Intel® Manycore Testing Lab (Archived)

Find out the C-state of a Processor

Is there a way in Linux to know exactly which state a core or a processor is residing in ?? I saw some pointers to Mointor/MWAIT Instructions. Is there some documentation on these instructions related to power management particularly. May be someone can point to C code where these instructions are used

Per Core Power Measurement / Intel Power Sensors ??

I would like to know if there a way to measure power for any of Intel processors on per core basis. People normaylly use external power analayzers to measure total power for all cores. But not on per-core basis. Are there any "electrical" power sensors in any of microprocessors just like the DTS sensors ???

Job not running

Hello, I am having a problem running a job. I get an 'o' file, which provides details such as user, ncpus, etc. But I do not get any results from my program. My program sorts a file and writes the results to another text file. Any idea why it is not running?

Below is the "qstat -Q" result.



Job id Name User Time Use S Queue ---------------- ---------------- ---------------- -------- - ------

limiting the execution to 32 cores

Hi, so, all the nodes now have 40 cores, great :D but I still want to know if when I state "#PBS -l select=1:ncpus=32", but my application launches 64 threads, what is going to happen? (1) all 40 cores are going to be used (2) the submission system in conjunction with the OS is smart enough to make use of ONLY the 32 cores that specified. thanks, Daniel Nicacio

Insufficient memory for the Java Runtime Environment to continue

Hi All.

I suddenly started seeing this error for all invocations of the Java command that refer to the 1.6 and 1.7 versions. Even a simple "java -version" will give the same problem.

This was not happening earlier today and physical memory should not be a problem:

[ufab@acano01 ~]$ free
total used free shared buffers cached
Mem: 264145020 99303160 164841860 0 165132 94458896
-/+ buffers/cache: 4679132 259465888
Swap: 266207224 0 266207224

Does anyone know what might be wrong (or a workaround for that)?

Thank you!

Andrey

S’abonner à Intel® Manycore Testing Lab (Archived)