Skip to content

[REFACTOR][IR] Initialize Python Op APIs from registered namespaces #20539

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

Merged
tqchen merged 5 commits into from
Oct 4, 2026
+516 −928
Merged

[REFACTOR][IR] Initialize Python Op APIs from registered namespaces #20539

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
b4d2a3e
[REFACTOR][IR] Initialize Python Op APIs from registered namespaces
tqchen Oct 3, 2026
76c87f9
[REFACTOR][IR] Reconcile canonical Op exposure with shared constructors
tqchen Oct 3, 2026
048d80d
[REFACTOR][IR] Keep Op API coverage in existing suites
tqchen Oct 3, 2026
1d925a0
[REFACTOR][CUDA] Chain metadata on intrinsic registrations
tqchen Oct 3, 2026
b793a05
[FIX][CUDA] Preserve composed leader-address construction
tqchen Oct 4, 2026
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
6 changes: 3 additions & 3 deletions python/tvm/backend/cuda/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,7 +73,7 @@ def register_backend():
pass
register_device_target_detector("cuda", _detect_target_from_device)
for name, namespace in script_namespaces().items():
builder_op.register_script_namespace(name, namespace)
builder_op.register_script_namespace(name, namespace, canonical_op_names=name == "cuda")

# script_namespaces() above pulls in ptx, which only imports the shared
# codegen layer -- not the device-helper modules. This import is the sole
Expand All @@ -85,16 +85,16 @@ def register_backend():

def script_namespaces(**_):
"""Return CUDA-owned TVMScript namespaces."""
from . import script # pylint: disable=import-outside-toplevel
from .ptx import PTXNamespace # pylint: disable=import-outside-toplevel
from .script import ( # pylint: disable=import-outside-toplevel
CUDANamespace,
NVSHMEMNamespace,
PTXLegacyNamespace,
STIRNamespace,
)

return {
"cuda": CUDANamespace(),
"cuda": script,
"nvshmem": NVSHMEMNamespace(),
"ptx_legacy": PTXLegacyNamespace(),
"ptx": PTXNamespace(),
Expand Down
Loading
Loading