Skip to content

Fix brittle layer count #4576

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
jloftin-nv wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix brittle layer count #4576

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
47 changes: 12 additions & 35 deletions tests/py/ts/api/test_classes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,25 +7,6 @@
from torch_tensorrt.dynamo.runtime._TorchTensorRTModule import TorchTensorRTModule


def is_blackwell():
"""
Check if running on NVIDIA Blackwell architecture (sm_100+).

Blackwell architecture adds input/output reformat layers in TensorRT engines.

Returns:
bool: True if running on Blackwell (sm_100+), False otherwise
"""
if not torch.cuda.is_available():
return False

device_properties = torch.cuda.get_device_properties(0)
compute_capability = device_properties.major * 10 + device_properties.minor

# Blackwell is sm_100 and above
return compute_capability >= 100


@unittest.skipIf(
not torchtrt.ENABLED_FEATURES.torchscript_frontend,
"TorchScript Frontend is not available",
Expand Down Expand Up @@ -357,22 +338,12 @@ def test_get_layer_info(self):

import json

if is_blackwell():
# spellchecker:off
# blackwell has additional layers-
# Layer 0: __mye88_myl0_0 ← Input reformat layer
# Layer 1: aten__matmul(...) fc1 ← First matmul (fc1)
# Layer 2: aten__matmul(...) fc2 ← Second matmul (fc2)
# Layer 3: __mye90_myl0_3 ← Output reformat layer
# spellchecker:on
num_layers = 4
else:
num_layers = 2
for trt_mod in (
TestTorchTensorRTModule._get_trt_mod(),
TestTorchTensorRTModule._get_trt_mod(via_ts=True),
):
trt_json = json.loads(trt_mod.get_layer_info())
layer_info = trt_mod.get_layer_info()
trt_json = json.loads(layer_info)
self.assertIn("Layers", trt_json.keys(), "Key Layers is missing")
io_key = next(
(k for k in ("I/O Tensors", "Bindings") if k in trt_json.keys()),
Expand All @@ -381,10 +352,16 @@ def test_get_layer_info(self):
self.assertIsNotNone(
io_key, "Neither 'I/O Tensors' nor 'Bindings' key is present"
)
self.assertTrue(
len(trt_json["Layers"]) == num_layers
), "Not enough layers found"
self.assertTrue(len(trt_json[io_key]) == 2, "Not enough I/O tensors found")
self.assertGreater(
len(trt_json["Layers"]),
0,
f"No layers found in layer info: {layer_info}",
)
self.assertEqual(
len(trt_json[io_key]),
2,
f"Expected 2 I/O tensors, got {trt_json[io_key]}",
)


if __name__ == "__main__":
Expand Down
Loading