* Support pipelining codegen and template specialization
* Support ReduceCopy pipelining for AllReduce, ReduceScatter, and Reduce (currently enabled for bfloat16)
* Remove need for FUNC_INDEX_TOTAL
* Add pipeline field to device function key construction logic
* Avoid unneeded codegen for LL/LL64 kernels
* Modify conditions and add pipeline dtypes env
* Optimize selection for both gfx942 and gfx950
* Increase pipeline bitfield width
* Use __forceinline__ for all device functions
* Realign reduceCopy with original form
* Add opt-out option to enable perf debugs
* Remove force-reduce-pipelining option from README
* Update CHANGELOG.md
---------
Co-authored-by: Jeffrey Novotny <jnovotny@amd.com>
* add direct allgather algorithm
* minor fix
* add debug print for memory allocation tracker
* add message size threshold for direct allgather
* scatter transfers across ranks
* update changelog
* minor fix
* Update CHANGELOG.md
Co-authored-by: Jeffrey Novotny <jnovotny@amd.com>
* enable direct AG when pxn is ON on MI300X or MI350
---------
Co-authored-by: Jeffrey Novotny <jnovotny@amd.com>
- Updated ncclDevFuncId to use a hash-based lookup with std::unordered_map.
- Keys are now 64-bit integers, which pack coll, algo, proto, devRedOp, and type fields.
- Improved flexibility and maintainability by moving away from row-based indexing.
- Added error handling for missing keys in the hash map.
- Aligned key generation logic with generate.py and updated generate.py.
* change gfx950 algo/proto selection for multinode allreduce, allgather, reduceScatter
* gfx950 tuning: enable tuning for broadcast, allreduce starts LL128 earlier and switches to ring earlier, change LL128 start for allgather and reduceScatter
* lower LL128 threshold
* update reduceScatter LL128 min to match LL max for consistency
* enable multinode PXN and increase chunksize for gfx950
* change LL128 start to 128KB, adjust ring-start according to node-count
* disable code-path for fused-AR on LL128 for gfx950
* use LL128 starting from 1KB for multinode allgather on gfx950
* start LL128 earlier for multinode reduceScatter on gfx950
* start LL128 earlier for multinode broadcast on gfx950
* set multinode allreduce to start simple on 64MB for gfx950
* start LL128 from 1KB for multinode broadcast on gfx950
* setting multinode AR to use tree instead of ring at 16MB, 64MB, 128MB
* set multinode broadcast to use LL for up to 256KB depending on node-count for gfx950
* adjust algo for 32MB multinode allreduce on gfx950
* make 32MB tree LL128 for multinode AR on gfx950
* make sure ring is not picked on 2N allreduce on small sizes
* Added useAcc as a template parameter to address the 2% performance regression in allreduceWithBias
---------
Co-authored-by: Marzieh Berenjkoub <mberenjk@amd.com>
* Support fused all reduce and elementwise operations
Add additional "acc" parameter to RCCL Replayer logs
Add flag which indicates availability of new API
* Fix Recorder json parsing
* Remove unreachable code
* Remove extra acc pointer check
* .
* Revert "[DEVICE] Adding ability to choose unroll factor at runtime (#1734)"
This reverts commit 9d72be7b2f.
* Use noinline to reduce kernels linking time
* Don't use noinline for gfx942 and gfx950 to avoid perf regression
---------
Co-authored-by: AtlantaPepsi <timhu102@amd.com>
Co-authored-by: BertanDogancay <bertan.dogancay@gmail.com>
Boosts single node bfloat16 allreduce performance by up to 20% for some data sizes and provides gating with the RCCL_GFX942_CHEAP_FENCE_OFF environment variable
* Add support for extended fine-grained system memory pool
* Use hipHostRegisterUncached
* Add "sc0 sc1" flags for LL store on gfx950
* Update after HIP flag is changed to hipExtHostRegisterUncached
This feature tracks the proxy events and status of each send/recv op. ProxyTrace keeps a fixed number of active ops in host mem and dumps the status of each op when the program crashes or hangs.
* First version of new replayer, with comments on future TODOs
* plus minor fixes for UT
* Updated format of recorder, especially in binary department, according to replayer's need
* Reapply "[AG and RS channel tuning] Add thread work threshold to tuning models and precompute reg index in LL128 (#1641)"
This reverts commit 943ad6f7820739385a0b54e81f823d0df1dbf71c.
* Decreasing NCCL_LL128_SHMEM_ELEMS_PER_THREAD from 16 to 8
* Revert "Revert "replacing rccl_float8 with hip_fp8 and address compatibility …"
This reverts commit 824b81c034.
* [UT] Modify max stack size to 496
* adding a check for OCP type and replacing ROCM_VERSION with HIP_VERSION
* addressing the ci failure
* Adding the device tag
---------
Co-authored-by: Marzieh Berenjkoub <mberenjk@amd.com>
* Update LL128 elems per thread
* Precompute ix[g] in LL128 prim
* Make Threadthreshold part of tuning models
* Ignore channel tuning when channels are env controlled
* Tune LL128 max limit for AG
* Tune LL128 max limit for RS
* Retune AR LL128 limits due to changes
* Update CHANGELOG.md
---------
Co-authored-by: Jeffrey Novotny <jnovotny@amd.com>
This commit handles DMABUF initialization and call appropriate handling function. This fixes crash in OS with no peermem support and relying on only DMABUF.
* Initial test commit
* Handling Dmabuf_fd opening and closing
* Cleanup
* Use DMABuff or Peermem as needed
* Using user input for ibDmaBufSupportInitOnce
* Revert all changes to rocmwrap.cc
* Revert all changes to rocmwrap.cc
* Changing to func definition braces
* Reverting line removal in utils.h
* useDmaBuf to calculate flushEnabled
* Internal RCCL/NCCL functionality exposed when RCCL_EXPOSE_STATIC is enabled
* Algo/protocol/max channels can be obtained with the new RCCL API
* Introduce rccl_static and rccl_static_inline macros to work around invisible functions in core source files like enqueue.cc
* Add usage example in topo-explorer tool
Minimize the performance impact of the device kernel profiling support when
the profiler plugin is not loaded.
Reduce the overheads of CUDA graph capturing, which increased in NCCL
2.26.2 for large graphs.
Fix the exchange of enhanced connection establishment (ECE) options to
address potential slowdowns on networks utilizing RoCE.
Test if cuMem host allocations work and if not, disable them. Enabled by
default since NCCL 2.24 if the CUDA driver version is at least 12.6, such
allocations rely on NUMA support, which is by default not available under
Docker. We recommend invoking Docker with "--cap-add SYS_NICE" to enable
it.
Fix an initialization error when running with NCCL_NET_GDR_C2C=1 on
multiple MNNVL domains with non-uniform network configurations across
nodes.
Fix the printing of sub-seconds in the debug log when using a custom
NCCL_DEBUG_TIMESTAMP_FORMAT setting.
* Enabling LL128 by default on MI300
* Add missing CUDACHECK
* Adjust BW correction factors to fix the Tree->Ring switching point
* Refactor and add ll128 AR logarithmic factor to tuning models
* Move RCCL tuning changes to a separate file
* Use enum for tunable indexing
* Use explicit indexing in tuning models to avoid mismatch issues
* Place rcclGetSizePerRank in a function
* Remove HIP ifdef for rccl-only call
---------
Co-authored-by: Mustafa Abduljabbar <mustafa.abduljabbar@amd.com>
Nusrat Islam
Mustafa Abduljabbar
isaki001
mberenjk
Karthikeyan Arumugam
Avinash
ycui1984
Atul Kulkarni
Wenkai Du
alex-breslow-amd
Nilesh M Negi
Bertan Dogancay
ryanhankins
Dingming Wu
BertanDogancay
Sarat Kamisetty
Tim
Richard Barnes
Arm Patinyasakdikul
corey-derochie-amd
Pedram Alizadeh
Kamil Iskra