search

kymata-atlas / kymata-core

Core Kymata codebase, including statistical analysis and plotting tools
https://kymata.org
MIT License
5 stars 0 forks source link

Do we need to reduce memory overhead when working with source-space gridsearch? #196

Open caiw opened 6 months ago

caiw commented 6 months ago

Removed the following comment from load_single_emeg() as part of #173/#184:

            # TODO: currently this goes OOM (node-h04 atleast):
            #       looks like this will be faster when split up anyway
            #       note, don't run the inv_op twice for rh and lh!

However I'm not sure it's relevant any more.

Context:

            lh_emeg, rh_emeg, ch_names = inverse_operate(evoked[0], inverse_operator, snr, morph_map=morph_map)
            ...
            # TODO: currently this goes OOM (node-h04 atleast):
            #       looks like this will be faster when split up anyway
            #       note, don't run the inv_op twice for rh and lh!
            ...
            emeg = np.concatenate((lh_emeg, rh_emeg), axis=0)
            del lh_emeg, rh_emeg