[MITgcm-support] 答复: Workstation advice for MITgcm passive tracer runs

[QI Yuanzhi]

Hi Dimitris,

Thank you for the concrete hardware suggestions. They are very helpful.

Best,
Yuanzhi

发件人: MITgcm-support <mitgcm-support-bounces at mitgcm.org> 代表 Dimitris Menemenlis <dmenemenlis at gmail.com>
日期: 星期六, 2025年11月22日 09:57
收件人: MITgcm Support <mitgcm-support at mitgcm.org>
主题: Re: [MITgcm-support] Workstation advice for MITgcm passive tracer runs

It depends what resolution you want to run.

3-degree global, e.g., https://github.com/MITgcm-contrib/ecco_darwin/tree/master/v05/3deg
should run fine on your laptop

1-degree ECCO setup, e.g., https://github.com/MITgcm-contrib/llc_hires/blob/master/llc_90/ecco_v4r5/readme_v4r5_eady.txt
should run fin on a 32-core workstation with 256 GB

Dimitris


On Nov 21, 2025, at 4:51 PM, QI Yuanzhi <qi.yz at outlook.com> wrote:

Dear MITgcm community,

My name is Yuanzhi Qi. I am planning to use MITgcm to simulate the distribution and transport of radioactive tracers in the ocean. Specifically, I aim to model the historical concentration changes of Iodine-129 (I-129) in the Arctic Ocean over the past 80 years. This work is intended to serve as a foundation for future global-scale I-129 simulations and, potentially, for using I-129 as a tracer to study the broader marine iodine cycle.

My main question is: Is it feasible to run such 3D simulations on a single workstation? If so, could you provide specific hardware recommendations—such as CPU, GPU, RAM, and storage?

For instance, should I consider a CPU with 16 cores, 32 cores, or more? Would 128 GB of RAM be sufficient, or would I need 256 GB or more?

Additionally, if there are any example setups or configurations for similar oceanic passive tracer studies beyond the User Manual, I would be very grateful for any references or pointers.

Thank you in advance for your guidance.

Best regards,

Yuanzhi


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20251123/d88c5b7a/attachment.html>