GitLab

Signed-off-by: Tangao <2205747538@qq.com>

Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Add stream synchronization before destroying MPI communicator (#1979)
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* feat: Implement column-wise userbuffer overlap for comm+GEMM operations

Add support for overlapping column-wise allgather communication with GEMM
operations to improve training performance:

* **Core infrastructure changes:**
  - Update bulk_overlap_columnwise_ag() to accept explicit stream parameter
  - Modify userbuffers send/recv loops to use rank-ordered iteration
  - Add userbuffers_send_all/recv_all function declarations

* **Python integration:**
  - Add bulk_overlap_ag_with_external_gemm() C++ extension function
  - Expose new overlap function via pybind11 bindings
  - Update overlap method configurations to include more ring_exchange ops

* **LayerNorm MLP optimization:**
  - Enable column-wise quantization for FC2 gradient output
  - Implement overlap of allgather communication with FC2 DGRAD GEMM
  - Use fill_userbuffers_buffer_for_all_gather for efficient buffering

This optimization allows overlapping communication and computation phases
more effectively, reducing training wall-clock time by hiding allgather
latency behind GEMM execution.
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Working userbuffer overlapping API
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Fix overwriting bulk overlap UB object for layernormLinear
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Update external overlap to use tp size instead of nvsize to determine number of copies
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Fix linter error
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Explanatory comments of overlap logic
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Fix the UB fused ops tests
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* fix: Fix linter errors
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

---------
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* fix current device for cuDNN/cuBLAS handles
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* add unit test
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* use weight device and improve tests
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



---------
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>

* minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>

* refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>

* Merge remote-tracking branch 'origin/main' into save_bf16_in_fp8_gemm
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* decouple linear bwd with save_original_input; clean up UTs
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



---------
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
Signed-off-by: hx <hongxiaob@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Xin Yao <xiny@nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

Signed-off-by: wenjh <wenjh@sugon.com>

* Replace `is_float8_tensor` with `is_quantized_tensor`

Replace free function `is_float8_tensor` with `is_quantized_tensor` in `_common.py` and use it throughout the `ops` codebase to check if a tensor is a (possibly internal) quantized tensor
Signed-off-by: Jan Bielak <jbielak@nvidia.com>

* Pass next and previous op quantizers directly to op_forward and fuser_forward

Change interface of `fuser_forward` and `op_forward` to no longer take preceding and following ops and instead take the following op's input quantizer and preceding op's input gradient's quantizer directly
Signed-off-by: Jan Bielak <jbielak@nvidia.com>

* Remove use redundant `detach` in `BasicLinear`

Remove use of `detach` in `BasicLinear` for improved performance (enabled by not passing prev_op to backward)
Signed-off-by: Jan Bielak <jbielak@nvidia.com>

* Handle saving internal tensors

Handle saving internal tensors in `_OperationFuserAutogradFunction` using `prepare_for_saving` and `restore_from_saved`
Signed-off-by: Jan Bielak <jbielak@nvidia.com>

* Use internal tensors

Enable use of internal tensors in `BasicLinear` quantizers and fix issues resulting from internal tensors not having methods that regular tensors have
Signed-off-by: Jan Bielak <jbielak@nvidia.com>

* Apply suggestions from code review
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Jan Bielak <jbielak@nvidia.com>

---------
Signed-off-by: Jan Bielak <jbielak@nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Use FP16 tols for tests with TF32
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Use uniform init instead of constant init
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Revert constant init test, but reduce value
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Add FP8 current scaling to te.Sequential tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Helper function for test/ref tensors does not produce quantized tensor by default
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add FP8 current scaling to distributed te.Sequential tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add FP8 current scaling to Userbuffers te.Sequential tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Debug MXFP8 tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* Add support for overlapping wgrad NCCL AG with dgrad GEMM
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* Remove unused wait on memcpy API from UB
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

* Add better commenting to MXFP8 overlap
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>

---------
Signed-off-by: djns99 <40156487+djns99@users.noreply.github.com>
Co-authored-by: dastokes <dastokes@dastokes-dvt-01.nvidia.com>

[PyTorch] FP8 Subchannel Recipe With FP8 Gather And Configurable Scaling Factor Tensor Swizzling (#1707)

* functional kernel for columnwise + no-transpose option, still hacky
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* pass all quantizer unit tests
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* refactor, add gemm ready api
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* make format options private members, simplify api
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* swizzle scales right before gemm
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* bug fix of single layer test
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* attempt to fix lint issue
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fp8 gather pass, need minor refine
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix return_layernorm_output_gathered case
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* remove special cases, add sanity check before gemm
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* fix lint
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* lint ungrouped imports
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Implement dequantize for compact 1D blocks.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* add more unit test with dequantize compact supported
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* lint again
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* make ag for subchannel respect async
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* zero tolerance in distributed test
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* fix zero tolerance test
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* resolve rebase issues
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* lint & format
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* fix lint
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* clean up
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* bug fix
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* relax rtol for fp32 distributed test
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* fix some ci issue
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* fix ci test failure in debug mode
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* Force row-wise and column-wise data to have same data format

Prototype "all-gather usage" in quantizer.
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Remove dead logic for high-precision AGs
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Debug FP8 block-wise tensor tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug distributed test
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Handle case where LayerNormLinear returns gathered norm output
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* fix debug mode
Signed-off-by: zhongboz <zhongboz@nvidia.com>

---------
Signed-off-by: zhongboz <zhongboz@nvidia.com>
Signed-off-by: Keith Wyss <kwyss@nvidia.com>
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Keith Wyss <kwyss@nvidia.com>
Co-authored-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

Use ocp fp8.
Workaround: test_cast_float8blockwise.cu link wrong std::max
Signed-off-by: wenjh <wenjh@sugon.com>

* Modify the test cases
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Make the tests reproducible on different machines
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Fixed the cache of the gamma_in_weight_dtype setting
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Reinstate the tests
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* More verbose code and comments
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

---------
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* Fix split_overlap_rs aggregate=True chunk offset calculation
Signed-off-by: Guyue Huang <guyueh@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add unit test for aggregate=True
Signed-off-by: Guyue Huang <guyueh@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Fix unit test
Signed-off-by: Guyue Huang <guyueh@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



---------
Signed-off-by: Guyue Huang <guyueh@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* tests drop
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* move dir
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* tests fox
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



---------
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Przemek Tredak <ptredak@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Initial work toward restoring UB support in te.Sequential
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Forward UB linear runs, but has numerical error
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug UB forward tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Minor tweaks
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Remove Python checks for MXFP8 UB linear forward
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add dim check for MXFP8 full tiles
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Move QuantizedTensor logic out of UB comm and into Python helper function
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Support MXFP8 AGs
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Coalesce NCCL all-gathers for MXFP8 all-gather
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Initial impl of backward UB linear in te.Sequential
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug UB linear backward with no quantization

dgrad GEMM + dx RS is still broken.
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Fix chunk dims for dgrad GEMM + dx RS
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debugging MXFP8 UB cases

Still failing with dy AG + wgrad GEMM
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Use NCCL to overlap dy AG with dgrad GEMM
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug UB GEMM tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Initial refactoring of linear module forward
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Refactor linear module backward
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug linear module UB tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Tweak test tensor dims
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Do not store autograd context within wgrad GEMM closure
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Fix linter warnings
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Update LayerNormLinear
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Update LayerNormMLP
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug UB tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Fix linter warnings
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Debug test failures
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Minor style tweaks
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Fix incorrect usage for GEMM input with block-scaled FP8
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Fix RS out dims
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Disable dgrad GEMM + UB AG + NCCL AG overlapping
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Disable dgrad GEMM + UB AG + NCCL AG overlap in te.Sequential
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Restore support for internal quantized tensors
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add tests for MXFP8 GEMM with UB
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Fix linter warnings
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug test failures
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Debug test failures
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* Changes to Linear
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Fix
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Removing unnecessary check
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Relax the absolute tolerance in FP32 distributed test
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add QuantizedTensorBase class
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* Change the blockwise tensor.
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* A little cleaning
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Fixes
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>

---------
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Check if FP8 block scaling is supported in tests
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Update tests/pytorch/distributed/test_cast_master_weights_to_fp8.py
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Add fp8_primary_weights support for blockwise scaling
Signed-off-by: kunlunl <kunlunl@nvidia.com>

custom fsdp
Signed-off-by: kunlunl <kunlunl@nvidia.com>

[pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



Add view to blockwise fp8 tensor
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Fix columnwise_shape in backward of view()
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Add comments to the unit of start_offset
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Add test for view and reshape for blockwise fp8 tensor
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Add implementation for self._columnwise_scale_inv is not existed
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Track down checks for _columnwise_data is None and adding checks for  _columnwise_invalid
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add assertion to check whether ._quantizer is None
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* rename partial_cast.cu -> fp8_block_scaling_partial_cast.cu
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* rename partial_cast kernel to fp8_block_scaling_partial_cast kernel
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add shfl_sync in partial cast kernel
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Remove columnwise_invalid flag
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add comments about out-of-bounds write
Signed-off-by: kunlunl <kunlunl@nvidia.com>

---------
Signed-off-by: kunlunl <kunlunl@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* split wgrad for GroupedLinear
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* support wgrad split for linear and ln_linear
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* add comments and fix WeightGradStore
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* support bias and fix unit tests
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* minor fix
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* support fuse_grad_accumulation=false
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* add wgrad split for layernorm_mlp
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* minor fix
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix unittest
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* add unittest for distributed interface apply Dener's suggestion
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* minor fix
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* replace split_bw with delay_wgrad_compute
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Update transformer_engine/pytorch/module/layernorm_mlp.py
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Update transformer_engine/pytorch/module/linear.py
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Update transformer_engine/pytorch/module/layernorm_linear.py
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* remove comments
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>

---------
Signed-off-by: Hongbin Liu <hongbinl@nvidia.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Hongbin Liu <hongbinl@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Add fp8 weight transpose cache check in backward, and regenerated it if it does not exist
Signed-off-by: jianbinc <shjwudp@gmail.com>

* Properly handle fsdp shard model weight input.
Signed-off-by: jianbinc <shjwudp@gmail.com>

* move Float8Tensor to QuantizedTensor in cast_master_weights_to_fp8 UT
Signed-off-by: jianbinc <shjwudp@gmail.com>

* handle Float8TensorBase issue
Signed-off-by: jianbinc <shjwudp@gmail.com>

* fix bug in activation recompute
Signed-off-by: jianbinc <shjwudp@gmail.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



---------
Signed-off-by: jianbinc <shjwudp@gmail.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Add GEMM logic for blockwise quantized tensors.

GEMM test cases included in pytorch integration.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update NVTE_BLOCK_SCALING for GEMM.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Gate feature on CUDA 12.9
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Gemm typo.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Remove unecessary type converter change.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Reflect epilogue availability and test supported epilogues.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* GEMM simplifications from recipe branch.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Format py code.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update GEMM DGelu tests to match support depending on output dtype.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Force pow2Scales in GEMM
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Add GEMM test to pytorch test suite.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Add copyright to GEMM test.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update import for GEMM test.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Add license.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update test gemm supported predicate.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Use sgemm like interfaces and naming.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Rewrite GEMM comment.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* MR Feedback.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Recipe setup for Linear modules.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Use 12.9 feature test.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Run against tensor dumps from internal library.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update FIXME to TODO with linked issue.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update full recompute feature to save recipe.

The recompute context uses the same recipe
and fp8 settings as the original fwd pass.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* MR Feedback. Avoid reusing quantizer objects.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update logic in module.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Format py.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update for PP bug.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update test numerics.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update force_power_of_2 scales in the recipe.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update usage method to satisfy upstream changes.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* fix subchannel recipe in distributed test with bf16 gather
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* Edit and cleanup BF16 gather code.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update test import.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* support columnwise only mode to 1D quantize kernel
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* Format and move enum
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Skip alloc.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* try async bf16 gather
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* Format python code.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Document and type code.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update pytorch lint errors.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Dont set high precision dtype.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Add test for sanity and CG; fix CG for sequential?
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Keep make_quantizers API stable

Update num_quantizers instead to pass cuda_graph tests.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Fix import name.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Rename recipe method.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Skip grouped linear sanity test.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Set usage before BF16 gather.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* refactor for nvte_quantize_v2
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* Format code.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Cleanup nvte_quantize_v2
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Test fp32 scales.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Disable CUDA graph.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Simplify layernorm linear
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Cleanup layernorm linear.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* LayerNorm linear bwd gather logic.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Communication updates.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update transformer_engine/pytorch/ops/op.py

Apply MR comment change.
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: kwyss-nvidia <kwyss@nvidia.com>

* Lint fix.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* MR feedback.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Enable cuda graph tests.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Reduce chance of spurious failure and reword.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Review suggestions from @timmoon10
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Update CPP tests.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update common.h
Signed-off-by: Xin Yao <yaox12@outlook.com>

* Update test_float8blockwisetensor.py
Signed-off-by: Xin Yao <yaox12@outlook.com>

---------
Signed-off-by: Keith Wyss <kwyss@nvidia.com>
Signed-off-by: zhongboz <zhongboz@nvidia.com>
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: kwyss-nvidia <kwyss@nvidia.com>
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Signed-off-by: Xin Yao <yaox12@outlook.com>
Co-authored-by: zhongboz <zhongboz@nvidia.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: Xin Yao <yaox12@outlook.com>

Support fp8 primary weight in fsdp training
Signed-off-by: jianbinc <shjwudp@gmail.com>
Co-authored-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Handle case where FP8 current scaling quantizer gets default process group
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Fix linter warning
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Avoid canonicalizing TP group since it may not be initialized
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Enable fp8_primary_weights for current scaling
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Use different cast_master_weights_to_fp8 functions depending on the type of quantizer
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* All amaxes of model_weights should participate in reduce-max
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Clear _high_precision_init_val automatically in cast_master_weights_to_fp8 function
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Merge all all-reduce on amaxes into one NCCL kernel
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add unit tests for multi_tensor_compute_scale_and_scale_inv and preserve_high_precision_init_val
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Fix conflicts
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add unit test for cast_master_weights_to_fp8
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* use mock group to initialize fp8_autocast to avoid reduction of amax_history by fp8_autocast_exit
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Remove with_computing_amax and with_computing_scale
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Move replace_raw_data from QuantizedTensor to utils.py
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Remove allow_empty_output argument from nvte_compute_amax and set it always be true
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Rename import guard of recipe_common.cuh to be align with other import guards
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Add unit test for replace_raw_data
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Add test_replace_raw_data into qa/L0_pytorch_unittest/test.sh
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Minor changes in comments
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Add randomness to the unit test of replace_raw_data
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* (Maybe need revert) Add tex.quantize_to_fragment
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* (Maybe needsto rrevert) Use nvte_quantize_noop in quantize_to_fragment
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Fix lint error
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Move high_precision_init_val test and replace_raw_data test to test_sanity.py
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Remove test_fp8_model_init.py and test_replace_raw_data.py
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Remove cast_master_weights_to_fp8 and replace_raw_data from __all__ of tensor.__init__.py
Signed-off-by: kunlunl <kunlunl@nvidia.com>

* Move FP8 casting logic back from C++ tex funcs to Python
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Remove unimplemented function from header
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: kunlunl <kunlunl@nvidia.com>
Signed-off-by: Kunlun Li <94586211+kunlunl@users.noreply.github.com>
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Co-authored-by: Tim Moon <tmoon@nvidia.com>

* support tp-comm-overlap in Current Scaling recipe
Signed-off-by: Li Tao <lit@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* clean
Signed-off-by: Li Tao <lit@nvidia.com>

* fix test recipe argument to generalize to MXFP8
Signed-off-by: Li Tao <lit@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Reduce duplicated transpose in certain cases
Signed-off-by: Li Tao <lit@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Use per_tensor_scaling() to judge DS or CS
Signed-off-by: Li Tao <lit@nvidia.com>

* minor fixes
Signed-off-by: Li Tao <lit@nvidia.com>

* change comment description
Signed-off-by: Li Tao <lit@nvidia.com>

* add multi-layer unit test for tp overlap
Signed-off-by: Li Tao <lit@nvidia.com>

* support test case that run for several times
Signed-off-by: Li Tao <lit@nvidia.com>

* avoid save ub tensor in prepare_for_saving
Signed-off-by: Li Tao <lit@nvidia.com>

* fix
Signed-off-by: Li Tao <lit@nvidia.com>

* switch to a simple fix
Signed-off-by: Li Tao <lit@nvidia.com>

* formatting
Signed-off-by: Li Tao <lit@nvidia.com>

* simply test cases; avoid additional clone()
Signed-off-by: Li Tao <lit@nvidia.com>

* fall back to get_buffer in layernormmlp
Signed-off-by: Li Tao <lit@nvidia.com>

* use 2 layers for fp8 tpoverlap multi-layer test for better tolerance, limit max gpus for test
Signed-off-by: zhongboz <zhongboz@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



---------
Signed-off-by: Li Tao <lit@nvidia.com>
Signed-off-by: zhongboz <zhongboz@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: zhongboz <zhongboz@nvidia.com>

* Add options to comm overlap tests
Signed-off-by: Vasudevan Rengasamy <vrengasamy@nvidia.com>

* Fix Typo
Signed-off-by: Vasudevan Rengasamy <vrengasamy@nvidia.com>

* Update tests/pytorch/distributed/run_layer_with_overlap.py
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

---------
Signed-off-by: Vasudevan Rengasamy <vrengasamy@nvidia.com>
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>