Profiling in CULA

General CULA Dense (LAPACK & BLAS) support and troubleshooting. Use this forum if you are having a general problem or have encountered a bug.

Profiling in CULA

Postby temp0712 » Tue Dec 11, 2012 11:02 pm

Hi,

I'm running CULA R16a on top of CUDA 5.0, on Ubuntu 12.04 (kernel 3.2.0.24, gcc 4.6.3). My GPU is an 8600 GT M.

Is there any specific way I can profile my CULA code to find out its running time? I've tried using 2 techniques -

1. cudaEvent_t, cudaEventCreate, cudaEventRecord and cudaEventElapsedTime
2. cutStartTimer, cutStopTimer and cutGetTimerValue

Both of these return the "xyz was not declared in this scope" error (where xyz is the name of the functions I'm trying to use). I'm using the nvcc compiler in the CULA makefile for my project.

I've verified that CUDA is working correctly as profiling is working on standalone CUDA code.

KIndly let me know what could possibly be wrong here.
temp0712
 
Posts: 4
Joined: Thu Dec 06, 2012 10:18 pm

Re: Profiling in CULA

Postby john » Wed Dec 12, 2012 8:13 am

You can use the cut timers or any other CPU timers you'd like for this, but you'll need to integrate them yourself. You can check out examples/benchmark which has our own timer code in it if you'd like.
john
Administrator
 
Posts: 587
Joined: Thu Jul 23, 2009 2:31 pm

Re: Profiling in CULA

Postby temp0712 » Fri Dec 28, 2012 2:40 am

Got it, using gettimeofday() fixed it. Thanks!
temp0712
 
Posts: 4
Joined: Thu Dec 06, 2012 10:18 pm


Return to CULA Dense Support

Who is online

Users browsing this forum: No registered users and 3 guests

cron