GitLab

* shardy + quantize_layout rework
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* add assertion for NVFP4 in fused act and fused norm primitive
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

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

* [JAX] Support recipe flags for disabling SR, RHT, and 2D quantization
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

* Fix issue with SR state being erased due to pytree handling of NVFP4Quantizer
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Add test for SR state preservation across VJP boundaries
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix sharding of SR rng state
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

* update tolerances slightly now that SR is enabled
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

* Use hashlib for deterministic hashes across runs for SR
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* rename uses_rht on scaled tensors to has_applied_rht
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

* Move decision of whether to use RHT into helper.py and add dedicated RHT tests
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

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

* fix pure-jax rht usage criteria
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Adjust tolerances after rebase
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

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

* reuse amax for current scaling
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

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

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

* Rework shardy rules

* WAR for compound factor=1
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* Adding Amax Primitive and related args.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Enable local-amax for current-scaling and optionally run AR aross FSDP/TP/SP.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding doc for Amax Primitive.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix the function name conflict.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Modification as feedback suggested.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix errors from lint.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix the wrong amax-scope in the bwd.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Added more description for amax-scope
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix the wrong attribute name.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Keep dim for AmaxCalcuation.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Remove keepDim and add shardy_rule
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix shardy_rule
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Remove extra-collective bytes from ref_coll_count due to local amax.
Signed-off-by: Ming Huang <mingh@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>
Signed-off-by: Ming-Xu Huang <mingh@nvidia.com>
Co-authored-by: Phuong Nguyen <phuonguyen@nvidia.com>

* remove import jax.extend.ffi
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* Custom call tests passing
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

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

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

* Support using amax on HighPrecision tensor if it exists instead of recomputing for current scaling
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix shardy issue with amax being shape 1,1,1 instead of shape (1,)
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Add higher-precision VJP tests to test_distributed_layernorm_mlp
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Cast non-quantized kernels to input dtype in VJPs
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Rename HighPrecisionTensor to NoScaleTensor
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Use NoScaleTensor in pure JAX impls where it was missing
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

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

* add amax input to DBiasQuantizePrimitive and FFI
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

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



* make sure amax is init with zero
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* fix sharding rule
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* FP8 AllGather in FP8 GroupedGEMM

1. Support current scaling FP8 quantation with a given amax.
2. Support FP8 AG in fwd and BF16 RS in bwd.
3. The workflow is AR-max -> FP8 Quant -> FP8 AG -> FP8 GroupedGEMM.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Slightly refactor
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding documents of new args.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding unit-tests.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding license.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Move unit-tests to L1.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Move quantizaer store/reset into FP8 only.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding all layout support for Blackwell+
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adopt the feedback from code-review.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fixed the wrong stream used by d2d in groupedGEMM FFI.
Signed-off-by: Ming Huang <mingh@nvidia.com>

---------
Signed-off-by: Ming Huang <mingh@nvidia.com>
Co-authored-by: Phuong Nguyen <phuonguyen@nvidia.com>

* added amax as an optional arg
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* add manage_primitives() helper

* disable GEMM primitives for non-MXFP8 recipes

* implement the NVTE_JAX_CUSTOM_CALLS + deprecate NVTE_JAX_CUSTOM_CALLS_RE

* replace NVTE_JAX_CUSTOM_CALLS_RE with NVTE_JAX_CUSTOM_CALLS in TE tests and examples

* fix use_jax_gemm contextmanager
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* added XLA FFI custom op for TE/common nvte_cublas_gemm
Signed-off-by: Alp Dener <adener@nvidia.com>

started GemmPrimitive, abstract done
Signed-off-by: Alp Dener <adener@nvidia.com>

gemm custom op working with BF16, needs testing for FP8/MXFP8
Signed-off-by: Alp Dener <adener@nvidia.com>

converted TE GEMM API to use ScaledTensor and added os ENV flag to use TE GEMM under general gemm() call
Signed-off-by: Alp Dener <adener@nvidia.com>

BF16 tests passing, FP8 tests should be passing but contracting_dims has a scoping issue
Signed-off-by: Alp Dener <adener@nvidia.com>

fp8 tests passing for E4M3, getting CUBLAS_STATUS_NOT_SUPPORTED for E5M2
Signed-off-by: Alp Dener <adener@nvidia.com>

updated GEMM API to use separate LHS and RHS quantizers instead of a QuantizerSet
Signed-off-by: Alp Dener <adener@nvidia.com>

new GemmPrimitive passing all Dense tests
Signed-off-by: Alp Dener <adener@nvidia.com>

import cleanup and reverted code chunk movement
Signed-off-by: Alp Dener <adener@nvidia.com>

removed unused .transpose() implementations from ScaledTensors
Signed-off-by: Alp Dener <adener@nvidia.com>

all custom call tests passing on Hopper, GEMM-related tests cover both GemmPrimitive and native JAX impl
Signed-off-by: Alp Dener <adener@nvidia.com>

removed direct calls to GemmPrimitive.enabled() from outside of cpp_extensions
Signed-off-by: Alp Dener <adener@nvidia.com>

removed unused changes to ScaledTensor classes and debug prints
Signed-off-by: Alp Dener <adener@nvidia.com>

* minor unit test cleanup
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* FP8 tests passing on Blackwell but MXFP8 outputs NaN
Signed-off-by: Alp Dener <adener@nvidia.com>

* reverted dense and fuseddense changes, FP8 test passing on Hopper and Blackwell, MXFP8 has issues with E5M2
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* MXFP8 issue traced to scale factor padding with NaNs instead of zeros
Signed-off-by: Alp Dener <adener@nvidia.com>

* padding scale with 2^-127 instead of nans
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* fix bug on rhs_scale_inv usage
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* cleanup E8M0 type converter use it in gemm.cpp
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* segfault fixed, passing all unittests on Blackwell
Signed-off-by: Alp Dener <adener@nvidia.com>

* fix for fuseddense tests
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* fix workspace alignment
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

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



* fixed GemmPrimitive custom partitioning to match jax.nn.scaled_matmul
Signed-off-by: Alp Dener <adener@nvidia.com>

all unit tests passing on H100x8 node
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



linting fixes
Signed-off-by: Alp Dener <adener@nvidia.com>

fixed batch dimension numbers
Signed-off-by: Alp Dener <adener@nvidia.com>

fixed FP8 scale sharding rule when there are no FP8 scales
Signed-off-by: Alp Dener <adener@nvidia.com>

added error message for unsupported Shardy partitioner
Signed-off-by: Alp Dener <adener@nvidia.com>

fixed test tolerances for FP8 cases
Signed-off-by: Alp Dener <adener@nvidia.com>

fixed shardy test skip cases
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* moved reshape of encoder output in encoder examples to make custom partitioning rules work correctly
Signed-off-by: Alp Dener <adener@nvidia.com>

* added helper functions for padding and unpadding block scales, changed GemmPrimitive to accept unpadded scales and pad them after sharding
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* updated shardy rules for all custom ops to decouple block scale rules from their tensors
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* fixed linting errors
Signed-off-by: Alp Dener <adener@nvidia.com>

* changed unit test use_jax_gemm option to be a context to preserve external custom op settings, tightened multi-GPU encoder test tolerances, changed gemm() API to use contracting_dims and batched_dims separately instead of dimension_numbers
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* fixed typo in test utils
Signed-off-by: Alp Dener <adener@nvidia.com>

* added sequence-first input warnings
Signed-off-by: Alp Dener <adener@nvidia.com>

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

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



* fixed datasets version for JAX examples
Signed-off-by: Alp Dener <adener@nvidia.com>

* reverting modification to force_1x_quantization decision
Signed-off-by: Alp Dener <adener@nvidia.com>

* corrected gemm function syntax in unit tests
Signed-off-by: Alp Dener <adener@nvidia.com>

---------
Signed-off-by: Alp Dener <adener@nvidia.com>
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Phuong Nguyen <phuonguyen@nvidia.com>

* fixes for jittable grouped_quantize

* fixes for jittable grouped_gemm

* fix contracting_dim for wgrad gemm

* exclude jitted grouped_gemm from the unit test as it does not work cudaGraph

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

* Implemented GroupedDense and TestGroupedDense for BF16, FP16, and FP8 
* Fix GroupedGemmFFI cuBLAS workspace alignment bug
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

Revert "[JAX] GroupedDense v.2 without dynamic shape (#1721)"

This reverts commit 5d01ef21

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

* Implemented GroupedDense and TestGroupedDense for BF16, FP16, and FP8 
* Fix GroupedGemmFFI cuBLAS workspace alignment bug
Signed-off-by: Hua Huang <huah@nvidia.com>
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* refactor the multi_stream utils + implement nvte_multi_tensor_quantize in TE/Common

* implement GroupedQuantizer and grouped_quantize in jaxx

* fix logical_axes_names for transpose tensor in ScaledTensor
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: Ming Huang <mingh@nvidia.com>

* Fix 1x quantize kernel availability check on hopper

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

* Use 1x quantization + jax transpose on BW for performance
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Use 1x quantization on Hopper as well as it is also faster
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Undo architecture check helper function
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>

Make primitive names more granular for better disabling granularity
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

Fix distributed layernorm test failure
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.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>

* [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>

* Add experimental Shardy support.

Production use is not yet recommended.

---------
Signed-off-by: Johannes Reifferscheid <jreiffers@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>

* 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>

* 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>

Make ffi compatible with jax 0.4
Signed-off-by: Reese Wang <rewang@nvidia.com>
Co-authored-by: Phuong Nguyen <phuonguyen@nvidia.com>

* Fix wheel install after src install
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

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

* switch order of dirs for finding so
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Use existing dir src build
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

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

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

* fixes L1 test

* fix test_multigpu_encoder

* fixes for other multi-encoder tests

* jax.extend.ffi to jax.ffi

* initialization with float32

* add init_dtype as an optional arg to all modules

* update use_scan query from xla flags

* relax threshold for test_encoder fp8

* relax the tols

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

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

[JAX] XLA Custom Calls with FFI for FusedAttnFwd, Quantize, Transpose, ActLuFP8, LayerNormForwardFP8FFI, and LayerNormBackwardFFI (#1263)

* Add TransposeFFI, test passed
Signed-off-by: Hua Huang <huah@nvidia.com>

* Add ActLuFP8FFI; fix TransposeFFI
Signed-off-by: Hua Huang <huah@nvidia.com>

* Add QuantizeFFI
Signed-off-by: Hua Huang <huah@nvidia.com>

* Add FusedAttnForwardFFI and some unit tests
Signed-off-by: Hua Huang <huah@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: Hua Huang <huah@nvidia.com>

* Add LayerNormForwardFP8FFI & LayerNormBackwardFFI
Signed-off-by: Hua Huang <huah@nvidia.com>

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

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



* Revise FusedAttnForwardFFI()
Signed-off-by: Hua Huang <huah@nvidia.com>

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

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



* Add FFI_CudaGraph_Traits

All tests passed, ready for merge
Signed-off-by: Hua Huang <huah@nvidia.com>

* Bug fix for FFI data type mismatch

Also add a safeguard on the entrance to FFI function
Signed-off-by: Hua Huang <huah@nvidia.com>

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

Signed-off-by: Frederic Bastien <fbastien@nvidia.com>
Co-authored-by: Phuong Nguyen <36155692+phu0ngng@users.noreply.github.com>

* Add enabled() to BasePrimitive

* Add layernorm/rmsnorm fallback

* Add cast_fp8 fallback

* Add transpose/cast_transpose XLA fall back

* Act_lu fallback

* Add transpose fallback

* Add softmax fallback

* Unify the use of _cast_fp8

* Add tests for NVTE_JAX_CUSTOM_CALLS_RE

---------
Signed-off-by: Reese Wang <rewang@nvidia.com>
Co-authored-by: Phuong Nguyen <36155692+phu0ngng@users.noreply.github.com>

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

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

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

* Splitted cpp_extensions.py, renamed mlp.py and fused_attn.py
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* fixed import in tests
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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