Skip to content

[AUTOTUNER] adding simple report flag for autotuner runs #3411

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 4 commits into from
Apr 2, 2024
Merged

[AUTOTUNER] adding simple report flag for autotuner runs #3411

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ab55087
[AUTOTUNER] adding simple report flag for autotuner runs
bringlein Mar 19, 2024
661a373
use existing bench_time; rename flag; update docstring
bringlein Mar 20, 2024
9b740cf
autotuner print stats controlled by env variable TRITON_PRINT_AUTOTUNING
bringlein Mar 27, 2024
23308be
Address review comments
jlebar Apr 2, 2024
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
6 changes: 6 additions & 0 deletions python/triton/runtime/autotuner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
from __future__ import annotations

import builtins
import os
import time
from typing import Dict

Expand Down Expand Up @@ -109,6 +110,7 @@ def kernel_call():

def run(self, *args, **kwargs):
self.nargs = dict(zip(self.arg_names, args))
used_cached_result = True
if len(self.configs) > 1:
all_args = {**self.nargs, **kwargs}
_args = []
Expand All @@ -122,6 +124,7 @@ def run(self, *args, **kwargs):
key = tuple(key)
if key not in self.cache:
# prune configs
used_cached_result = False
pruned_configs = self.prune_configs(kwargs)
bench_start = time.time()
timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
Expand All @@ -134,6 +137,9 @@ def run(self, *args, **kwargs):
else:
config = self.configs[0]
self.best_config = config
if os.getenv("TRITON_PRINT_AUTOTUNING", None) == "1" and not used_cached_result:
print(f"Triton autotuning for function {self.fn} finished after "
f"{self.bench_time:.2f}s; best config selected: {self.best_config};")
full_nargs = {**self.nargs, **kwargs, **self.best_config.kwargs}
if config.pre_hook is not None:
config.pre_hook(full_nargs)
Expand Down