Editing article

Title

Summary

<div class="block-paragraph_advanced">Today we’re excited to announce the general availability of Rocky Linux 8-based and CentOS 7-based HPC Virtual Machine (VM) images for high-performance computing (HPC) workloads, with a focus on tightly-coupled workloads, such as weather forecasting, fluid dynamics, and molecular modeling.
With the <a href="https://cloud.google.com/compute/docs/instances/create-hpc-vm">HPC VM image</a>, we have made it easy to build an HPC-ready VM instance, incorporating our <a href="https://cloud.google.com/solutions/hpc?hl=en#section-7">best practices running HPC on Google Cloud</a>, including:
<ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
VMs ready for HPC out-of-the-box - No need to manually tune performance, manage VM reboots, or stay up to date with the latest Google Cloud updates for tightly-coupled HPC workloads, especially with our regular HPC VM image releases. Reboots will be automatically triggered when tunings require them and this process will be managed for you by the HPC VM image. 
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Networking optimizations for tightly-coupled workloads - Optimizations that reduce latency for small messages are included, which benefits applications that are heavily dependent on point-to-point and collective communications.
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Compute optimizations - Optimizations that reduce system jitter are included, which makes single-node performance consistent, important to improving scalability.
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Improved application compatibility - Alignment with the node-level requirements of the Intel HPC platform specification enables a high degree of interoperability between systems.
</li>
</ul>
<h3>Performance measurement using HPC benchmarks</h3>
We have compared the performance of the HPC VM images against the default CentOS 7 and GCP-optimized Rocky Linux 8 images across Intel MPI Benchmarks (IMB).
The benchmarks were run against the following images.
HPC Rocky Linux 8
<ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Image name: hpc-rocky-linux-8-v20240126
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Image project: cloud-hpc-image-public
</li>
</ul>
Default GCP Rocky Linux 8
<ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Image name: rocky-linux-8-optimized-gcp-v20240111
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Image project: rocky-linux-cloud
</li>
</ul>
Each cluster of machines was deployed with compact placement with <code style="vertical-align: baseline;">max_distance=1</code>, meaning all VMs were placed on hardware that were physically on the same rack to minimize network latency.
<h3>Intel MPI Benchmark (IMB) Ping-Pong</h3>
IMB Ping-Pong measures the latency when transferring a fixed-sized message between two ranks on different VMs. We saw up to a 15% improvement when using the HPC Rocky Linux 8 image compared to the default GCP Rocky Linux 8 image.
Benchmark setup
<ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
2 x h3-standard-88
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI library: Intel OneAPI MPI library 2021.11.0
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI benchmarks application: Intel MPI Benchmarks 2019 Update 6
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI environment variables:
</li>
<ul>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
I_MPI_PIN_PROCESSOR_LIST=0
</li>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
I_MPI_FABRICS=shm:ofi
</li>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
FI_PROVIDER=tcp
</li>
</ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Command line: mpirun -n 2 -ppn 1 -bind-to core -hostfile &lt;hostfile&gt; IMB-MPI1 Pingpong -msglog 0:16 -iter 50000
</li>
</ul>
Results</div>
<div class="block-image_full_width">

<img
 src="https://storage.googleapis.com/gweb-cloudblog-publish/images/1_X6ust7I.max-1000x1000.png"
 
 alt="1">
 
 </a>
 
 <figcaption class="article-image__caption ">Pingpong 1 PPN - Rocky Linux 8 (lower is better)</figcaption>
 
 </figure>

</div>
 </div>

</div>
<div class="block-paragraph_advanced"><h3>Intel MPI Benchmark (IMB) AllReduce - 1 process per node</h3>
The IMB AllReduce benchmark measures the collective latency among multiple ranks across VMs. It reduces a vector of a fixed length with the MPI_SUM operation.
To isolate networking performance, we initially show 1 PPN (process-per-node) results (1 MPI rank) on 8 VMs.
We saw an improvement of up to 35% when comparing the HPC Rocky Linux 8 image to the default GCP Rocky Linux 8 image.
Benchmark setup
<ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
8 x h3-standard-88
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
1 process per node
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI library: Intel OneAPI MPI library 2021.11.0
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI benchmarks application: Intel MPI Benchmarks 2019 Update 6
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI environment variables:
</li>
<ul>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
I_MPI_FABRICS=shm:ofi
</li>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
FI_PROVIDER=tcp
</li>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
I_MPI_ADJUST_ALLREDUCE=11
</li>
</ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Command line: mpirun -n 008 -ppn 01 -bind-to core -hostfile &lt;hostfile&gt; IMB-MPI1 Allreduce -msglog 0:16 -iter 50000 -npmin 008
</li>
</ul>
Results</div>
<div class="block-image_full_width">

<img
 src="https://storage.googleapis.com/gweb-cloudblog-publish/images/2_b9MR5jY.max-1000x1000.png"
 
 alt="2">
 
 </a>
 
 <figcaption class="article-image__caption ">Allreduce 1 PPN - Rocky Linux 8 (lower is better)</figcaption>
 
 </figure>

</div>
 </div>

</div>
<div class="block-paragraph_advanced"><h3>Intel MPI Benchmark (IMB) AllReduce - 1 process per core (88 processes per node)</h3>
We show 88 PPN results where there are 88 MPI ranks/node and 1 thread/rank (704 ranks).
For this test, we saw an improvement of up to 25% when comparing the HPC Rocky Linux 8 image to the default GCP Rocky Linux 8 image.
Benchmark setup
<ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
8 x h3-standard-88
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
1 process per core (88 processes per node)
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI library: Intel OneAPI MPI library 2021.11.0
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI benchmarks application: Intel MPI Benchmarks 2019 Update 6
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
MPI environment variables:
</li>
<ul>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
I_MPI_FABRICS=shm:ofi
</li>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
FI_PROVIDER=tcp
</li>
<li aria-level="2" style="list-style-type: circle; vertical-align: baseline;">
I_MPI_ADJUST_ALLREDUCE=11
</li>
</ul>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
Command line: mpirun -n 704 -ppn 88 -bind-to core -hostfile &lt;hostfile&gt; IMB-MPI1 Allreduce -msglog 0:16 -iter 50000 -npmin 704
</li>
</ul>
Results</div>
<div class="block-image_full_width">

<img
 src="https://storage.googleapis.com/gweb-cloudblog-publish/images/3_IOxuhIk.max-1000x1000.png"
 
 alt="3">
 
 </a>
 
 <figcaption class="article-image__caption ">Allreduce 88 PPN - Rocky Linux 8 (lower is better)</figcaption>
 
 </figure>

</div>
 </div>

</div>
<div class="block-paragraph_advanced">The latency, bandwidth, and jitter improvements in the HPC VM Image have resulted in <a href="https://cloud.google.com/blog/topics/hpc/introducing-hpc-vm-images">historically higher MPI workload performance</a>. We plan to update this blog as more performance results become available.
<h3>Cloud HPC Toolkit and the HPC VM image</h3>
You can use the HPC VM image through the <a href="https://cloud.google.com/hpc-toolkit/docs/overview">Cloud HPC Toolkit</a>, an open-source tool that simplifies the process of deploying environments for a variety of workloads, including HPC, AI, and machine learning. In fact, the Toolkit blueprints and Slurm images based on Rocky Linux 8 and CentOS 7 use the HPC VM image by default. Using the Cloud HPC Toolkit, you can add customization on top of the HPC VM image, including installing new software and changing configurations, making it even more useful. 
By using the Cloud HPC Toolkit to customize images based on the HPC VM Image, it is possible to create and share blueprints for producing optimized and specialized images, improving reproducibility while reducing setup time and effort.
<h3>How to get started</h3>
You can create an HPC-ready VM by using the following options:
<ul>
<li role="presentation"><a href="https://cloud.google.com/hpc-toolkit/docs/overview#:~:text=Cloud%20HPC%20Toolkit%20is%20open,performance%20computing%20(HPC)%20environments.">Cloud HPC Toolkit</a></li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
<a href="https://cloud.google.com/compute/docs/gcloud-compute">Google Cloud CLI</a>
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
<a href="https://console.cloud.google.com">Google Cloud console</a> - Note: the image is available through <a href="https://pantheon.corp.google.com/marketplace/product/click-to-deploy-images/hpc-vm-image-rocky-linux-8" rel="noopener" target="_blank">Cloud Marketplace</a> in the console.
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
<a href="https://cloud.google.com/blog/topics/hpc/introducing-the-latest-slurm-on-gcp-scripts">SchedMD's Slurm workload manager</a>, which uses the HPC VM image by default. For more information, see <a href="https://pantheon.corp.google.com/marketplace/product/click-to-deploy-images/hpc-vm-image-rocky-linux-8?mods=logs_tg_prod&amp;project=cloud-hpc-image-devel" rel="noopener" target="_blank">Creating Intel Select Solution verified clusters</a>.
</li>
<li aria-level="1" style="list-style-type: disc; vertical-align: baseline;">
<a href="https://gcp.cloudycluster.com/" rel="noopener" style="font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, 'Open Sans', 'Helvetica Neue', sans-serif;" target="_blank">Omnibond CloudyCluster</a>, which uses the HPC VM image by default.
</li>
</ul></div>

Content

Author

Link

Published date

Image url

Feed url

Guid

Hidden blurb

--- !ruby/object:Feedjira::Parser::RSSEntry
published: 2024-03-22 16:00:00.000000000 Z
entry_id: !ruby/object:Feedjira::Parser::GloballyUniqueIdentifier
 guid: https://cloud.google.com/blog/topics/hpc/ga-rocky-linux-8-and-centos-7-versions-of-hpc-vm-image/
title: Rocky Linux 8 and CentOS 7 versions of HPC VM image now generally available
categories:
- Compute
- HPC
carlessian_info:
 news_filer_version: 2
 newspaper: Google Cloud Blog
 macro_region: Technology
summary: "<div class=\"block-paragraph_advanced\">Today we’re excited to announce the general availability of Rocky Linux
 8-based and CentOS 7-based HPC Virtual Machine (VM) images for high-performance
 computing (HPC) workloads, with a focus on tightly-coupled workloads, such as weather
 forecasting, fluid dynamics, and molecular modeling.\nWith the <a href=\"https://cloud.google.com/compute/docs/instances/create-hpc-vm\">HPC VM image</a>, we have made it easy to build an HPC-ready
 VM instance, incorporating our <a href=\"https://cloud.google.com/solutions/hpc?hl=en#section-7\">best practices running
 HPC on Google Cloud</a>, including:\n<ul>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nVMs ready for
 HPC out-of-the-box - No need
 to manually tune performance, manage VM reboots, or stay up to date with the latest
 Google Cloud updates for tightly-coupled HPC workloads, especially with our regular
 HPC VM image releases. Reboots will be automatically triggered when tunings require
 them and this process will be managed for you by the HPC VM image. \n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nNetworking optimizations
 for tightly-coupled workloads
 - Optimizations that reduce latency for small messages are included, which benefits
 applications that are heavily dependent on point-to-point and collective communications.\n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nCompute optimizations
 - Optimizations that reduce system
 jitter are included, which makes single-node performance consistent, important to
 improving scalability.\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nImproved application compatibility - Alignment with the node-level requirements of the Intel HPC platform
 specification enables a high degree of interoperability between systems.\n</li>\n</ul>\n<h3>Performance measurement using HPC benchmarks</h3>\nWe have compared the performance of the HPC
 VM images against the default CentOS 7 and GCP-optimized Rocky Linux 8 images across
 Intel MPI Benchmarks (IMB).\nThe
 benchmarks were run against the following images.\nHPC Rocky Linux 8\n<ul>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nImage name: hpc-rocky-linux-8-v20240126\n</li>\n<li aria-level=\"1\"
 style=\"list-style-type: disc; vertical-align: baseline;\">\nImage project: cloud-hpc-image-public\n</li>\n</ul>\nDefault GCP Rocky Linux 8\n<ul>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nImage name: rocky-linux-8-optimized-gcp-v20240111\n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nImage project: rocky-linux-cloud\n</li>\n</ul>\nEach cluster of machines was deployed with compact
 placement with <code style=\"vertical-align: baseline;\">max_distance=1</code>, meaning all VMs were placed on hardware that
 were physically on the same rack to minimize network latency.\n<h3>Intel MPI Benchmark (IMB) Ping-Pong</h3>\nIMB Ping-Pong measures the latency when transferring
 a fixed-sized message between two ranks on different VMs. We saw up to a 15% improvement
 when using the HPC Rocky Linux 8 image compared to the default GCP Rocky Linux 8
 image.\nBenchmark
 setup\n<ul>\n<li aria-level=\"1\" style=\"list-style-type: disc;
 vertical-align: baseline;\">\n2 x h3-standard-88\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nMPI library: Intel OneAPI MPI library 2021.11.0\n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nMPI benchmarks application:
 Intel MPI Benchmarks 2019 Update 6\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nMPI environment variables:\n</li>\n<ul>\n<li aria-level=\"2\"
 style=\"list-style-type: circle; vertical-align: baseline;\">\nI_MPI_PIN_PROCESSOR_LIST=0\n</li>\n<li
 aria-level=\"2\" style=\"list-style-type: circle; vertical-align: baseline;\">\nI_MPI_FABRICS=shm:ofi\n</li>\n<li
 aria-level=\"2\" style=\"list-style-type: circle; vertical-align: baseline;\">\nFI_PROVIDER=tcp\n</li>\n</ul>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nCommand line: mpirun
 -n 2 -ppn 1 -bind-to core -hostfile &lt;hostfile&gt; IMB-MPI1 Pingpong -msglog 0:16
 -iter 50000\n</li>\n</ul>\nResults</div>\n<div
 class=\"block-image_full_width\">\n\n\n\n\n\n\n \n <div class=\"article-module
 h-c-page\">\n <div class=\"h-c-grid\">\n \n\n <figure class=\"article-image--large\n
 \ \n \n h-c-grid__col\n h-c-grid__col--6 h-c-grid__col--offset-3\n
 \ \n \n \"\n >\n\n \n \n \n <img\n
 \ src=\"https://storage.googleapis.com/gweb-cloudblog-publish/images/1_X6ust7I.max-1000x1000.png\"\n
 \ \n alt=\"1\">\n \n </a>\n \n <figcaption
 class=\"article-image__caption \">Pingpong 1 PPN - Rocky
 Linux 8 (lower is better)</figcaption>\n \n </figure>\n\n \n </div>\n
 \ </div>\n \n\n\n\n\n</div>\n<div class=\"block-paragraph_advanced\"><h3>Intel MPI Benchmark (IMB) AllReduce - 1 process
 per node</h3>\nThe IMB AllReduce
 benchmark measures the collective latency among multiple ranks across VMs. It reduces
 a vector of a fixed length with the MPI_SUM operation.\nTo isolate networking performance, we initially show 1 PPN (process-per-node)
 results (1 MPI rank) on 8 VMs.\nWe
 saw an improvement of up to 35% when comparing the HPC Rocky Linux 8 image to the
 default GCP Rocky Linux 8 image.\nBenchmark setup\n<ul>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\n8 x h3-standard-88\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\n1 process per node\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nMPI library: Intel OneAPI MPI library 2021.11.0\n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nMPI benchmarks application:
 Intel MPI Benchmarks 2019 Update 6\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nMPI environment variables:\n</li>\n<ul>\n<li aria-level=\"2\"
 style=\"list-style-type: circle; vertical-align: baseline;\">\nI_MPI_FABRICS=shm:ofi\n</li>\n<li
 aria-level=\"2\" style=\"list-style-type: circle; vertical-align: baseline;\">\nFI_PROVIDER=tcp\n</li>\n<li
 aria-level=\"2\" style=\"list-style-type: circle; vertical-align: baseline;\">\nI_MPI_ADJUST_ALLREDUCE=11\n</li>\n</ul>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nCommand line: mpirun
 -n 008 -ppn 01 -bind-to core -hostfile &lt;hostfile&gt; IMB-MPI1 Allreduce -msglog
 0:16 -iter 50000 -npmin 008\n</li>\n</ul>\nResults</div>\n<div class=\"block-image_full_width\">\n\n\n\n\n\n\n
 \ \n <div class=\"article-module h-c-page\">\n <div class=\"h-c-grid\">\n
 \ \n\n <figure class=\"article-image--large\n \n \n h-c-grid__col\n
 \ h-c-grid__col--6 h-c-grid__col--offset-3\n \n \n \"\n
 \ >\n\n \n \n \n <img\n src=\"https://storage.googleapis.com/gweb-cloudblog-publish/images/2_b9MR5jY.max-1000x1000.png\"\n
 \ \n alt=\"2\">\n \n </a>\n \n <figcaption
 class=\"article-image__caption \">Allreduce 1 PPN -
 Rocky Linux 8 (lower is better)</figcaption>\n \n </figure>\n\n \n
 \ </div>\n </div>\n \n\n\n\n\n</div>\n<div class=\"block-paragraph_advanced\"><h3>Intel MPI Benchmark (IMB) AllReduce - 1 process
 per core (88 processes per node)</h3>\nWe
 show 88 PPN results where there are 88 MPI ranks/node and 1 thread/rank (704 ranks).\nFor this test, we saw an improvement of up to
 25% when comparing the HPC Rocky Linux 8 image to the default GCP Rocky Linux 8
 image.\nBenchmark
 setup\n<ul>\n<li aria-level=\"1\" style=\"list-style-type: disc;
 vertical-align: baseline;\">\n8 x h3-standard-88\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\n1 process per core (88 processes per node)\n</li>\n<li aria-level=\"1\"
 style=\"list-style-type: disc; vertical-align: baseline;\">\nMPI library: Intel OneAPI MPI library 2021.11.0\n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nMPI benchmarks application:
 Intel MPI Benchmarks 2019 Update 6\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\nMPI environment variables:\n</li>\n<ul>\n<li aria-level=\"2\"
 style=\"list-style-type: circle; vertical-align: baseline;\">\nI_MPI_FABRICS=shm:ofi\n</li>\n<li
 aria-level=\"2\" style=\"list-style-type: circle; vertical-align: baseline;\">\nFI_PROVIDER=tcp\n</li>\n<li
 aria-level=\"2\" style=\"list-style-type: circle; vertical-align: baseline;\">\nI_MPI_ADJUST_ALLREDUCE=11\n</li>\n</ul>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\nCommand line: mpirun
 -n 704 -ppn 88 -bind-to core -hostfile &lt;hostfile&gt; IMB-MPI1 Allreduce -msglog
 0:16 -iter 50000 -npmin 704\n</li>\n</ul>\nResults</div>\n<div class=\"block-image_full_width\">\n\n\n\n\n\n\n
 \ \n <div class=\"article-module h-c-page\">\n <div class=\"h-c-grid\">\n
 \ \n\n <figure class=\"article-image--large\n \n \n h-c-grid__col\n
 \ h-c-grid__col--6 h-c-grid__col--offset-3\n \n \n \"\n
 \ >\n\n \n \n \n <img\n src=\"https://storage.googleapis.com/gweb-cloudblog-publish/images/3_IOxuhIk.max-1000x1000.png\"\n
 \ \n alt=\"3\">\n \n </a>\n \n <figcaption
 class=\"article-image__caption \">Allreduce 88 PPN -
 Rocky Linux 8 (lower is better)</figcaption>\n \n </figure>\n\n \n
 \ </div>\n </div>\n \n\n\n\n\n</div>\n<div class=\"block-paragraph_advanced\">The latency, bandwidth, and jitter improvements
 in the HPC VM Image have resulted in <a href=\"https://cloud.google.com/blog/topics/hpc/introducing-hpc-vm-images\">historically higher
 MPI workload performance</a>. We
 plan to update this blog as more performance results become available.\n<h3>Cloud HPC Toolkit and the HPC VM image</h3>\nYou can use the HPC VM image through the <a
 href=\"https://cloud.google.com/hpc-toolkit/docs/overview\">Cloud HPC Toolkit</a>, an open-source tool that simplifies the process of deploying environments
 for a variety of workloads, including HPC, AI, and machine learning. In fact, the
 Toolkit blueprints and Slurm images based on Rocky Linux 8 and CentOS 7 use the
 HPC VM image by default. Using the Cloud HPC Toolkit, you can add customization
 on top of the HPC VM image, including installing new software and changing configurations,
 making it even more useful. \nBy
 using the Cloud HPC Toolkit to customize images based on the HPC VM Image, it is
 possible to create and share blueprints for producing optimized and specialized
 images, improving reproducibility while reducing setup time and effort.\n<h3>How to get started</h3>\nYou can create an HPC-ready VM by using the following options:\n<ul>\n<li
 role=\"presentation\"><a href=\"https://cloud.google.com/hpc-toolkit/docs/overview#:~:text=Cloud%20HPC%20Toolkit%20is%20open,performance%20computing%20(HPC)%20environments.\">Cloud HPC Toolkit</a></li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\n<a href=\"https://cloud.google.com/compute/docs/gcloud-compute\">Google Cloud CLI</a>\n</li>\n<li
 aria-level=\"1\" style=\"list-style-type: disc; vertical-align: baseline;\">\n<a href=\"https://console.cloud.google.com\">Google Cloud console</a> - Note: the image is available through <a href=\"https://pantheon.corp.google.com/marketplace/product/click-to-deploy-images/hpc-vm-image-rocky-linux-8\"
 rel=\"noopener\" target=\"_blank\">Cloud Marketplace</a>
 in the console.\n</li>\n<li aria-level=\"1\" style=\"list-style-type:
 disc; vertical-align: baseline;\">\n<a href=\"https://cloud.google.com/blog/topics/hpc/introducing-the-latest-slurm-on-gcp-scripts\">SchedMD's Slurm
 workload manager</a>, which uses
 the HPC VM image by default. For more information, see <a href=\"https://pantheon.corp.google.com/marketplace/product/click-to-deploy-images/hpc-vm-image-rocky-linux-8?mods=logs_tg_prod&amp;project=cloud-hpc-image-devel\"
 rel=\"noopener\" target=\"_blank\">Creating Intel Select Solution verified clusters</a>.\n</li>\n<li aria-level=\"1\" style=\"list-style-type: disc;
 vertical-align: baseline;\">\n<a href=\"https://gcp.cloudycluster.com/\"
 rel=\"noopener\" style=\"font-family: -apple-system, BlinkMacSystemFont, 'Segoe
 UI', Roboto, Oxygen, Ubuntu, Cantarell, 'Open Sans', 'Helvetica Neue', sans-serif;\"
 target=\"_blank\">Omnibond
 CloudyCluster</a>, which uses
 the HPC VM image by default.\n</li>\n</ul></div>"
rss_fields:
- title
- url
- summary
- author
- categories
- published
- entry_id
url: https://cloud.google.com/blog/topics/hpc/ga-rocky-linux-8-and-centos-7-versions-of-hpc-vm-image/
author: Rohit Ramu

Language

Active

Ricc internal notes

Ricc source

Show this article Back to articles