Create accountLogin

Recent searches

No recent searches

Search options

Only available when logged in.
en.osm.town is one of the many independent Mastodon servers you can use to participate in the fediverse.
An independent, community of OpenStreetMap people on the Fediverse/Mastodon. Funding graciously provided by the OpenStreetMap Foundation.

Administered by:

Server stats:

254
active users

en.osm.town: AboutProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.3.7

#gh200

0 posts0 participants0 posts today
Benjamin Carr, Ph.D. 👨🏻‍💻🧬<p><a href="https://hachyderm.io/tags/NVIDIA" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NVIDIA</span></a> <a href="https://hachyderm.io/tags/GH200" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GH200</span></a> Grace <a href="https://hachyderm.io/tags/CPU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>CPU</span></a> vs. <a href="https://hachyderm.io/tags/AMD" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AMD</span></a> <a href="https://hachyderm.io/tags/EPYC" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>EPYC</span></a> 9005 <a href="https://hachyderm.io/tags/Turin" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Turin</span></a> CPU Performance Review<br>The article is focusing just on the CPU performance across these servers. Of course, the main selling point of the GH200 Superchip is when leveraging the Hopper GPU in tandem. But for those curious how well Grace / Neoverse-V2 compares to the new <a href="https://hachyderm.io/tags/EPYC9005" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>EPYC9005</span></a> series in single socket configurations, hopefully today's numbers provided some insight. <br><a href="https://www.phoronix.com/review/nvidia-grace-epyc-turin" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">phoronix.com/review/nvidia-gra</span><span class="invisible">ce-epyc-turin</span></a></p>
Benjamin Carr, Ph.D. 👨🏻‍💻🧬<p>A Quick Introduction to the <a href="https://hachyderm.io/tags/NVIDIA" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NVIDIA</span></a> <a href="https://hachyderm.io/tags/GH200" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GH200</span></a> aka <a href="https://hachyderm.io/tags/GraceHopper" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GraceHopper</span></a><br>If you want Arm CPU with LPDDR5X memory &amp; high-speed interconnect to a Hopper GPU today, you are looking at NVIDIA GH200. NVIDIA has two key innovations here that really set it apart from what you can get elsewhere. First, LPDDR5X memory is soldered onto the package. Second, the NVLink-C2C is a high-bandwidth interface between the CPU and the GPU. The 72 Arm v9 cores are Arm Neoverse V2 from 2022 so not exciting.<br><a href="https://www.servethehome.com/a-quick-introduction-to-the-nvidia-gh200-aka-grace-hopper-arm/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">servethehome.com/a-quick-intro</span><span class="invisible">duction-to-the-nvidia-gh200-aka-grace-hopper-arm/</span></a></p>
Benjamin Carr, Ph.D. 👨🏻‍💻🧬<p><a href="https://hachyderm.io/tags/GraceHopper" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GraceHopper</span></a>, <a href="https://hachyderm.io/tags/Nvidia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Nvidia</span></a>’s Halfway <a href="https://hachyderm.io/tags/APU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>APU</span></a><br>Grace Hopper is an attempt to get Nvidia’s CPU efforts into high performance territory. <a href="https://hachyderm.io/tags/GH200" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GH200</span></a> bundles a <a href="https://hachyderm.io/tags/CPU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>CPU</span></a> and <a href="https://hachyderm.io/tags/GPU" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GPU</span></a> together. The Grace CPU consists of 72 <a href="https://hachyderm.io/tags/Neoverse" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Neoverse</span></a> V2 cores running at up to 3.44 GHz, supported by 114 MB of L3 cache. Cores and L3 cache sit on top of Nvidia’s Scalable Coherency Fabric (SCF). SCF is a mesh interconnect, with cores and L3 cache slices attached to mesh stops. <br><a href="https://chipsandcheese.com/2024/07/31/grace-hopper-nvidias-halfway-apu/" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">chipsandcheese.com/2024/07/31/</span><span class="invisible">grace-hopper-nvidias-halfway-apu/</span></a> <a href="https://hachyderm.io/tags/HPC" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>HPC</span></a></p>
Simon B<p>Took delivery of a couple of servers containing the new <a href="https://fosstodon.org/tags/Nvidia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Nvidia</span></a> Grace Hopper <a href="https://fosstodon.org/tags/GH200" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GH200</span></a> superchip which is a 72 core ARM CPU with an nvlink connected H100 GPU (<a href="https://resources.nvidia.com/en-us-grace-cpu/grace-hopper-superchip" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">resources.nvidia.com/en-us-gra</span><span class="invisible">ce-cpu/grace-hopper-superchip</span></a>). These will also be used for the forthcoming UK national <a href="https://fosstodon.org/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> supercomputer. Haven't found any pictures online yet, beyond the fancy nvidia marketing, so here's some real world pictures. OS going on shortly, benchmarks soon.</p>
Benjamin Carr, Ph.D. 👨🏻‍💻🧬<p>21 <a href="https://hachyderm.io/tags/ExaFLOP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>ExaFLOP</span></a> <a href="https://hachyderm.io/tags/Isambard" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Isambard</span></a> <a href="https://hachyderm.io/tags/AI" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>AI</span></a> <a href="https://hachyderm.io/tags/Supercomputer" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Supercomputer</span></a> Uses 5,448 <a href="https://hachyderm.io/tags/NVidia" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NVidia</span></a> <a href="https://hachyderm.io/tags/GH200" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GH200</span></a> <a href="https://hachyderm.io/tags/GraceHopper" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>GraceHopper</span></a> Superchips<br>U.K.'s most powerful supercomputer boasts 200 <a href="https://hachyderm.io/tags/PetaFLOP" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>PetaFLOP</span></a> <a href="https://hachyderm.io/tags/FP64" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>FP64</span></a> compute.<br>The machine, which is set to be built by <a href="https://hachyderm.io/tags/HPE" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>HPE</span></a>, cost £225 million ($273.38 million) <br><a href="https://www.tomshardware.com/pc-components/gpus/21-exaflop-isambard-ai-supercomputer-uses-5448-gh200-grace-hopper-superchips" rel="nofollow noopener noreferrer" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="ellipsis">tomshardware.com/pc-components</span><span class="invisible">/gpus/21-exaflop-isambard-ai-supercomputer-uses-5448-gh200-grace-hopper-superchips</span></a></p>
ExploreLive feeds
About