Skip to content

Pytato exec: don't wait #309

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
inducer merged 1 commit into from
Mar 28, 2025
Merged

Pytato exec: don't wait #309

Show file tree
Hide file tree
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
3 changes: 1 addition & 2 deletions arraycontext/impl/pytato/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -546,10 +546,9 @@ def _to_frozen(key: tuple[Any, ...], ary) -> TaggableCLArray:
self._dag_transform_cache[normalized_expr])

assert len(pt_prg.bound_arguments) == 0
evt, out_dict = pt_prg(self.queue,
_evt, out_dict = pt_prg(self.queue,
allocator=self.allocator,
**bound_arguments)
evt.wait()
assert len(set(out_dict) & set(key_to_frozen_subary)) == 0

key_to_frozen_subary = {
Expand Down
14 changes: 2 additions & 12 deletions arraycontext/impl/pytato/compile.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -636,15 +636,10 @@ def __call__(self, arg_id_to_arg) -> ArrayContainer:
input_kwargs_for_loopy = _args_to_device_buffers(
self.actx, self.input_id_to_name_in_program, arg_id_to_arg, fn_name)

evt, out_dict = self.pytato_program(queue=self.actx.queue,
_evt, out_dict = self.pytato_program(queue=self.actx.queue,
allocator=self.actx.allocator,
**input_kwargs_for_loopy)

# FIXME Kernels (for now) allocate tons of memory in temporaries. If we
# race too far ahead with enqueuing, there is a distinct risk of
# running out of memory. This mitigates that risk a bit, for now.
evt.wait()

def to_output_template(keys, _):
name_in_program = self.output_id_to_name_in_program[keys]
return self.actx.thaw(to_tagged_cl_array(
Expand Down Expand Up @@ -680,15 +675,10 @@ def __call__(self, arg_id_to_arg) -> ArrayContainer:
input_kwargs_for_loopy = _args_to_device_buffers(
self.actx, self.input_id_to_name_in_program, arg_id_to_arg, fn_name)

evt, out_dict = self.pytato_program(queue=self.actx.queue,
_evt, out_dict = self.pytato_program(queue=self.actx.queue,
allocator=self.actx.allocator,
**input_kwargs_for_loopy)

# FIXME Kernels (for now) allocate tons of memory in temporaries. If we
# race too far ahead with enqueuing, there is a distinct risk of
# running out of memory. This mitigates that risk a bit, for now.
evt.wait()

return self.actx.thaw(to_tagged_cl_array(out_dict[self.output_name],
axes=get_cl_axes_from_pt_axes(
self.output_axes),
Expand Down
Loading