07:02 <_whitenotifier-e> [scopehal] azonenberg opened issue #621: Oscilloscope: save ADC configuration to scopesession files - https://github.com/glscopeclient/scopehal/issues/621

07:02 <_whitenotifier-e> [scopehal] azonenberg labeled issue #621: Oscilloscope: save ADC configuration to scopesession files - https://github.com/glscopeclient/scopehal/issues/621

07:04 <_whitenotifier-e> [scopehal] azonenberg labeled issue #622: MockOscilloscope: preserve trigger information when loading/saving scopesession files - https://github.com/glscopeclient/scopehal/issues/622

07:04 <_whitenotifier-e> [scopehal] azonenberg opened issue #622: MockOscilloscope: preserve trigger information when loading/saving scopesession files - https://github.com/glscopeclient/scopehal/issues/622

07:06 <_whitenotifier-e> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±2] https://github.com/glscopeclient/scopehal/compare/c4fe7d44eee2...663cee6f5998

07:06 <_whitenotifier-e> [scopehal] azonenberg 663cee6 - MockOscilloscope: preserve sample rate, memory depth, and transport information

07:06 <_whitenotifier-e> [scopehal] azonenberg commented on issue #492: MockOscilloscope: store original connection information (if applicable) so we can reconnect in the future - https://github.com/glscopeclient/scopehal/issues/492#issuecomment-1145656047

07:06 <_whitenotifier-e> [scopehal] azonenberg closed issue #492: MockOscilloscope: store original connection information (if applicable) so we can reconnect in the future - https://github.com/glscopeclient/scopehal/issues/492

07:07 <_whitenotifier-e> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±3] https://github.com/glscopeclient/scopehal-apps/compare/f51a1dbf97cb...899d4bf55225

07:07 <_whitenotifier-e> [scopehal-apps] azonenberg 899d4bf - OscilloscopeWindow: use new MockOscilloscope features for preserving transport information

07:07 <_whitenotifier-e> [scopehal] azonenberg commented on issue #544: Refactoring: migrate old import stuff to import filters - https://github.com/glscopeclient/scopehal/issues/544#issuecomment-1145657048

07:08 <_whitenotifier-e> [scopehal] azonenberg commented on issue #485: CSV import: detect digital channels and import correctly - https://github.com/glscopeclient/scopehal/issues/485#issuecomment-1145657293

08:10 <_whitenotifier-e> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/glscopeclient/scopehal-apps/compare/899d4bf55225...1ff92c8e7a0b

08:10 <_whitenotifier-e> [scopehal-apps] azonenberg 1ff92c8 - Updated unit test for new MockOscilloscope constructor

20:31 <_whitenotifier-e> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/glscopeclient/scopehal/compare/663cee6f5998...11037202475e

20:32 <_whitenotifier-e> [scopehal] azonenberg 1103720 - DeEmbedFilter: fixed crash when falling back due to clfftBakePlan failure

22:00 <azonenberg> So this is interesting

22:00 <azonenberg> clFFT is giving a "not implemented" error in clfftBakePlan

22:00 <azonenberg> whenever the input is more than 16M points

22:01 <azonenberg> https://github.com/clMathLibraries/clFFT/blob/master/src/library/plan.cpp#L496

22:02 <azonenberg> and this is the root cause

22:02 <azonenberg> The "next release" never came

22:03 <_whitenotifier-e> [scopehal] azonenberg opened issue #623: clFFT cannot handle FFTs of >16M points. Figure out how to handle this - https://github.com/glscopeclient/scopehal/issues/623

22:03 <_whitenotifier-e> [scopehal] azonenberg labeled issue #623: clFFT cannot handle FFTs of >16M points. Figure out how to handle this - https://github.com/glscopeclient/scopehal/issues/623

22:04 <azonenberg> right now we fall back to FFTS, but this is not ideal because the whole point of gpu accelerating is to make things faster

22:07 <azonenberg> i guess a related issue: for de-embedding we currently do the FFT stuff on the entire waveform at once

22:07 <azonenberg> We can probably get better results by chunking the waveform up into, say, 1M point blocks and doing them separately

22:07 <azonenberg> the question being how to handle edge effects

22:07 <azonenberg> we don't want discontinuities when the blocks meet

22:09 <azonenberg> so i'm not sure what the right way of implementing this would be across frequency bands

22:10 <azonenberg> another option to look at is using cuFFT if there's an nvidia card available

22:13 <azonenberg> But either way i want to look at algorithmic optimizations too

22:34 <_whitenotifier-e> [scopehal] azonenberg opened issue #624: Create FFT convolution filter for applying narrowband filters etc - https://github.com/glscopeclient/scopehal/issues/624

22:34 <_whitenotifier-e> [scopehal] azonenberg labeled issue #624: Create FFT convolution filter for applying narrowband filters etc - https://github.com/glscopeclient/scopehal/issues/624

22:35 <_whitenotifier-e> [scopehal] azonenberg opened issue #625: De-embed / channel emulation: look into overlap/add + FFT convolution algorithms for chunking huge datasets into smaller blocks - https://github.com/glscopeclient/scopehal/issues/625

22:35 <_whitenotifier-e> [scopehal] azonenberg labeled issue #625: De-embed / channel emulation: look into overlap/add + FFT convolution algorithms for chunking huge datasets into smaller blocks - https://github.com/glscopeclient/scopehal/issues/625

22:40 <_whitenotifier-e> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/glscopeclient/scopehal/compare/11037202475e...6adfa3874665

22:40 <_whitenotifier-e> [scopehal] azonenberg 6adfa38 - DeEmbedFilter: added detection for excessively long input and fall back to software FFT for now