Issues while RUNNING PARBOIL's Bencmarks

Accelerated Computing CUDA CUDA Setup and Installation
1

In order to run benchmarks, when i give the command,
oprime@ubuntu:~/parboil$ ./parboil run BENCHMARK cuda default

Where BENCHMARK is replaced by [ bfs, cutcp, lbm, mri-gridding, mri-q, sgemn, spmv, tpacf,histo, sad, stencil ]

IT GIVES TWO KINDS OF ERRORS.

For these benchmarks ( bfs, cutcp, lbm, mri-gridding, mri-q, sgemn, spmv, tpacf )
Here, im showing for “tpacf” benchmark. Same error for others too.

oprime@ubuntu:~/parboil$ ./parboil run tpacf cuda default
Parboil parallel benchmark suite, version 0.2

Cannot find data set

AND
for benchmarks ( histo, sad, stencil )

Parboil parallel benchmark suite, version 0.2

Resolving CUDA runtime library...
	libcudart.so.4 => not found
build/cuda_default/stencil: error while loading shared libraries: libcudart.so.4: cannot open shared object file: No such file or directory
make: *** [run] Error 127
Run failed!

HOW TO TACKLE WITH THESE ERRORS? ANYONE, PLEASE???

2

Hi Oprimoyst,

I am also working with the Parboil benchmark. For your first question you need to run:

./parboil run tpacf cuda small default

You need to put the datasets folder in the parboil folder and it should look like this:

benchmarks/ common/ datasets/ DIRECTORIES driver/ LICENSE parboil* README README.benchmarks README.buildsystem

For the second question, I am having the same issue, so if you found the solution I would appreciate it if you tell me.

3

Hello Oprimoyst and Perruchin,

Did you find the solution to the second issue, error while loading shared mem?

It’d be great if you can let me know how to fix it.

4

Hello,

Sorry I have the same problem as your second problem. Have you found out how to solve that issue?