Dear Tracy et al.

Estimating a determinstic DCM for fMRI should be quite quick – but this is dependent on the amount of data and complexity of the model. I would hope it would use 100% of the processing core for maximum efficiency. That also means it’s easy to parallelise – you can put your DCM estimation in a parfor loop and it will run them in parallel. An option to do this automatically will be included in the next SPM release.

Can you be more specific about how long your estimation is taking, and how much data you have in the model? If it’s taking a long time to converge, it could mean your model is having problems.

Best

Peter

From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Miller Tracy
Sent: 19 August 2018 04:21
To: [log in to unmask]
Subject: [SPM] 转发：[SPM] high CPU usage when DCM estimation

Dear DCM experts,

I have similar question with Xinqi. It takes a long time to estimate one model.
I was wondering if this question could be solved by optimizing our DCM algorithm.
And how long does it usually take to estimate one model with 4 ROIs for task-fMRI?
Thanks in advance.

Best,
Tracy

-------- 原始邮件 --------
主题：[SPM] high CPU usage when DCM estimation
发件人：Xinqi Zhou
收件人：[log in to unmask]
抄送：

Dear all,

I just do my first DCM analysis. A full model with 5 nodes and 3 conditions was estimated via spm_dcm_estimate function. It took too much time (around 40min) and CPU resourses (my CPU is i7) per subject. I have multiple models and subjects to do. So, does anyone have any suggestions for me on code or something else?

Thank you.

Best,
Xinqi Zhou