Revise IPC benchmark and add a new benchmark on multiplexing.
o Changed IPC benchmark clients to add throughput measurement and
share it for benchmarks on multiplexing.
o Added a stand-alone benchmark binary bm/main.go for clean histogram
outputs. (With many benchmarks, it is very hard to read the
benchmark results when printing histogram in each benchmark
trials.)
o Added the latest benchmark results to RESULTS.txt
Note:
o There will be one follow-up CL to add a new benchmark to bmclient/bmserver.
o I'm not sure when we got the BM results in README.txt, but a new BM results
shows that the current IPC performance is 2 ~ 4X slower than before even
with a better machine.
Change-Id: I66942c33a849af2b37990d5368ede85cdb9b76cc
6 files changed