get the execution time in nvprof

Question

Is there a way to get the kernel execution time in nvprof like for a metric?

for example, to get the dram read transactions I type:

nvprof --metrics dram_read_transactions ./myprogram

My question is: is there something like

nvprof --metrics execution_time ./myprogram

I would like to collect a small set of metrics in one command line instead of having to use

nvprof ./myprogram

as a separate command.

Answer 1

I believe you are looking for: nvprof --print-gpu-trace ./myprogram

Answer 2

You should read this post on nVIDIA's "CUDA Pro Tip" blog:

CUDA Pro Tip: nvprof is Your Handy Universal GPU Profiler

It walks you through some basics of how to use nvprof to profile and time your application. Specifically, if you write something like:

nvprof --print-gpu-trace ./nbody --benchmark -numdevices=2 -i=1

(the example is for an n-body physics problem simulator), your output will include something like the following:

...
==4125== Profiling application: ./nbody --benchmark -numdevices=2 -i=1
==4125== Profiling result:
   Start  Duration            Grid Size      Block Size     Regs*    SSMem*    DSMem*      Size  Throughput           Device   Context    Stream  Name
260.78ms     864ns                    -               -         -         -         -        4B  4.6296MB/s   Tesla K20c (0)         2         2  [CUDA memcpy HtoD]
260.79ms     960ns                    -               -         -         -         -        4B  4.1667MB/s  GeForce GTX 680         1         2  [CUDA memcpy HtoD]
260.93ms     896ns                    -               -         -         -         -        4B  4.4643MB/s   Tesla K20c (0)         2         2  [CUDA memcpy HtoD]
260.94ms     672ns                    -               -         -         -         -        4B  5.9524MB/s  GeForce GTX 680         1         2  [CUDA memcpy HtoD]
268.03ms  1.3120us                    -               -         -         -         -        8B  6.0976MB/s   Tesla K20c (0)         2         2  [CUDA memcpy HtoD]
268.04ms     928ns                    -               -         -         -         -        8B  8.6207MB/s  GeForce GTX 680         1         2  [CUDA memcpy HtoD]
268.19ms     864ns                    -               -         -         -         -        8B  9.2593MB/s   Tesla K20c (0)         2         2  [CUDA memcpy HtoD]
268.19ms     800ns                    -               -         -         -         -        8B  10.000MB/s  GeForce GTX 680         1         2  [CUDA memcpy HtoD]
274.59ms  2.2887ms             (52 1 1)       (256 1 1)        36        0B  4.0960KB         -           -   Tesla K20c (0)         2         2  void integrateBodies(vec4::Type*, vec4::Type*, vec4::Type*, unsigned int, unsigned int, float, float, int) [242]
274.67ms  981.47us             (32 1 1)       (256 1 1)        36        0B  4.0960KB         -           -  GeForce GTX 680         1         2  void integrateBodies(vec4::Type*, vec4::Type*, vec4::Type*, unsigned int, unsigned int, float, float, int) [257]
276.94ms  2.3146ms             (52 1 1)       (256 1 1)        36        0B  4.0960KB         -           -   Tesla K20c (0)         2         2  void integrateBodies(vec4::Type*, vec4::Type*, vec4::Type*, unsigned int, unsigned int, float, float, int) [275]
276.99ms  979.36us             (32 1 1)       (256 1 1)        36        0B  4.0960KB         -           -  GeForce GTX 680         1         2  void integrateBodies(vec4::Type*, vec4::Type*, vec4::Type*, unsigned int, unsigned int, float, float, int) [290]

which is a timing of all of your kernels.

It's also useful to run nvprof --help and spend 5-10 minutes reading through the options; for example, you'll find the switch for printing your trace in CSV format if you want to process it in a script.