Hi Fabien, Both these items are on my todo list. 1. Loading the sep4_0 driver 2. Setting perf_event_paranoid to 0 Should get to them sometime next week. Ben
On Jul 22, 2016, at 3:55 PM, Delalondre Fabien Jonathan <[email protected]> wrote:
Thanks All.
So I face a couple of issues. Please let me know if this is coming from me or something else:
On knl00 (configured, SNC4-Cache) I use: compiler (2017): source /soft/compilers/intel_beta/compilers_and_libraries/linux/bin/compilervars.sh intel64 source /soft/compilers/intel_beta/compilers_and_libraries/linux/mpi/bin64/mpivars.sh vtune (2016 v4): source /soft/compilers/intel/vtune_amplifier_xe_2016.4.0.470476/amplxe-vars.sh
When I execute the an MPI+OpenMP application 1) It works (running basic vtune hot spot analysis) amplxe-cl -collect hotspots -- mpirun -n $NODES <EXEC> <ParamEXEC>
2) It used to work in May (with Vtune patched NDA - now it returns an error) amplxe-cl -c memory-access -knob analyze-mem-objects=true -knob sampling-interval=10 -data-limit=0 -d 60 -- mpirun -n $NODES <EXEC> <ParamEXEC>
amplxe: Error: Cannot enable Hardware Event-based Sampling: problem with the driver (sep*/sepdrv*). Check that the driver is running and the driver group is in the current user group list. See "Building and Managing the Sampling Driver" help topic for further details.
3) Instructions provided on wiki by Vitali do not work for me due to lack of permission (set to 1 in X file)
mpirun -n $NODE_COUNT -env KMP_AFFINITY=balanced,granularity=fine -gtool "/soft/restricted/intel-vtune-nda/vtune_amplifier_xe/bin64/amplxe-cl -collect memory-access -knob sampling-interval=10 -knob analyze-mem-objects=true -data-limit=1500 -mrte-mode=native -result-dir=Result4:0-3" <EXEC> <EXEC_PARAM>
Error message: amplxe: Error: Not enough permissions to run uncore events collection. Consider setting the /proc/sys/kernel/perf_event_paranoid value to 0 or less.
Any hint ? Can we change the permissions to run uncore events collection on all KNL? Let me know if I am doing something wrong.
Thanks,
Fabien
________________________________________ From: Allen, Benjamin S. <[email protected]> Sent: Friday, July 22, 2016 3:15 PM To: Delalondre Fabien Jonathan Cc: [email protected]; Kumaran, Kalyan Subject: Re: [JLSE-Intel-NDA] Last version of compiler & Vtune working with memory analysis
VTune 2016 Update 4 added KNL support, so the NDA'ed add-on bits are no longer needed. The compilers in 2017 and 2016 haven't been updated recently, so they should be the same as when you last worked with them.
Ben
On Jul 22, 2016, at 7:15 AM, Kumaran, Kalyan <[email protected]> wrote:
You can check, but I think Release 16 is a sure bet and had full KNL support.
-Kumar
On Jul 21, 2016, at 6:00 PM, Delalondre Fabien Jonathan <[email protected]> wrote:
Hi All,
Sorry to ask but ... Can someone confirm which version of the compiler and Vtune work fine together for the data memory analysis.
I double check Vitali's comments on the wiki but I have been told everything should work fine with the latest 2017 versions.
Thanks !
Fabien _______________________________________________ Intel-nda mailing list [email protected] https://lists.jlse.anl.gov/mailman/listinfo/intel-nda
_______________________________________________ Intel-nda mailing list [email protected] https://lists.jlse.anl.gov/mailman/listinfo/intel-nda