On Sun, 2007-09-16 at 10:30 -0500, Nate Straz wrote:
> On Sun, Sep 16, 2007 at 08:49:14PM +0530, Subrata Modak wrote:
> > I have observed that ./runltp (and pan for that matter) does not create
> > multiple instances of testcases to run if -x option is specified. I am
> > also sceptical how the results/output are logged when -x is specified.
>
> That's not the intent of -x N. The intent it to keep at least N tags
> running at any one time. The results are sent to the log file and
> output file directly or buffered in a temporary file then flushed when
> the tag completes.
Thanks, i found out the same.
>
> To run that tags more than once, you need to use the -i M option, then
> pan will run each tag M times. Although the algorithm will complete all
> tags once before starting the next iteration.
>
> Nate
But i think, -i M through runltp will actually:
1) try to fork 1 genload process spinning indefinitely on sync(), and,
2) try to fork 1 genload process spinning indefinitely on write() to
write M MB of random data to a temporary file.
Thatś what line no.s 189 and 190 of runltp specify.
Now, regarding this (genload: ltp/tools/genload/stress.c), i have sent a
mail earlier. If you see what <genload --help> displays:
[EMAIL PROTECTED] ltp-full-20070831]$ ./tools/genload/stress
--help
`stress' imposes certain types of compute stress on your system
Usage: stress [OPTION [ARG]] ...
-?, --help show this help statement
--version show version statement
-v, --verbose be verbose
-q, --quiet be quiet
-n, --dry-run show what would have been done
--no-retry exit rather than retry non-critical errors
--retry-delay n wait n us before continuing past error
-t, --timeout n timeout after n seconds
--backoff n wait for factor of n us before starting work
-c, --cpu n spawn n procs spinning on sqrt()
-i, --io n spawn n procs spinning on sync()
-m, --vm n spawn n procs spinning on malloc()
--vm-chunks c malloc c chunks (default is 1)
--vm-bytes b malloc chunks of b bytes (default is 256MB)
--vm-hang hang in a sleep loop after memory allocated
-d, --hdd n spawn n procs spinning on write()
--hdd-noclean do not unlink file to which random data written
--hdd-files f write to f files (default is 1)
--hdd-bytes b write b bytes (default is 1GB)
Infinity is denoted with 0. For -m, -d: n=0 means infinite redo,
n<0 means redo abs(n) times. Valid suffixes are m,h,d,y for time;
k,m,g for size.
I think we do not exploit the entire set of such options through runltp.
Now, we need to do that too.
--Subrata--
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Ltp-list mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/ltp-list
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
