[PyTorch] Miscellanous fixes for FP8 DPA module (#804)

* initialize tp_group for FP8 DPA

Signed-off-by: Charlene Yang <[email protected]>

* fix cuDNN version in unit tests for cuDNN v9

Signed-off-by: Charlene Yang <[email protected]>

* add hook to ignore missing fused_attn._extra_states if training from old checkpoints

Signed-off-by: Charlene Yang <[email protected]>

* remove test and redundant implementation from last commit

Signed-off-by: Charlene Yang <[email protected]>

* remove warning message and replace with docstring

Signed-off-by: Charlene Yang <[email protected]>

* remove tp_size/tp_group in FusedAttention; amax reduction is handled with fp8_group

Signed-off-by: Charlene Yang <[email protected]>

* move core_attention.fused_attention._extra_state to core_attention._extra_state

Signed-off-by: Charlene Yang <[email protected]>

* simplify post_state_dict_hooks between FU and DPA

Signed-off-by: Charlene Yang <[email protected]>

* add temporary test

Signed-off-by: Charlene Yang <[email protected]>

* remove previous attempts to move core_attention.fused_attention to core_attention; keep the test

Signed-off-by: Charlene Yang <[email protected]>

* remove the test

Signed-off-by: Charlene Yang <[email protected]>

* disable pylint self arg for hook which is required by hook

Signed-off-by: Charlene Yang <[email protected]>

---------

Signed-off-by: Charlene Yang <[email protected]>
Signed-off-by: cyanguwa <[email protected]>

Author: cyanguwa

tests/pytorch/fused_attn/test_fused_attn.py

@@ -70,7 +70,8 @@ def reset_global_fp8_state():
 def _cudnn_version() -> Tuple[int, int, int]:
     """Runtime cuDNN version (major, minor, patch)"""
     encoded_version = ext.get_cudnn_version()
-    major, encoded_version = divmod(encoded_version, 1000)
+    major_version_magnitude = 1000 if encoded_version < 90000 else 10000
+    major, encoded_version = divmod(encoded_version, major_version_magnitude)
     minor, patch = divmod(encoded_version, 100)
     return (major, minor, patch)
 

transformer_engine/pytorch/attention.py

@@ -2711,6 +2711,17 @@ def __init__(
             if os.environ["NVTE_FUSED_ATTN_FORCE_WORKSPACE_OPT"] == "1":
                 os.environ["CUDNN_FRONTEND_ATTN_DP_WORKSPACE_LIMIT"] = "-1"
 
+        def remove_extra_states_check(self, incompatible_keys): # pylint: disable=unused-argument
+            """
+            Temporarily remove fused_attention._extra_state as a missing key
+            when loading older TransformerEngine checkpoints. Will phase out
+            this hook in TransformerEngine 2.0.
+            """
+            for key in incompatible_keys.missing_keys:
+                if 'fused_attention._extra_state' in key:
+                    incompatible_keys.missing_keys.remove(key)
+        self.register_load_state_dict_post_hook(remove_extra_states_check)
+
     def get_fp8_weights_scratchpad(
         self,
         is_first_microbatch: Union[bool, None],
@@ -3063,6 +3074,7 @@ def __init__(
                                                   layer_number=layer_number,
                                                   deterministic=self.deterministic,
                                                   **attn_kwargs)
+
         self.unfused_attention = UnfusedDotProductAttention(
             norm_factor, **attn_kwargs, layer_number=layer_number)