About simulate NVL72 on Astra-sim

[Zhongkai Yu]

Dear Astra-sim Team,

My name is Zhongkai Yu, and I am a second-year Ph.D. student in Computer Science and Engineering at UC San Diego. I am currently working on deploying Astra-sim to simulate the performance of the NVL72 system with 72 GPUs.

However, I noticed that the simulation relies on trace files generated from PyTorch, and I could only find trace files for systems with 2 to 16 GPUs in the github repo. Since we do not have access to a 72-GPU system, I was wondering if you have any suggestions or recommendations on how to generate or obtain a suitable trace file for a 72-GPU setup.

Thank you for your time and for providing such a well-designed tool that has saved us a great deal of effort in our research.

Best regards,
Zhongkai Yu

[Yoo, Jinsun]

Hi Zhongkai, 

Thank you for reaching out to the mailing list!

There are two ways you could generate pre-execution Chakra graphs: 

• Symbolic Tensor Graph: Generate Chakra graphs from a symbolic representation of the model: This method is more stable and has a Github repo you can try: link here
• FX Graphs: Another method is to generate chakra graphs directly from model source code using torch.compile's FX graphs: This method is currently undergoing work, although we are expecting to release more details soon.

Both methods do not need a GPU cluster, as the Chakra graph can be collected pre-execution.

Please feel free to reach out if you have any other questions.

Best

Jinsun

---

From: 'Zhongkai Yu' via ASTRA-sim Users <astrasi...@googlegroups.com>
Sent: Sunday, November 2, 2025 10:06 PM
To: astrasi...@googlegroups.com <astrasi...@googlegroups.com>
Subject: About simulate NVL72 on Astra-sim

 

--
You received this message because you are subscribed to the Google Groups "ASTRA-sim Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrasim-user...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/astrasim-users/CAHJKqK-i-KhNe5V%3Duv2yXR%3DJeZpBd6mY6ABEhTQjFCKJSo3i1Q%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.