Add Support for Google Cloud A4 and A4X Machine Types #39
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 4 commits
October 20, 2025 00:47
Add SQL configuration for Google Cloud A4 machine series: - Accelerator-optimized family with NVIDIA B200 Blackwell GPUs - Sapphire Rapids CPU platform - 12TB Local SSD storage - 3.6 Tbps network bandwidth - Spot VM support enabled Reference: https://cloud.google.com/compute/docs/accelerator-optimized-machines#a4-vms
Add SQL configuration for Google Cloud A4X machine series: - Accelerator-optimized family with NVIDIA GB200 Grace Blackwell Superchips - ARM Neoverse V2 CPU platform - 12TB Local SSD storage - 2 Tbps network bandwidth - ARM architecture support - Spot VM support enabled Reference: https://cloud.google.com/compute/docs/accelerator-optimized-machines#a4x-vms
Add GPU type mappings for latest NVIDIA accelerators: - NVIDIA H200 141GB (nvidia-h200-141gb) - NVIDIA B200 Blackwell (nvidia-b200) - NVIDIA GB200 Grace Blackwell Superchip (nvidia-gb200) These GPUs are used in A3 Ultra, A4, and A4X machine series respectively. Reference: https://cloud.google.com/compute/docs/gpus
Update instances documentation: - Add A4 and A4X to machine types list - Fix A3 link (was incorrectly pointing to a2.sql) - Update resources section to reference A3, A4, and A4X accelerator-optimized machines This brings the documentation in sync with the newly added machine type configurations.
|
Thanks for the pull. As I understand it, you can only get A4 and A3 if you are special activated and have a separate contract. Am I right? How can we calculate the list price? Please see: Cyclenerd/google-cloud-pricing-cost-calculator#279 and Cyclenerd/google-cloud-pricing-cost-calculator#309 |
|
Note: The machine type |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
This PR adds support for the newly released Google Cloud Compute Engine machine types A4 and A4X, along with their associated NVIDIA GPU models (B200, GB200, and H200).
Motivation
Google Cloud has recently announced the A4 and A4X machine series featuring the latest NVIDIA Blackwell GPU architecture. These new accelerator-optimized machine types are designed for foundation model training and serving, representing a significant advancement in AI/ML compute capabilities.
Reference: https://cloud.google.com/compute/docs/gpus/
Changes Made
New Machine Type Configurations
1. A4 Machine Series (
instances/series/a4.sql)a4-highgpu-8g2. A4X Machine Series (
instances/series/a4x.sql)a4x-highgpu-4gGPU Model Support
Added support for the following NVIDIA GPU models in
instances/series/gpu/gpu_names.sql:nvidia-h200-141gb) - Used in A3 Ultranvidia-b200) - Used in A4nvidia-gb200) - Used in A4XDocumentation Updates
Updated
instances/README.mdto:a2.sql)Testing
All SQL files follow the existing project patterns and schema:
References
Checklist
Additional Notes
These machine types represent Google Cloud's latest offerings for AI/ML workloads:
Both machine types require capacity reservation or specific provisioning methods as outlined in the Google Cloud documentation.