GPU runtime optimisation #896
Open
+651
−191
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
|
@slab-ci compiler-cpu-build-distributed |
|
@slab-ci compiler-cpu-build |
antoniupop
force-pushed
the
antoniu/gpu_optimization_wip
branch
3 times, most recently
from
93575ee
to
13dc8d8
Compare
antoniupop
force-pushed
the
antoniu/gpu_optimization_wip
branch
5 times, most recently
from
1a6375b
to
6b6cf2a
Compare
antoniupop
force-pushed
the
antoniu/gpu_optimization_wip
branch
from
6b6cf2a
to
c3a776a
Compare
BourgerieQuentin
requested changes
Jul 4, 2024
compilers/concrete-compiler/compiler/lib/Dialect/SDFG/Transforms/SDFGBufferOwnership.cpp
Outdated
Show resolved
Hide resolved
compilers/concrete-compiler/compiler/lib/Dialect/SDFG/Transforms/SDFGBufferOwnership.cpp
Outdated
Show resolved
Hide resolved
…ns on multiple keys in KS/BS operations for GPU.
…s to the GPU backend when already lowering through the SDFG dialect. Otherwise this forces GPU offload even for very fine granularity operations, bypassing SDFG and Batching.
…Add environment variable CONCRETELANG_TIMING_ENABLED as prerequisite to activation of timing logs.
…he SDFG runtime on Put operations. This allows to avoid an extra copy of the data for use in asynchronous operations.
…rging SDFG batch outputs in place.
… by default and use runtime environment variable to activate as required.
antoniupop
force-pushed
the
antoniu/gpu_optimization_wip
branch
from
023db39
to
2c244df
Compare
antoniupop
force-pushed
the
antoniu/gpu_optimization_wip
branch
from
2c244df
to
3346127
Compare
BourgerieQuentin
approved these changes
Jul 5, 2024
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Outdated suggestions cannot be applied.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.
This PR reduces the number of copy operations by transferring ownership of data buffers to the runtime system, and by merging the outputs of SDFG compute graphs in place.
It also fixes some bugs and minor issues.