monitoring large numbers of short-lived threads

[jmellorcrummey]

Profiling and tracing large numbers of short-lived threads is suboptimal. Currently, we generate one trace file per short-lived thread. Each such thread consumes a new line in a visualization. Perhaps we should collapse things in some way ...
DMRG-trace.pdf

[jmellorcrummey]

this is handled by laksono's many threads implementation