<div dir="ltr">Please keep the list updated on what you find,</div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, 31 Jul 2023 at 20:08, Andrew Falgout <<a href="mailto:andrew.falgout@gmail.com">andrew.falgout@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Not ignoring you guys, literally have been moving. We had to downsize, I've got no power and even my main computer is still powered off. There's nothing more eerie than a quiet computer room. I'm on an old laptop that I threw Linux Mint 21 on to be here now. Okay.. to introduce myself a bit more.<br>I've been doing linux for a long time, but been in a silo for a long time. I feel like I've not used so many skills that I can't trust them anymore. So I'm mentally just marking my cache as dirty and going to relearn as much as I can. Great information so far. <br clear="all"><div><div dir="ltr" class="gmail_signature"><div dir="ltr"><div dir="ltr"><p style="margin:0in 0in 12pt;font-size:11pt;font-family:Calibri,sans-serif"><span style="font-size:11pt">I have hardware and storage space to play with. (Dell R930 112 core/600gb of ram) The issue is getting a graphics card in them for compute is really not proving to be ideal. I have about 4 of these machines, and I'd like to play around with clustering. Learning how to properly and securely plan and implement them. I've played around with docker, and have used multiple docker servers with portainer. Next, when I get an electrician to install power, is to try to setup a kubernetes cluster. <br>When I can get something with some decent compute (not this laptop), I'd like to learn how to train a small llm model using the cluster if possible. I know I can do a good bit slowly with the CPU. If I can get a GPU in the mix, doing that to speed things up.<br>Again.. I would like to apologize for being quiet for so long. I'll try to toss an "ack" in there from my phone if nothing else.</span></p><p style="margin:0in 0in 12pt;font-size:11pt;font-family:Calibri,sans-serif"><span style="font-size:11pt"><br>./Andrew Falgout<br></span>KG5GRX</p></div></div></div></div><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jul 31, 2023 at 6:10 AM John Hearns <<a href="mailto:hearnsj@gmail.com" target="_blank">hearnsj@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto">A quick ack would be nice. </div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, 28 Jul 2023, 06:38 John Hearns, <<a href="mailto:hearnsj@gmail.com" target="_blank">hearnsj@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Andrew, the answer is very much yes. I guess you are looking at the interface of 'traditional' HPC which uses workload schedulers and Kubernetes style clusters which use containers.<div>Firstly I would ask if you are coming from the point of view of someone who wants to build a cluster in your home or company using kit which you already have.</div><div>Or are you a company which wants to set up an AI infrastructure?</div><div><br></div><div>By the way, I think you are thinking on a CPU cluster and scaling out using Beowulf concepts.</div><div>In that case you are looking at Horovod <a href="https://github.com/horovod/horovod" rel="noreferrer" target="_blank">https://github.com/horovod/horovod</a></div><div>One thing though - for AI applications it is common to deploy Beowulf clusters which have servers with GPUs as part of their specification.</div><div><br></div><div><br><div>I think it will be clear to you soon that you will be overwhelmed with options and opinions.</div><div>Firstly join the hpc.social community and introduce yourself on the Slack channel introductions</div><div>I would start with the following resources:</div><div><br></div><div><a href="https://www.clustermonkey.net/" rel="noreferrer" target="_blank">https://www.clustermonkey.net/</a><br></div><div><a href="https://www.nvidia.com/en-gb/data-center/bright-cluster-manager/" rel="noreferrer" target="_blank">https://www.nvidia.com/en-gb/data-center/bright-cluster-manager/</a><br></div><div><a href="https://catalog.ngc.nvidia.com/containers" rel="noreferrer" target="_blank">https://catalog.ngc.nvidia.com/containers</a><br></div><div><a href="https://openhpc.community/" rel="noreferrer" target="_blank">https://openhpc.community/</a><br></div><div><a href="https://ciq.com/" rel="noreferrer" target="_blank">https://ciq.com/</a><br></div><div><a href="https://qlustar.com/" rel="noreferrer" target="_blank">https://qlustar.com/</a><br></div><div><a href="https://www.delltechnologies.com/asset/en-nz/products/ready-solutions/technical-support/omnia-solution-overview.pdf" rel="noreferrer" target="_blank">https://www.delltechnologies.com/asset/en-nz/products/ready-solutions/technical-support/omnia-solution-overview.pdf</a><br></div><div><a href="https://omnia-doc.readthedocs.io/en/latest/index.html" rel="noreferrer" target="_blank">https://omnia-doc.readthedocs.io/en/latest/index.html</a><br></div><div><br></div><div>Does anyone know if the Bright Easy8 licenses are available? I would say that building test cluster with Easy 8 would be the quickest way to get some hands on experience.</div><div><br></div><div>You should of course consider cloud providers:</div><div><a href="https://aws.amazon.com/hpc/parallelcluster/" rel="noreferrer" target="_blank">https://aws.amazon.com/hpc/parallelcluster/</a><br></div><div><a href="https://azure.microsoft.com/en-gb/solutions/high-performance-computing/#intro" rel="noreferrer" target="_blank">https://azure.microsoft.com/en-gb/solutions/high-performance-computing/#intro</a><br></div><div><a href="https://cloud.google.com/solutions/hpc" rel="noreferrer" target="_blank">https://cloud.google.com/solutions/hpc</a><br></div><div><a href="https://go.oracle.com/LP=134426" rel="noreferrer" target="_blank">https://go.oracle.com/LP=134426</a><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div><div><br></div></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, 28 Jul 2023 at 01:10, Andrew Falgout <<a href="mailto:andrew.falgout@gmail.com" rel="noreferrer" target="_blank">andrew.falgout@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">So I'm interested to see if a Beowulf Cluster could be used for Machine Learning, LLM training, and LLM inference. Anyone know where a good entry point is for learning Beowulf Clustering? <br clear="all"><div><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><p style="margin:0in 0in 12pt;font-size:11pt;font-family:Calibri,sans-serif"><span style="font-size:11pt"><br>./Andrew Falgout<br></span>KG5GRX</p></div></div></div></div></div><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Wed, Jul 26, 2023 at 8:39 AM Michael DiDomenico <<a href="mailto:mdidomenico4@gmail.com" rel="noreferrer" target="_blank">mdidomenico4@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">just a mailing list as far as i know. it used to get a lot more<br>
traffic, but seems to have simmered down quite a bit<br>
<br>
On Tue, Jul 25, 2023 at 6:50 PM Andrew Falgout <<a href="mailto:andrew.falgout@gmail.com" rel="noreferrer" target="_blank">andrew.falgout@gmail.com</a>> wrote:<br>
><br>
> Just curious, do we have a discord channel, or just a mailing list?<br>
><br>
><br>
> ./Andrew Falgout<br>
> KG5GRX<br>
><br>
><br>
><br>
> On Fri, Jul 21, 2023 at 9:12 AM Michael DiDomenico <<a href="mailto:mdidomenico4@gmail.com" rel="noreferrer" target="_blank">mdidomenico4@gmail.com</a>> wrote:<br>
>><br>
>> ugh, as someone who worked the front lines in the 00's i got front row<br>
>> seat to the interconnect mud slinging... but franky if they're going<br>
>> to come out of the gate with a product named "Ultra Ethernet", i smell<br>
>> a loser... :) (sarcasm...)<br>
>><br>
>> <a href="https://www.nextplatform.com/2023/07/20/ethernet-consortium-shoots-for-1-million-node-clusters-that-beat-infiniband/" rel="noreferrer noreferrer" target="_blank">https://www.nextplatform.com/2023/07/20/ethernet-consortium-shoots-for-1-million-node-clusters-that-beat-infiniband/</a><br>
>> _______________________________________________<br>
>> Beowulf mailing list, <a href="mailto:Beowulf@beowulf.org" rel="noreferrer" target="_blank">Beowulf@beowulf.org</a> sponsored by Penguin Computing<br>
>> To change your subscription (digest mode or unsubscribe) visit <a href="https://beowulf.org/cgi-bin/mailman/listinfo/beowulf" rel="noreferrer noreferrer" target="_blank">https://beowulf.org/cgi-bin/mailman/listinfo/beowulf</a><br>
</blockquote></div>
_______________________________________________<br>
Beowulf mailing list, <a href="mailto:Beowulf@beowulf.org" rel="noreferrer" target="_blank">Beowulf@beowulf.org</a> sponsored by Penguin Computing<br>
To change your subscription (digest mode or unsubscribe) visit <a href="https://beowulf.org/cgi-bin/mailman/listinfo/beowulf" rel="noreferrer noreferrer" target="_blank">https://beowulf.org/cgi-bin/mailman/listinfo/beowulf</a><br>
</blockquote></div>
</blockquote></div>
</blockquote></div>
</blockquote></div>