[racket] Performance measurement per each thread

If I understand, that result is the ratio of the change in the thread's
process time to the change in real time. For example, if a thread's
process time grows by 500 milliseconds over one second, then that's
50% CPU usage.

At Sat, 17 Nov 2012 20:07:22 +0900, 주대연 wrote:
> Thanks for Matthew.
> 
> I have one more question.
> I want to obtain cpu usage rate you can see it in TaskManager in windows.
> 
> TaskManager shows us a cpu usage per thread.
> I looked the reference manuals in Web Site but I didn't find it.
> 
> So How can I obtain that rate value per thread?
> 
> Thanks for your reply.
> Have a nice day.
> 
> Sent from my iPad
> 
> On 2012. 11. 16., at 오후 10:28, Matthew Flatt <mflatt at cs.utah.edu> wrote:
> 
> > At Fri, 16 Nov 2012 15:33:39 +0900, 주대연 wrote:
> >> Hi I am a student in master course for Computer Engineering.
> >>
> >> I am studying Racket for my research.
> >> My one of goals is to measure the performance(cpu usages, memory, i/o..etc)
> >> of each thread.
> >>
> >> So I have two questions...
> >>
> >> 1) CPU, I/O usages
> >>
> >>    Is there any way to measure CPU usage per each thread in run-time by
> >> Racket?
> >
> > The `current-process-milliseconds' function returns a thread-specific
> > value if you give it a thread argument.
> >
> >> 2) Memory usages
> >>
> >>   Also memory usages per each thread in rum-time??
> >>   As I know Linux doesn't provide the way to measure memory usage per each
> >> thread.
> >>   So How about Racket?
> >
> > You can't ask for thread-specific memory use directly, but you can ask
> > for custodian-specific memory use, and every thread can have its own
> > custodian.
> >
> > Beware of sharing among threads and how that influences memory
> > accounting. For more information about custodian-based memory
> > accounting, see
> >
> >  "Memory Accounting without Partitions"
> >  Wick and Flatt, ISMM'04
> >  http://dl.acm.org/authorize?730730
> >  Addendum:
> >    http://www.cs.utah.edu/plt/publications/ismm04-addendum.txt
> >