GitLab

* Try async copy of grouped GEMM group_sizes data
Signed-off-by: Hua Huang <huah@nvidia.com>

---------
Signed-off-by: Hua Huang <huah@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>

* not fuse bias for output all reduction case + unit tests
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* norm to reduce dgamma along tpsp as well
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* clean up tests
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* fix test_distributed_layernorm byte counts
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* increase tols for jax_gemm
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

Signed-off-by: Varun Thumbe <vthumbe@nvidia.com>
*Jax integration for clamped swiglu. This is the continuation of PR which added Clamped Swiglu(used in GPT OSS) support in TE along with Pytorch integration. This PR hooks up the clamped swiglu and dswiglu's nvte APIs to TE Jax.

fix rng_state shape
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>
Co-authored-by: jberchtold-nvidia <158520091+jberchtold-nvidia@users.noreply.github.com>

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

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

* init cgemm + unit tests

* UB bootstrap with NCCL, no MPI dependency

* add NVLINK-P2P check + error message

* skip tests if no NVLINK available

* use std::vector to store ncclComm_t

* update misuse of TP warning
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>

* Applying the original precision as Norm outputs' and activation compuations.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Adding knob to control norm output precision.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Removing the knob and applying lower-precision norm with current-scaling only.
Signed-off-by: Ming Huang <mingh@nvidia.com>

* Fix the error when quantizer==None
Signed-off-by: Ming Huang <mingh@nvidia.com>

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

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

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

* clean up 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>

* Fix failing tests for dropout=0.1 and bias for fused attn for blackwell
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

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

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



* Fix the skip message
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

* Assert in fused attn bwd pass for sm100
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

Add check for sm100
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

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

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



* Add support to get all devs in the process for jax
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

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

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



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

* Make get_all_device_compute_capability more pythonic, thereby avoiding unnecessary type conversion
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

* Represent attn bias using enum instead of string
Signed-off-by: Kshitij Lakhani <klakhani@nvidia.com>

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

* Decouple recipe and scaling mode
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Expose global QuantizeConfig instance as a getter
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

* Merge branch 'main' into dev/jberchtold/jax-scaling-mode-and-recipe-decoupling
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

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

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>
Signed-off-by: jberchtold-nvidia <158520091+jberchtold-nvidia@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>

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

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

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

* rework test for MeshResource
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* add mesh_resource into fp8_autocast in test_helper.py
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

---------
Signed-off-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>

* added shardy warning
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>


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

[JAX] Error checking for mesh resource and update GemmPrimitive to use global_mesh_resource().fsdp_resource (#2088)

* Enforce global MeshResource is set
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Use global_mesh_resource().fsdp_resource in gemm primitive
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

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

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

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

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

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

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

* Support setting collection name for quantizer set Flax variables in TransformerEngineBase flax module
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Support creating quantizer set from a recipe directly
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix debug error format string in gemm.py
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>

* enabled TE GEMM for all recipes
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* add warnings
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



* fix lint
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>

* rm batch_dim, sequence_dim, sequence_parallel_output
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* rm unnecessary transpose_batch_sequence arg from some modules
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

Revert "[JAX] Disable TE Norm Custom Calls (#1993)"

This reverts commit 6c970612

.

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

Disable Norm custom calls
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* new gemm operand specs processing
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

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

[JAX] Fixing GemmPrimitive partitioning rules to handle tensor-parallelism correctly for sequence-parallel inputs (#1980)

* updated GemmPrimitive partitioning rules to explicitly control all-reduce vs. reduce-scatter for sequence-parallelism
Signed-off-by: Alp Dener <adener@nvidia.com>

* corrected handling of FSDP sharding for the RHS operand
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



* use correct logical axes variable to identify sequence-parallel dim in LayerNormDenseGeneral
Signed-off-by: Alp Dener <adener@nvidia.com>

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

* added assert on sequence-parallel options when GemmPrimitive is disabled
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



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

* remove unnecessary padding
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

* adapt the test_distributed_layernorm byte count
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>

* Fix JAX grouped gemm error on CUDA 12.9.1 & later by using 16B alignment for scale ptr
Signed-off-by: Hua Huang <huah@nvidia.com>

* Pad MXFP8 scales with 2*-127 instead of NaNs
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>

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

* TensorUsage + FP8 GEMM with all layouts handling on BW
Signed-off-by: Phuong Nguyen <phuonguyen@nvidia.com>


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

* Support MXFP8 and handle empty matrices
Signed-off-by: Hua Huang <huah@nvidia.com>

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

* Add support for Fused Attn MLA head_dim_qk != head_dim_v
	Modify is_fused_attn_kernel_available() to accept different head_dims for qk and v
	Modify FusedAttnHelper to accept different head_dims for qk and v and modify assert dims checks in parse_qkv_aval()
	Modify FusedAttnFwdPrimitive and FusedAttnBwdPrimitive to accept different head_dims for qk and v
	Modify Fused Attn related cpp and csrc extension API calls to accept different head_dims for qk and v
	Modify DotProductAttention call() to extract head dims separately for qk and v
	Modify the FusedAttn Tests to accommodate for API changes in FusedAttn API
	Add test case for head_dim_qk != head_dim_v (failing)
	Modify the baseline JAX appropriately to reshape the output vector based on v dims and not q dims
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 context dims in general DPA in test_fused_attn
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Fix dim for output tensor by replacing with v head dim rather than q head dim
Add test cases for jax fused attn where head_dim_qk != head_dim_v for a combination of data types and attention type
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Modify the fused attn jax unit test case for head dim qk != head dim v
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Use new FusedAttnRunner function signature for separate hidden dim for qk and v in Fused Attn distributed tests
Code clean up
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Fix usage of is_fused_attn signature in distributed tests
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

* Remove unnecessary assert
Signed-off-by: Kshitij Janardan Lakhani <klakhani@nvidia.com>

---------
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>

* add support for head dim > 128
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* remove debugging
Signed-off-by: Charlene Yang <charleney@nvidia.com>

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

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



* raise tols slightly to tolerate 1/2048 mismatches
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* fix is_training for test_te_layer
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* add bprop support for blackwell
Signed-off-by: Charlene Yang <charleney@nvidia.com>

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

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



* minor tweak for format
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* fix backend selection results
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* bump sm100 to sm100+
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* add sq=1 test for MLA
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* enable sq=1 for bprop
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* minor tweak in comments
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* fix head_dim logic and remove pytest skip
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* add FE fix for d>128
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* update FE again to take in small fixes
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* add cuDNN version info in L0 tests
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* increase tols for Unfused + large dim
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* Revert "add cuDNN version info in L0 tests"

This reverts commit 3e1b426ca5319a2c0540b9e73bba7047d0e583e5.
Signed-off-by: Charlene Yang <charleney@nvidia.com>

* fix tols for Unfused
Signed-off-by: Charlene Yang <charleney@nvidia.com>

---------
Signed-off-by: Charlene Yang <charleney@nvidia.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>

* 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>