Closed leewujung closed 6 years ago
I'm looking into this (as of 2018-07-03 at about 09h15). The version as of 2018-07-02 at about 17h00 appears to have a file writing lag of 2 seconds according to the timestamp, which is highly undesirable. I'm considering OpenMP for multi-threading or some other alternative (e.g., rewriting the code structure) such that we can have parallel buffers running.
UPDATE (2018-07-03 at 16h00): This is a very important (and good) question. Presently, the delay for 3 channels is 1-2 seconds. It seems to vary file to file but not considerably with more channels. I suspect we are losing some data in the meantime of reading and writing. I'll have to take some quantitative measurements--hopefully @funnyabc can help with this as it is testing/performance related.
UPDATE (2018-07-05 at 09h30): I was reading the timestamp incorrectly; the latency is not as high as I reported. Upon closer inspection there seems to be about a 0.05 s lag. I don't know what is acceptable or satisfactory.
Will address this in #8 (during testing) with @funnyabc
What's the lag time between loops for saving files? how does this scale with the number of channels being recorded? Loop as described in #1 later comments.