From my experience, RoCE will be just as fast if not faster than IB, inside a
single Ethernet switch, it’s when you go outside the switch you lose out.
The trick has been finding NICs that are supported natively by OFED. I tend to
still find the Mellanox NICs the most reliable and well supported.
> On 22 Nov 2022, at 06:16, Christopher Samuel wrote:
>
> On 11/21/22 4:39 am, Scott Atchley wrote:
>
>> We have OpenMPI running on Frontier with libfabric. We are using HPE's CXI
>> (Cray eXascale Interface) provider instead of RoCE though.
>
> Yeah I'm curious to know if Matt's issues are
Hi Jörg
> On 30 Jun 2020, at 10:09 am, Jörg Saßmannshausen
> wrote:
>
> Dear all,
>
> we are currently planning a new cluster and this time around the idea was to
> use OpenStack for the HPC part of the cluster as well.
>
> I was wondering if somebody has some first hand experiences on the
NVIDIA has a version of HPL floating around, but will only supply it under NDA,
and you’re definitely not allowed to share the version you have. Not that that
doesn’t happen of course, but NVIDIA would definitely prefer you didn’t.
Matt.
—
Matt Wallis
ma...@madmonks.org
> On 15 Aug 2
On 12/06/2018 6:06 PM, John Hearns via Beowulf wrote:
My personal take is that heirarchical storage is the answere,
automatically pushing files to slower and cheaper tiers.
This is my preference as well, if manual intervention is required, it
won't get done, but you do need to tune it a fair
Hi Jonathan,
> On 7 Feb 2015, at 6:20 pm, Jonathan Aquilina wrote:
>
> Can someone explain to me what exactly the purpose of hadoop is and what we
> mean when we say big data? Is this for data storage and retrieval? Number
> crunching?
Hadoop can be thought of as HTPC, High Throughput Computi
Hi Prentice,
> On 7 Feb 2015, at 9:35 am, Prentice Bisbal
> wrote:
>
> Do any of you disable swap on your compute nodes?
>
> I brought this up in a presentation I gave last night on HPC system
> administration, but realized I never actually did this, or know of anyone who
> has. I would twe
On 2 Jul 2014, at 4:19 pm, Jonathan Aquilina wrote:
> How would the same arguments apply if you are just dealing with dns
> servers web servers databases etc.
If you're just dealing with standard services then you're not really doing high
performance clustering, but fault tolerance.
In a typi
On 1 Jul 2014, at 5:11 pm, Christopher Samuel wrote:
> There were releases in 2011 and 2012 and the list is still active.
>
> http://sourceforge.net/projects/modules/files/Modules/
>
> The 3.2.10 release in 2012 fixed the infamous segfault bug that could
> result in corrupted environment varia
On 1 Jul 2014, at 4:36 pm, Olli-Pekka Lehto wrote:
> On a sidenote, EasyBuild was mentioned here earlier and it seems that they
> have IMO right idea of simplifying installs of the type of environment that
> Chris
> was describing and many (including us) seem to hold as best practice. Just
>
ation.
And then when it comes time to debug why the application doesn't work,
commercial compiler suites often come with very good debugging tools.
Matt.
--
Matt Wallis
ma...@madmonks.org
___
Beowulf mailing list, Beowulf@beowulf.org sponsored
11 matches
Mail list logo
