search for: undercount

...ng metric could include disparity between fields: hot fields interleaved with cold fields, with a total access count high enough to matter. There are a few potential problems with this simple approach: - Overcount: a GEP instruction does not necessarily mean a field access. - Undercount: a GEP instruction may lead to multiple field accesses, especially if the address is passed to another function as an argument. - Racy update by multiple threads. We want to keep the instrumentation simple in our initial implementation for both robustness and performance reasons, so we w...

...old fields, with a total access count high > > enough to matter. > > > There are a few potential problems with this simple approach: > > > * > > > Overcount: a GEP instruction does not necessarily mean a field > > access. > > > * > > > Undercount: a GEP instruction may lead to multiple field accesses, > > especially if the address is passed to another function as an > > argument. > You can ignore my message; I misread the distinction here. -Hal > > * > > I can't tell from your description, but it sound...

...s: hot fields interleaved with cold > fields, with a total access count high enough to matter. > > > There are a few potential problems with this simple approach: > > - > > Overcount: a GEP instruction does not necessarily mean a field access. > - > > Undercount: a GEP instruction may lead to multiple field accesses, > especially if the address is passed to another function as an argument. > > > > I changed my mind; don't ignore what I said. But let me propose a > slightly-different design: > > 1. Just as you've describe...

FYI, I posted a blog a few days ago about a DTrace provider for NFS that is currently in development: http://blogs.sun.com/roller/page/samf?entry=a_dtrace_provider_for_nfs Let''s discuss any questions, comments, etc. here. I also advertised this on nfs-discuss at opensolaris.org. Naturally, I would expect the discussion here to be more on the specifics of DTrace, and the