Maximum variable size is exceeded vs. memory error.

[jvoigts]

I'm getting a max variable size error here on a long-ish recording (~7hrs) that seems to come from a large value of nBatches:
time 1591.36, pre clustered 48001 / 48215 batches Error using gpuArray.zeros Maximum variable size allowed on the device is exceeded. Error in clusterSingleBatches (line 111) ccb = gpuArray.zeros(nBatches, 'single');

I am using ops.ntbuff = 64; ops.NT = 32*1024+ ops.ntbuff; , and if I increase NT i'm starting to get out of memory errors. Does this impose a restriction on total recording duration unless I grab a card with more ram or is there some way to circumvent this limit?

[marius10p]

how many channels do you have?

[jvoigts]

This is with 64 channels, all tetrodes. As far as i see it, the ideal solution for us would be to just be able to sort 4 channels at a time, which currently doesnt work - seems related to #25

[LiangShe]

@jvoigts try to increase the batch size more you will get no error.

[marius10p]

Sorting 4 channels at a time now works, but it might still not fix your problem. Re-open this issue if needed.