OK, I've been tearing my hair out all morning over this stupid sp vs mp thing. I've unified the code so that the sp version is doing exactly the same work as the mp version. mp is still faster until you reduce the effort per chunk to virtually nil, when the cost of the context switch takes over. Changing the amount of optimisation doesn't help either. The only difference I can see is that when the mp code is running most of the cpu time is marked as "system" in my load meter, is signal'd code run in a higher priority or something? The only other thing I considered is that the cost of context switching a larger process is disproportionately high, but I tried making the sp process smaller by removing all but one of the processing functions and it didn't make enough difference to explain it. Argh! - Steve
- Re: [linux-audio-dev] Re: high level laaga Steve Harris
- Re: [linux-audio-dev] Re: high level laaga Paul Davis
- Re: [linux-audio-dev] Re: high level laaga Steve Harris
- Re: [linux-audio-dev] Re: high level laaga Paul Davis
- Re: [linux-audio-dev] Re: high level laaga Steve Harris
- Re: [linux-audio-dev] Re: high level laaga Steve Harris
- Re: [linux-audio-dev] Re: high level laaga Paul Davis
- Re: [linux-audio-dev] Re: high level ... Steve Harris
- Re: [linux-audio-dev] Re: high level laaga Paul Davis
- Re: [linux-audio-dev] Re: high level laaga Steve Harris
- Re: [linux-audio-dev] Re: high level ... Steve Harris
- Re: [linux-audio-dev] Re: high level ... Paul Davis
- Re: [linux-audio-dev] Re: high level laaga Jim Peters
- Re: [linux-audio-dev] Re: high level laaga Steve Harris
- [linux-audio-dev] Re: high level laaga Paul Davis
- [linux-audio-dev] Re: high level laaga Abramo Bagnara
- Re: [linux-audio-dev] Re: high level laaga Abramo Bagnara
- Re: [linux-audio-dev] Re: high level laaga Paul Davis
- [linux-audio-dev] costs of IPC Paul Davis
- [linux-audio-dev] Re: costs of IPC Abramo Bagnara
- [linux-audio-dev] Re: costs of IPC Paul Davis
