* Clarified comments regarding power limit retrieval and its support on virtualized systems.
* Change unsupported comment to UINT32_MAX
---------
Signed-off-by: josnarlo <Joseph.Narlo@amd.com>
Signed-off-by: Arif, Maisam <Maisam.Arif@amd.com>
* Clarified comments regarding power limit retrieval and its support on virtualized systems.
* Change unsupported comment to UINT32_MAX
---------
Signed-off-by: josnarlo <Joseph.Narlo@amd.com>
Signed-off-by: Arif, Maisam <Maisam.Arif@amd.com>
[ROCm/amdsmi commit: 460cfcba1f]
this test will prefetch SVM memory, and then verify the memory is sourced
from the expected numa node.
Signed-off-by: Sunday Clement <Sunday.Clement@amd.com>
- make adding PMIx library to compile time based on the result of
finding PMIx support. This is required eg if compiling rocSHMEM with ompi
4.0/4.1, which do not have a built-in PMIx version.
- when setting USE_EXTERNAL_MPI=OFF which ensures that we do not
check for external MPI libraries (even if one would be available).
[ROCm/rocshmem commit: ed957302d4]
- make adding PMIx library to compile time based on the result of
finding PMIx support. This is required eg if compiling rocSHMEM with ompi
4.0/4.1, which do not have a built-in PMIx version.
- when setting USE_EXTERNAL_MPI=OFF which ensures that we do not
check for external MPI libraries (even if one would be available).
* Add rdhc script in to rocm-core package
* Create the rdhc symlink within the package itself.
* Removed hard-coding of rocm-core name, used CORE_TARGET instead.
* [RDHC] Check if the required pip pkgs are present and warn .
rdhc checks the required pip packages are present or not.
if not warns the user and exits gracefully.
Signed-off-by: Saravanan Solaiyappan <saravanan.solaiyappan@amd.com>
Tools running for sanity checks are
detecting buffer overrun which is
not the case. Still getting rid of
function which is causing the issue
removed and making the code more robust
using defensive programming so that
any tool is not able to detect issues
hereafter. Fixed comments, corrected
typos and added a new return type
which is required as per refactoring
Signed-off-by: Ashutosh Mishra <ashutosh.mishra@amd.com>
* Split ionic code to a subdirectory; dyld libionicl; move the fntable to provider_gda_xxx.hpp
pass the pattr to ionic_setup_pd, include endian.hpp
Enable building IONIC conduit for runtime selection
* Uniform style for the fntable between ionic and the rest
* Move mlx5 gda conduit to a subdir; resolve conflict with backend_can_run
function
* Don't forget to init qp for ionic, move mlx5 specialized init qp code to
the mlx5 subdir
* Don't add cmakecaches...
Typo: GDA_BXNT
* Add gda-ionic to all_backends build scripts
* Apply suggestion from reviews
Co-authored-by: Omri Mor <omri50@gmail.com>
Co-authored-by: Edgar Gabriel <edgar.gabriel@amd.com>
* Remove duplicate definitiion of DLSYM macros
---------
Co-authored-by: Omri Mor <omri50@gmail.com>
Co-authored-by: Edgar Gabriel <edgar.gabriel@amd.com>
[ROCm/rocshmem commit: 3cfe76522e]
* Split ionic code to a subdirectory; dyld libionicl; move the fntable to provider_gda_xxx.hpp
pass the pattr to ionic_setup_pd, include endian.hpp
Enable building IONIC conduit for runtime selection
* Uniform style for the fntable between ionic and the rest
* Move mlx5 gda conduit to a subdir; resolve conflict with backend_can_run
function
* Don't forget to init qp for ionic, move mlx5 specialized init qp code to
the mlx5 subdir
* Don't add cmakecaches...
Typo: GDA_BXNT
* Add gda-ionic to all_backends build scripts
* Apply suggestion from reviews
Co-authored-by: Omri Mor <omri50@gmail.com>
Co-authored-by: Edgar Gabriel <edgar.gabriel@amd.com>
* Remove duplicate definitiion of DLSYM macros
---------
Co-authored-by: Omri Mor <omri50@gmail.com>
Co-authored-by: Edgar Gabriel <edgar.gabriel@amd.com>
* Updated groups printing
* added parameters to check_required_groups
* two device groups since kfd and render require the same group
---------
Signed-off-by: Maisam Arif <Maisam.Arif@amd.com>
Co-authored-by: Maisam Arif <Maisam.Arif@amd.com>
* Updated groups printing
* added parameters to check_required_groups
* two device groups since kfd and render require the same group
---------
Signed-off-by: Maisam Arif <Maisam.Arif@amd.com>
Co-authored-by: Maisam Arif <Maisam.Arif@amd.com>
[ROCm/amdsmi commit: ee1445e2cc]
Query IPC handles on shared memory export/import for any metadata as a
means to uniquely identify handles that happen to be backed by buffers
that point to the same memory.
- Earlier fix PR1788 is no longer necessary after ROCr fix and pre-ROCr fix workaround
- Inserts an s_waitcnt vmcnt(0), which fixes a data corruption issue in MSCCL
[ROCm/rccl commit: 154350baaf]
- Earlier fix PR1788 is no longer necessary after ROCr fix and pre-ROCr fix workaround
- Inserts an s_waitcnt vmcnt(0), which fixes a data corruption issue in MSCCL
* Gate based on ROCM version, safe for ROCm 7.0.2 and beyond.
* Updates naming to gfx9CheapFenceOff since we use this for gfx942 and gfx950. Thanks Nilesh.
* Add info logging statement to NCCL_INIT to print whether enabled when INFO logging is enabled.
[ROCm/rccl commit: c70f5b4621]
* Gate based on ROCM version, safe for ROCm 7.0.2 and beyond.
* Updates naming to gfx9CheapFenceOff since we use this for gfx942 and gfx950. Thanks Nilesh.
* Add info logging statement to NCCL_INIT to print whether enabled when INFO logging is enabled.
- On some hosts the wget can finish too soon and PAPI doesn't catch even a single network event.
- On some hosts, there are multiple default NICs and the scripts didn't work in that case.
- The test script was writing the output of wget to /tmp directory, which causes a problem if another user tries to run the same test. Because the output file with the same name already exists in the same directory, but with a different owner, the test fails
---------
Co-authored-by: David Galiffi <David.Galiffi@amd.com>
Arm Patinyasakdikul
Narlo, Joseph
Sunday Clement
Milan Radosavljevic
Edgar Gabriel
solaiys
Ashutosh Mishra
Venkateshwar Reddy Kandula
Rahul Vaidya
Bindhiya Kanangot Balakrishnan
Aurelien Bouteiller
Pryor, Adam
marantic-amd
David Galiffi
Dimple Prajapati
Young Hui - AMD
vedithal-amd
Alysa Liu
alex-breslow-amd
Saurabh Verma
gilbertlee-amd
Mythreya Kuricheti
Gerardo Hernandez
Danylo Lytovchenko
ajanicijamd
axie_amdeng