Skip to content

[mlir][python] remove eager loading of dialect module (for type and value casting) #72338

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
merged 2 commits into from
Nov 21, 2023
Merged

[mlir][python] remove eager loading of dialect module (for type and value casting) #72338

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8ae7d2d
[mlir][python] automatically bundle builtin dialect with core
makslevental Nov 15, 2023
dc6a222
[mlir][python] loading dialect module is best effort only
makslevental Nov 15, 2023
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
9 changes: 4 additions & 5 deletions mlir/lib/Bindings/Python/IRModule.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -132,10 +132,8 @@ PyGlobals::lookupAttributeBuilder(const std::string &attributeKind) {

std::optional<py::function> PyGlobals::lookupTypeCaster(MlirTypeID mlirTypeID,
MlirDialect dialect) {
// Make sure dialect module is loaded.
if (!loadDialectModule(unwrap(mlirDialectGetNamespace(dialect))))
return std::nullopt;

// Try to load dialect module.
(void)loadDialectModule(unwrap(mlirDialectGetNamespace(dialect)));
Comment on lines +135 to +136
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can't really think of a nice way to test this, so can you at least leave a comment here about why we continue on instead of returning.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No one has noticed it, but the Python API overhead for all of the fancy stuff is raising on my hit list. There are a wide swatch of use cases that need no frills, and I worry about the overhead of this kind of thing. For some cases, I've already wanted a big switch that just turns off all OpViews, for example.

const auto foundIt = typeCasterMap.find(mlirTypeID);
if (foundIt != typeCasterMap.end()) {
assert(foundIt->second && "type caster is defined");
Expand All @@ -146,7 +144,8 @@ std::optional<py::function> PyGlobals::lookupTypeCaster(MlirTypeID mlirTypeID,

std::optional<py::function> PyGlobals::lookupValueCaster(MlirTypeID mlirTypeID,
MlirDialect dialect) {
loadDialectModule(unwrap(mlirDialectGetNamespace(dialect)));
// Try to load dialect module.
(void)loadDialectModule(unwrap(mlirDialectGetNamespace(dialect)));
const auto foundIt = valueCasterMap.find(mlirTypeID);
if (foundIt != valueCasterMap.end()) {
assert(foundIt->second && "value caster is defined");
Expand Down