Edgar Gabriel c7297b1d6b Performance tuning for inter-node communication (#66) ...

This PR addresses two issues:
 - reduce the number of contexts supported by the host-interface by
   default to 1, we are not using those at the moment, and hence
   we now create fewer MPI_Win at the startup
 - introduces a micro-sleep in RO progress engine in case there are no
   pending requests. This leads significant performance improvements
   observed for inter-node communication with THor2 NICs.

[ROCm/rocshmem commit: 12561783de]

2025-03-26 21:09:26 -05:00

..

containers

Add AtomicWFQueue implementation and tests (#62)

2025-03-25 00:45:19 -05:00

host

Performance tuning for inter-node communication (#66)

2025-03-26 21:09:26 -05:00

ipc

add rocshmem_barrier() (#61)

2025-03-24 11:23:03 -05:00

memory

Fix/RO Backend Hang Issue (#53)

2025-03-24 22:54:07 -05:00

reverse_offload

Performance tuning for inter-node communication (#66)

2025-03-26 21:09:26 -05:00

sync

Update(DeviceProxy): Dynamically Determine Memory Allocation Size & Remove Compile-Time size Calculations (#48)

2025-02-24 15:11:46 -06:00

assembly.hpp

Remove support code for GFX940 and GFX941 targets (#55)

2025-03-21 14:31:49 -05:00

atomic_return.cpp

Transfer files from RAD repository

2024-07-01 09:57:08 -05:00

atomic_return.hpp

Transfer files from RAD repository

2024-07-01 09:57:08 -05:00

atomic.hpp

Add missing header file

2024-09-10 09:35:02 -07:00

backend_bc.cpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

backend_bc.hpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

backend_type.hpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

CMakeLists.txt

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

constants.hpp

Remove comparisons of signed to unsigned values

2024-12-12 10:21:08 -06:00

context_device.cpp

add rocshmem_barrier() (#61)

2025-03-24 11:23:03 -05:00

context_host.cpp

Rename config.h to roc_shmem_config.h

2024-12-06 01:08:13 +00:00

context_incl.hpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

context_tmpl_device.hpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

context_tmpl_host.hpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

context.hpp

add rocshmem_barrier() (#61)

2025-03-24 11:23:03 -05:00

device_proxy.hpp

Refactor RO backend data structures (#49)

2025-03-14 10:49:44 -05:00

fence_policy.hpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

hdp_policy.hpp

Rename config.h to roc_shmem_config.h

2024-12-06 01:08:13 +00:00

hdp_proxy.hpp

Update(DeviceProxy): Dynamically Determine Memory Allocation Size & Remove Compile-Time size Calculations (#48)

2025-02-24 15:11:46 -06:00

ipc_policy.cpp

Updated IPC detection logic (#51)

2025-03-17 11:36:11 -04:00

ipc_policy.hpp

Sync Reverse Offload Scripts (#52)

2025-03-19 14:31:07 -04:00

mpi_init_singleton.cpp

Set MPI threading mode for hyd-7c machine

2024-07-11 08:43:01 -07:00

mpi_init_singleton.hpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

rocshmem_calc.hpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

rocshmem_gpu.cpp

add rocshmem_barrier() (#61)

2025-03-24 11:23:03 -05:00

rocshmem.cpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

stats.hpp

Remove rocshmem_timer function

2025-02-17 17:10:51 +00:00

team_tracker.cpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

team_tracker.hpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

team.cpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

team.hpp

Removed GPU_IB (#59)

2025-03-24 09:04:52 -04:00

templates_host.hpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

templates.hpp

Use new naming scheme

2024-11-25 14:25:29 -06:00

util.cpp

Performance tuning for inter-node communication (#66)

2025-03-26 21:09:26 -05:00

util.hpp

Performance tuning for inter-node communication (#66)

2025-03-26 21:09:26 -05:00

wf_coal_policy.cpp

Transfer files from RAD repository

2024-07-01 09:57:08 -05:00

wf_coal_policy.hpp

Rename config.h to roc_shmem_config.h

2024-12-06 01:08:13 +00:00