GitLab

* Reapply "Allow NVTEShape to own data." (#1703)

This reverts commit 91405eb4

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

* Update code so that data is replaced by an array.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Specify unambiguous Tensor constructor in tests.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Fix assumption in test of 2D shape.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Remove row and col
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

---------
Signed-off-by: Keith Wyss <kwyss@nvidia.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>

* Update test_helper.py and add QuantizeConfig class for CurrentScaling
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* WIP distributed current scaling
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Distributed Current Scaling (debugging). Distributed implementation with replicated scale_inv works for layernorm_mlp but feels like a hack

Has different per-device scale_inv values, but jax.debug.print only shows one of them. Since we're telling JAX/XLA that this scale is replicated, I think it assumes all the values are equal. However, it doesn't actually check this, so it seems we are able to get away with per-device scales for current scaling but I am not sure how stable this will be and may randomly fail if us or the user changes partitioning at all or if XLA decides to actually act on the assumption that all these scale_invs are the same.
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Implement distributed current scaling by computing a global amax and scale before quantization
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Add encoder and mnist tests for current scaling
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Add primitive prefix to shardy unique_vars to prevent factor conflicts when performing unfused primitives for current scaling
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Remove scale_shape primitive arg that is no longer used
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Format
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix expected result on multiprocessing encoder test
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Lint fix
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Update multiprocessing current scaling tolerances
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Uncomment test case that was disabled for testing
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Remove commented out debug line
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Move MultiHeadAttention into its own file. Modify tests and files in t_e/pytorch to import from the new MHA module
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Resolving lost MHA changes from PR 1614 as a result of rebase
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Move context parallelism code into it's own file. Modify test and local imports of cp code accordingly
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Move softmax.py frm pytorch/ to pytorch/d_p_a
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Move Unfused and Fused attention to backends.py and some utils functions to pytorch/utils.py
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Resolving lost mark_activation_offload changes from PR 1678 as a result of rebase
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Code clean up
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Refactor attention dir
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Refactor dir structure. Make relevant symbols public in __init__ for attention and d_p_a dirs
Move FA package imports to backends.py
Code cleanup
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Modify tests to import attention modules correctly
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Lint fixes
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Code clean up and fix typo
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Allowing InferenceParams and RoPE imports from attention module and pytorch module
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Allow InferenceParams and RoPE imports via transformer_engine.pytorch and transformer_engine.pytorch.attention modules
Remove unnecessary checks for check_set_window_size in MHA and TL
Reorder backends such that smaller classes at the start and larger ones at the end
Code clean up
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Reinstating changes from PR 1478 for rope.py lost during rebase conflict resolution
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Fix lint issues
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* nit: Code clean up
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Make imports leaner
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



---------
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Introduce nvte_memset to provide a fill kernel that is faster than cudaMemsetAsync for small sizes (#1716)

* nvte_memset fills single float value
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Support larger sizes than a single value and add tests
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* add support for `sb1d` freqs tensor in Fused RoPE
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

* add `start_positions` variable to `apply_rotary_pos_emb` function to make staggered rope application faster
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* add pytorch path for `start_positions` and corresponding tests
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* add tests for start_positions with thd
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* fixes from feedback
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* remove start_positions from backward pass
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* from feedback
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* make notes shorter
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

* [JAX-Q] Single GPU current scaling for JAX
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix scale check dtype for MXFP8 scales affecting tests using assert_bitwise_scaled_tensors
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Address comments
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Remove cast to fp32 for norm primitives now that zero-centered gamma dtype issue is fixed
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix lint issue
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Remove unnecessary cast to fp32
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Lint
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

Check CuDNN version and apply unfused norm if
below a version with the fix
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* rtx5090 arch fix support
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

* apprend `nvte` to the function name so that its visible in framework specific dirs
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

* fix typo
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

* add filter for nvte_is_supported_nontn_fp8_gemm
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

* properly expose the api
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* feedback from PR
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* move the function to apt header/c files
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

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



* add more info
Signed-off-by: Sudhakar Singh <sudhakars@nvidia.com>

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

Revert "Allow NVTEShape to own data. (#1674)"

This reverts commit e61ce77c

.
Signed-off-by: Tim Moon <tmoon@nvidia.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>

* Allow NVTEShape to own data.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Convert repeated copy paths to nvte_make_shape calls.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

* Build fixes.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

---------
Signed-off-by: Keith Wyss <kwyss@nvidia.com>
Signed-off-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>

* Add a flag to support computing zero-centered gamma in weight dtype or compute dtype for CuDNN
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Address comments
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Add test cases for full coverage in jax/test_layer.py
- causal and window size None
- causal and window size default (-1,1)
- no_mask and window size default (-1,1)
- no_mask and window size default (2,2)
- padding and window size None
- padding_causal and window_size (2,2)
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Correct the condition where padding_causal_mask was being mapped to scaled upper triangle
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Fix Issue #1524
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Add a runner and test cases for jax.flax.module.Softmax class for fwd pass only
Segregate runner classes for Softmax module and softmax primitives
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Simplify logic when picking softmax primitives and softmax jax framework calls
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Simplify the logic for performing jax based softmax
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Code clean up
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Add support table for mask, SWA and Softmax type. Code linting
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Explicit SWA conditons in comments. Fix Typo
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



* Resolve typo to remove None in SWA comments section
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

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

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



---------
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* support adam bf16 state
Signed-off-by: XiaobingSuper <xiaobingzhangupc@gmail.com>

* use fp32 kernel but keep bf16 optimizer states to save memory
Signed-off-by: lit <lit@nvidia.com>

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

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



---------
Signed-off-by: XiaobingSuper <xiaobingzhangupc@gmail.com>
Signed-off-by: lit <lit@nvidia.com>
Co-authored-by: XiaobingSuper <xiaobingzhangupc@gmail.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>

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

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

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

* small changes
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



* test
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



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

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

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

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

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

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

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

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

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

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

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

* added test
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



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

* changed the test
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>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.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>

* Enable MXFP8 and Per-Tensor Current Scaling for Grouped Linear
Signed-off-by: Xin Yao <xiny@nvidia.com>

* enable float8blockwise
Signed-off-by: Xin Yao <xiny@nvidia.com>

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

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



* update
Signed-off-by: Xin Yao <xiny@nvidia.com>

* remove grouped linear parallel mode test
Signed-off-by: Xin Yao <xiny@nvidia.com>

* update test
Signed-off-by: Xin Yao <xiny@nvidia.com>

* resolve comments
Signed-off-by: Xin Yao <xiny@nvidia.com>

* internal=False for now
Signed-off-by: Xin Yao <xiny@nvidia.com>

* remove unused import
Signed-off-by: Xin Yao <xiny@nvidia.com>

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

* Add experimental Shardy support.

Production use is not yet recommended.

---------
Signed-off-by: Johannes Reifferscheid <jreiffers@nvidia.com>

* add support for new recipe on permute_fusion, rm fp unpermute
Signed-off-by: tongliu <tongliu@nvidia.com>

* fix lint
Signed-off-by: Xin Yao <xiny@nvidia.com>

* remove fp8 from index map
Signed-off-by: Xin Yao <xiny@nvidia.com>

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

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



* skip unsupported tests
Signed-off-by: Xin Yao <xiny@nvidia.com>

---------
Signed-off-by: tongliu <tongliu@nvidia.com>
Signed-off-by: Xin Yao <xiny@nvidia.com>
Co-authored-by: tongliu <tongliu@nvidia.com>
Co-authored-by: Xin Yao <xiny@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* Add option to cache activation input in FP8
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Avoid casting to FP8 transpose
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Skip input caching if device is not supported
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Add documentation that FP8 input caching is experimental
Signed-off-by: Tim Moon <tmoon@nvidia.com>

---------
Signed-off-by: Tim Moon <tmoon@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>

* Debug checkpointing with 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



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

Explicitly specify quantized tensor usages needed for linear op backward
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* scaling enum abstract

* rm NVTE_ from ScalingMode names

* rework scaling mode enum in grouped gemm

* fix norm sharding

---------
Signed-off-by: Phuong Nguyen <phuonguyen@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>

* Refactor GEMM param canonicalization

Configure A and B matrices separately. Have separate code path for each scaling mode.
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



* Prune number of tests.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

---------
Signed-off-by: Keith Wyss <kwyss@nvidia.com>
Signed-off-by: Tim Moon <tmoon@nvidia.com>
Co-authored-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>

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

* refactor to add cp support for sbhd/bshd
Signed-off-by: Xin Yao <xiny@nvidia.com>

* support interleaved
Signed-off-by: Xin Yao <xiny@nvidia.com>

* format
Signed-off-by: Xin Yao <xiny@nvidia.com>

* add interleaved to RotaryPositionEmbedding in test
Signed-off-by: Xin Yao <xiny@nvidia.com>

* update
Signed-off-by: Xin Yao <xiny@nvidia.com>

* merge sbhd/bshd and thd functions
Signed-off-by: Xin Yao <xiny@nvidia.com>

---------
Signed-off-by: Xin Yao <xiny@nvidia.com>

* rename QuantizeAxis to QuantizeLayout, get_layout to get_data_layout, q_axis to q_layout

* add fatten_axis option

* added gated act to test encoder

* sharding constraint fixes

* fix padding when flattening first dim needs to be padded

* update test sizes so that padding is tested

* rm output sharding as it can be done in the flax module

* sharding scale_inv for mxfp8

---------
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

MXFP8 flax layer tests
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Blockwise float8 quantizer and quantized tensor class.

The classes are configurable for 128x128 blocksize
and 1x128 blocksize via setting block_scaling_dim == 2,1 respectively.

Scale tensors are stored in a format emenable for matrix multiplication,
however the integration of matmul is deferred as a separate story.

Fusions of quantization and DBIAS or activation functions are not yet
implemented, and the dequantization is currently implemented in torch.

Tests for quantization are included in C++ and pytorch layers, with
exact comparison to reference quantizer behavior as well as an attempt
to hit interesting branches through the API such as tensor creation
in pytorch and CPP and dequantization of row and columnwise usage.

Two CUDA kernels for quantization are included, and are direct ports
of equivalents in the kitchen repository, where a subchannel recipe
has been used for end to end training.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Apply linting changes.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Alignment for 1D scaling for GEMM edge case.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

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

* Fix merge conflict with name change.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Use common tensor map API.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Change API to use two scaling mode enums.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

* Update some call sites.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Tests for torch tensor API surface.

Since the quantized tensor is a tensor
subclass, these tests exercise torch hooks.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Reuse scale calculation between quantizer refs.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Save memory by dropping reference to saved tensors.

Issues previously observed are solved.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Remove constexpr parameters from kernel.

Code size is reduced with fewer constexpr params.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Merge conflict from rebase.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Add shape implementations for block scaling.

nvte_shape was added upstream. Logic added
for block scaled fp8.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Move benchmark to te_playground
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Remove amax_epsilon and pow_2_scales from tensor.

Hardcodes the default values.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

* Fixup MR changes that broke.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Safer ifdef in kernel.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Documentation prose.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Reuse compute_scale function from Current Scaling.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Bugfix on inf_value scale refactor.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Remove qopt calls from test.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

* Add copyright to reference scale calc.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Use ptx.cuh functions instead of cde.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

* Update shape logic with allocation and reuse shape.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

* Copyright and header guard.
Signed-off-by: Keith Wyss <kwyss@nvidia.com>

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

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

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

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

* Update CS scale update test to use updated ref impl
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Update JAX scaling mode enum
Signed-off-by: Tim Moon <tmoon@nvidia.com>

* Skip tests on Lovelace
Signed-off-by: Tim Moon <tmoon@nvidia.com>

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

* refactor + mxfp8

* added grouped gemm

* rename linear to dense

* added cublas init phase for groupedGemm

* relax the tol of test encoder multiprocessing mxfp8 by 0.001
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

---------
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>
Co-authored-by: Hua Huang <huah@nvidia.com>
Co-authored-by: Jeremy Berchtold <jberchtold@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>

* Cleanup sanity tests and add CS recipe tests
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix sanity test
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix CG capture with CS recipe
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix ops for CG
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

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

* skip cuDNN 9.8 for KV caching
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* revert from max_seqlen_kv to max_sequence_length for InferenceParams
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* rename test_paged_attn to test_kv_cache
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove redundant None returns in bwd
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* add debug flags when no backend is found
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



* skip kv_cache_accuracy tests for cuDNN 9.8
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* truncate length of cu_seqlens for consistency with q/k/v shape
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* add back padding_brcm for fused attn tests
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* re-enable kv_cache_accuracy test for 9.8
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fix cuDNN search dir
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* fixes based on review
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

* remove extra empty line
Signed-off-by: Charlene Yang <8636796+cyanguwa@users.noreply.github.com>

---------
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>

* Defer torch compilation steps until first function call
Signed-off-by: Peter St. John <pstjohn@nvidia.com>

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

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



* Fix function call in smoke test
Signed-off-by: Peter St. John <pstjohn@nvidia.com>

---------
Signed-off-by: Peter St. John <pstjohn@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>

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

* fix dtypes of fused_attn_bwd in CP+A2A
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

* fix dtypes of fused_attn_bwd in CP+P2P
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

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

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



* fix amax_per_step
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

* clone scaling factors of fwd quantizers
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

* fix fwd quantizers of CP+P2P
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

* minor change
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

* dequantize fp8 out in CP unit test
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

* delete redundant None in FusedAttnFunc bwd
Signed-off-by: Xiaowei Ren <xren@nvidia.com>

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