E-B7106 1U barebones server.
FreeBSD 11.1 and the FreeBSD-derivative desktop/workstation-focused
TrueOS (formerly known as PC-BSD) were the primary candidates for
testing. TrueOS stable is currently tracking FreeBSD 12.0-CURRENT
development.
When comparing the out-of-the-box performance of FreeBSD/TrueOS and
even with running under the GCC compiler rather than LLVM Clang, the
Linux distributions were offering noticeably better performance on
this dual Intel Xeon Gold server.
_______________________________________________
[hidden email] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-performance
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: Xeon Gold 6138's Running FBSD 11.1 and TrueOS Stable [GCC 5.4]

Allan Jude-9
On 2017-10-08 21:03, grarpamp wrote:

> http://www.phoronix.com/vr.php?view=25148
>
> While we have tested a number of Linux distributions on Intel's new
> Xeon Scalable platform, here are some initial BSD tests using two Xeon
> Gold 6138 processors with the Tyan GT24E-B7106 1U barebones server.
> FreeBSD 11.1 and the FreeBSD-derivative desktop/workstation-focused
> TrueOS (formerly known as PC-BSD) were the primary candidates for
> testing. TrueOS stable is currently tracking FreeBSD 12.0-CURRENT
> development.
> When comparing the out-of-the-box performance of FreeBSD/TrueOS and
> even with running under the GCC compiler rather than LLVM Clang, the
> Linux distributions were offering noticeably better performance on
> this dual Intel Xeon Gold server.
> _______________________________________________
> [hidden email] mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-performance
> To unsubscribe, send any mail to "[hidden email]"
>
You might want to try enabling turbo boost.

sysctl dev.cpu.0.freq=9999

This will change the cpu frequency from the default (2000 mhz) to 2001
mhz, which will enable turbo boost (this specific CPU bursts to 3700
mhz), so this will likely make a very large difference in your benchmarks.

To get more consistent results, you may actually want to disable
turboboost in the bios, and rerun the benchmarks on ALL of the operating
systems.

--
Allan Jude


signature.asc (851 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Xeon Gold 6138's Running FBSD 11.1 and TrueOS Stable [GCC 5.4]

O. Hartmann-5
Am Sun, 8 Oct 2017 22:42:01 -0400
Allan Jude <[hidden email]> schrieb:

> On 2017-10-08 21:03, grarpamp wrote:
> > http://www.phoronix.com/vr.php?view=25148
> >
> > While we have tested a number of Linux distributions on Intel's new
> > Xeon Scalable platform, here are some initial BSD tests using two Xeon
> > Gold 6138 processors with the Tyan GT24E-B7106 1U barebones server.
> > FreeBSD 11.1 and the FreeBSD-derivative desktop/workstation-focused
> > TrueOS (formerly known as PC-BSD) were the primary candidates for
> > testing. TrueOS stable is currently tracking FreeBSD 12.0-CURRENT
> > development.
> > When comparing the out-of-the-box performance of FreeBSD/TrueOS and
> > even with running under the GCC compiler rather than LLVM Clang, the
> > Linux distributions were offering noticeably better performance on
> > this dual Intel Xeon Gold server.
> > _______________________________________________
> > [hidden email] mailing list
> > https://lists.freebsd.org/mailman/listinfo/freebsd-performance
> > To unsubscribe, send any mail to "[hidden email]"
> >  
>
> You might want to try enabling turbo boost.
>
> sysctl dev.cpu.0.freq=9999
>
> This will change the cpu frequency from the default (2000 mhz) to 2001
> mhz, which will enable turbo boost (this specific CPU bursts to 3700
> mhz), so this will likely make a very large difference in your benchmarks.
>
> To get more consistent results, you may actually want to disable
> turboboost in the bios, and rerun the benchmarks on ALL of the operating
> systems.
>
Wow, the difference between the FreeBSDs and Linux performance is amazing and for those
looking at the first time on such benchmarks not knowing much about the turbo boost issue
one would definitely choose the faster one :-(

I'd appreciate results of a benchmark considering no boost and with a light sched on the
scheduler (doesn't Linux have a very sophisticated scheduler dealing/scaling very
efficient with lots of threads/cpu cores? This could also be a very interesting benchmark
pointing to AMDs new Epyc platform comprised also from lots of cpu cores).

Kind regards,

oh

--
O. Hartmann

Ich widerspreche der Nutzung oder Übermittlung meiner Daten für
Werbezwecke oder für die Markt- oder Meinungsforschung (§ 28 Abs. 4 BDSG).

attachment0 (321 bytes) Download Attachment