I ran memory profiler on our examples to track down memory consuming functions. I thought that I'd share the results here if someone else wants to look at it. The examples listed here have a peak consumption of over 1 GiB.

• datasets/plot_brainstorm_data.py (MRG: Better memory usage (#2) #2977)
• datasets/plot_spm_faces_dataset.py (MRG: Better memory usage (#2) #2977)
• decoding/plot_decoding_spatio_temporal_source.py (MRG: Better memory usage (#2) #2977)
• decoding/plot_decoding_time_generalization.py (MRG: Better memory usage (#2) #2977)
• inverse/plot_covariance_whitening_dspm.py (MRG: Better memory usage (#2) #2977)
• inverse/plot_tf_lcmv.py (MRG: Better memory usage (#2) #2977)
• visualization/plot_meg_eeg_fields_3d.py (MRG: Better memory usage (#2) #2977)
• forward/plot_source_space_morphing.py (MRG: Less memory #2972)
• preprocessing/plot_resample.py (MRG: Less memory #2972)
• simulation/plot_simulate_raw_data.py (MRG: Less memory #2972)
• stats/plot_linear_regression_raw.py (MRG: Less memory #2972)
• time_frequency/plot_compute_raw_data_spectrum.py (MRG: Less memory #2972)
• time_frequency/plot_epochs_spectra.py (MRG: Less memory #2972)