Improve binary sizes by not globing MLIR dialects/passes #2414
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
boschmitt
force-pushed
the
improve_bin_sizes
branch
from
ba8fd31 to
8f43fee
Compare
schweitzpgi
reviewed
Nov 25, 2024
boschmitt
force-pushed
the
improve_bin_sizes
branch
3 times, most recently
from
e17b1c3 to
fe760c2
Compare
boschmitt
force-pushed
the
improve_bin_sizes
branch
from
fe760c2 to
4d8da08
Compare
By not globing all MLIR dialects/passes, many of which CUDAQ does not use, we can significantly improve binary sizes: ``` Filename Size (old) Size (new) Improvement -------- ----------- ----------- ----------- cudaq-opt 161.9 MB 39.6 MB -122.2 MB cudaq-quake 187.6 MB 108.6 MB -79.0 MB cudaq-translate 138.2 MB 76.1 MB -62.0 MB libcudaq-mlir-runtime.so 179.9 MB 111.8 MB -68.1 MB ``` This change also improves compilation time when building `llvm` along with CUDAQ, instead of using a pre-built version for llvm. It also improves the `--help` output for the `cudaq-*` tools, as they won't be "polluted" with many passes/options that are not relevant in the context of `nvq++`. Improving the python bindings situation, which can possibly profit from a similar change, will be left to a later commit. Signed-off-by: boschmitt <[email protected]>
boschmitt
force-pushed
the
improve_bin_sizes
branch
from
4d8da08 to
b1db6fe
Compare
bettinaheim
changed the title
bettinaheim
added
the
release notes
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
By not globing all MLIR dialects/passes, many of which CUDAQ does not use, we can significantly improve binary sizes:
This change also improves compilation time when building
llvmalong with CUDAQ, instead of using a pre-built version for llvm. It also improves the--helpoutput for thecudaq-*tools, as they won't be "polluted" with many passes/options that are not relevant in the context ofnvq++.Improving the python bindings situation, which can possibly profit from a similar change, will be left to a later commit.
This PR is part of a series that address the long standing need to refactor the build system.