Fixed device numbers for memory APIs.

vzakhari commented 3 years ago

Signed-off-by: Vyacheslav Zakharin vyacheslav.p.zakharin@intel.com

If this pull request is fixing a bug, please link the associated issue. The rest of this template does not apply.

If this pull request is providing a new implementation of the PRKs, please use the following template.

Note that checking all of the boxes is not required.

New PRK implementation checklist

Which kernels are implemented?

[ ] synch_p2p (p2p)
[ ] stencil
[ ] transpose
[ ] nstream
[ ] dgemm
[ ] reduce
[ ] sparse
[ ] branch
[ ] random
[ ] refcount
[ ] synch_global
[ ] PIC
[ ] AMR

Documentation and build examples

If your implementation uses a new programming model that is not ubiquitious (i.e. included in the system compiler on most systems) then you need to provide a link to the appropriate documentation for a new user to install it, etc.

We strongly recommend that you add the appropriate features to make.defs.${toolchain} if appropriate.

Do you certify that your contribution is made in good faith and does not attempt to introduce any negative behavior into this project?

[X ] Yes
[ ] No

jeffhammond commented 3 years ago

I am not suggesting your changes are not correct, but they are not sufficient to cause correct execution on TGL Gen12LP.

Are they working with another HW+SW combination?

jrhammon@tigerlake:~/PRK/C1z$ icx --version
Intel(R) oneAPI DPC++ Compiler 2021.1.2 (2020.10.0.1214)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /opt/intel/oneapi/compiler/2021.1.2/linux/bin

jrhammon@tigerlake:~/PRK/C1z$ LIBOMPTARGET_DEBUG=0 IGC_EnableDPEmulation=1 OverrideDefaultFP64Settings=1 ./nstream-target 10 10
Parallel Research Kernels version 2020
C11/OpenMP TARGET STREAM triad: A = B + scalar * C
Number of iterations = 10
Vector length        = 10
OpenMP Device        = 0
Solution validates
Rate (MB/s): 7.223774 Avg time (s): 0.000044

jrhammon@tigerlake:~/PRK/C1z$ LIBOMPTARGET_DEBUG=0 IGC_EnableDPEmulation=1 OverrideDefaultFP64Settings=1 ./nstream-alloc-target 10 10
Parallel Research Kernels version 2020
C11/OpenMP TARGET STREAM triad: A = B + scalar * C
Number of iterations = 10
Vector length        = 10
OpenMP Device        = 0
Segmentation fault (core dumped)

jrhammon@tigerlake:~/PRK/C1z$ LIBOMPTARGET_DEBUG=0 IGC_EnableDPEmulation=1 OverrideDefaultFP64Settings=1 ./nstream-memcpy-target 10 10
Parallel Research Kernels version 2020
C11/OpenMP TARGET STREAM triad: A = B + scalar * C
Number of iterations = 10
Vector length        = 10
OpenMP Device        = 0
Solution validates
Rate (MB/s): 5.807777 Avg time (s): 0.000055
Segmentation fault (core dumped)

jrhammon@tigerlake:~/PRK/C1z$ LIBOMPTARGET_DEBUG=0 IGC_EnableDPEmulation=1 OverrideDefaultFP64Settings=1 ./nstream-usm-target 10 10
Parallel Research Kernels version 2020
C11/OpenMP TARGET STREAM triad: A = B + scalar * C
Number of iterations = 10
Vector length        = 10
OpenMP Device        = 0
Failed Validation on output array
       Expected checksum: 880.000000
       Observed checksum: 0.000000
ERROR: solution did not validate

This one made my machine unresponsive with arguments 10 $((1024*1024*32)) and I had to power-cycle it to stop the program.

jrhammon@tigerlake:~/PRK/C1z$ LIBOMPTARGET_DEBUG=0 IGC_EnableDPEmulation=1 OverrideDefaultFP64Settings=1 ./nstream-ua-target 4 4
Parallel Research Kernels version 2020
C11/OpenMP TARGET STREAM triad: A = B + scalar * C
Number of iterations = 4
Vector length        = 4
OpenMP Device        = 0
Failed Validation on output array
       Expected checksum: 160.000000
       Observed checksum: 0.000000
ERROR: solution did not validate

vzakhari commented 3 years ago

I did not try it on Gen12LP yet. I used CML GEN11.

nstream-alloc-target: passes with 10 100 and 10 10 nstream-memcpy-target: passes with 10 100, but segfaults with 10 10 nstream-usm-target: does not validate with any input

I am only fixing obvious errors in this PR, and kind of paving the way for more PRs in future.

ParRes / Kernels