TODO.md

TODO list

2020-10-20 added

• Change rmm to a simple self written mem pool .

2020-10-16 added

• Change nanomsg to NNG.

2020-10-13 added

• Use Cap'n Proto or flatbuffers instead of Protobuf.

2020-9-25 added

• CMake use boost histogram version aaccording to libboost-dev version.

2020-9-18 added

• Allow user to choose 64bit or 32bit precision

2019-7-10 added

• try multi streamWorker workers

git branch multiThreadMultiStream e3cf1b4a92071e561ceed992cb18147915fd8f20 

• Test std::timed_mutex::try_lock_for vs std::mutex::try_lock

2019-6-18 added

• Use cudaStreamQuery to Query all stream, or cudaStreamAddCallback

2019-3-11 added

• Use streams in GPU.
• Run cudaOccupancyMaxPotentialBlockSize 1 time to get blockSize.
• Use shared memory in kernel to reduce the usage of reg in kernel thread.