From d8063cd38760d091dafe82caca6bb2ccc2cd40fd Mon Sep 17 00:00:00 2001
From: apbose <apbose694@gmail.com>
Date: Mon, 24 Feb 2025 09:26:56 -0800
Subject: [PATCH 1/2] removing the fuse distributed ops lowering pass for tegra
 platforms

---
 .../lowering/passes/_aten_lowering_pass.py    | 30 ++++++++++---------
 1 file changed, 16 insertions(+), 14 deletions(-)

diff --git a/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py b/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py
index f611c90f51..08c3eee3e7 100644
--- a/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py
+++ b/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py
@@ -17,20 +17,22 @@
 from .replace_max_pool_with_indices import replace_max_pool_with_indices
 from .view_to_reshape import view_to_reshape
 
-ATEN_POST_LOWERING_PASSES = DynamoPassManager.build_from_passlist(
-    [
-        remove_input_alias_fixing_clones,
-        constant_fold,
-        repair_input_as_output,
-        fuse_prims_broadcast,
-        fuse_distributed_ops,
-        replace_max_pool_with_indices,
-        lower_scaled_dot_product_attention,
-        view_to_reshape,
-        remove_assert_nodes,
-        accumulate_fp32_matmul,
-    ]
-)
+pass_list = [
+    remove_input_alias_fixing_clones,
+    constant_fold,
+    repair_input_as_output,
+    fuse_prims_broadcast,
+    replace_max_pool_with_indices,
+    lower_scaled_dot_product_attention,
+    view_to_reshape,
+    remove_assert_nodes,
+    accumulate_fp32_matmul,
+]
+
+if torch.cuda.get_device_capability() not in [(8, 7), (7, 2)]:
+    pass_list.append(fuse_distributed_ops)
+
+ATEN_POST_LOWERING_PASSES = DynamoPassManager.build_from_passlist(pass_list)
 
 ATEN_PRE_LOWERING_PASSES = DynamoPassManager.build_from_passlist(
     [

From 57943f30cfc03ede4c0f8a331aa1060a31b8f27a Mon Sep 17 00:00:00 2001
From: apbose <apbose694@gmail.com>
Date: Tue, 25 Feb 2025 06:09:33 -0800
Subject: [PATCH 2/2] utility function to detect tegra platform

---
 .../dynamo/lowering/passes/_aten_lowering_pass.py           | 3 ++-
 py/torch_tensorrt/dynamo/utils.py                           | 6 ++++++
 2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py b/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py
index 08c3eee3e7..600a130631 100644
--- a/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py
+++ b/py/torch_tensorrt/dynamo/lowering/passes/_aten_lowering_pass.py
@@ -3,6 +3,7 @@
 
 import torch
 from torch_tensorrt.dynamo._settings import CompilationSettings
+from torch_tensorrt.dynamo.utils import is_tegra_platform
 
 from .accumulate_fp32_matmul import accumulate_fp32_matmul
 from .constant_folding import constant_fold
@@ -29,7 +30,7 @@
     accumulate_fp32_matmul,
 ]
 
-if torch.cuda.get_device_capability() not in [(8, 7), (7, 2)]:
+if not is_tegra_platform():
     pass_list.append(fuse_distributed_ops)
 
 ATEN_POST_LOWERING_PASSES = DynamoPassManager.build_from_passlist(pass_list)
diff --git a/py/torch_tensorrt/dynamo/utils.py b/py/torch_tensorrt/dynamo/utils.py
index 1ff754532f..0d9cb8ca6f 100644
--- a/py/torch_tensorrt/dynamo/utils.py
+++ b/py/torch_tensorrt/dynamo/utils.py
@@ -799,3 +799,9 @@ def get_output_dtypes(output: Any, truncate_doulbe: bool = False) -> List[dtype]
             f"got unexpected type {type(output)}, expected type is a torch.fx.node.Node or a tuple/list of torch.fx.node.Node"
         )
     return output_dtypes
+
+
+def is_tegra_platform() -> bool:
+    if torch.cuda.get_device_capability() in [(8, 7), (7, 2)]:
+        return True
+    return False