[PyTorch] Minor optimizations to reduce CPU overheads in modules - #1191

Merged

timmoon10 merged 20 commits into

NVIDIA:mainfrom

timmoon10:linear-cpu-perf-optimization

Oct 4, 2024

Commits on Sep 17, 2024

CPU perf optimization in linear autograd function
timmoon10
committed
7789c7e
CPU perf optimization in prepare_forward function
timmoon10
committed
89628e0

Commits on Sep 18, 2024

Avoid module import in TE module forwards
timmoon10
committed
adab597
Merge branch 'main' into linear-cpu-perf-optimization
timmoon10
committed
fffbcf8
Use fast getter for params
timmoon10
committed
a4f2b3f

Commits on Sep 19, 2024

Reuse tensor dims in linear autograd func
timmoon10
committed
1b4ef9f
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot]
committed
2c464c5
Merge branch 'main' into linear-cpu-perf-optimization
timmoon10
authored
5905b01

Commits on Sep 20, 2024

Merge branch 'main' into linear-cpu-perf-optimization
timmoon10
committed
4e6d75b
Apply optimizations to grouped linear
timmoon10
committed
1763bdf

Commits on Sep 25, 2024

Merge branch 'main' into linear-cpu-perf-optimization
timmoon10
committed
ab6551a
Debug test failures
timmoon10
committed
92c443e
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot]
committed
d54cb00

Commits on Sep 27, 2024

Debug test failures
timmoon10
committed
b1db2ed
Merge branch 'main' into linear-cpu-perf-optimization
timmoon10
committed
c973cff
Fix linter warnings
timmoon10
committed
df8e7b0

Commits on Oct 1, 2024

Avoid deepcopy in tests
timmoon10
committed
706dcdf
Merge branch 'main' into linear-cpu-perf-optimization
timmoon10
committed
b29bae5

Commits on Oct 2, 2024

Move _fast_setattr logic to __setattr__ method
timmoon10
committed
ad13c35
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot]
committed
9458580