brig has many layers of streaming and for regression and performance debugging, we should provide automated benchmarks.
The following things should be measured in throughputs:
Throughput of our internal ipfs cat and ipfs add implementations.
Throughput of the compression/decompression layer.
Throughput of the encryption/decryption layer.
Throughput of TCP socket for the brig cat case.
Throughput of File reading from /tmp for the brig stage case.
Throughput of a simple copy from stdin to stdout 1
1 this is the I/O baseline your CPU can do - that's roughly 32 MB/s for my rather slow laptop.
sahib
commented
3 years ago
brighas many layers of streaming and for regression and performance debugging, we should provide automated benchmarks. The following things should be measured in throughputs:ipfs catandipfs addimplementations.brig catcase.brig stagecase.1 this is the I/O baseline your CPU can do - that's roughly 32 MB/s for my rather slow laptop.