Skip to content

Add NVTX marker instrumentation #2345

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

Open
iomaganaris wants to merge 5 commits into
base: main
Choose a base branch
from
+14 −1
Open

Add NVTX marker instrumentation #2345

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
ed529bc
Add NVTX marker instrumentation
iomaganaris Oct 24, 2025
494570b
Make sure that instrumentation reports from dace are not written in e…
iomaganaris Oct 27, 2025
9fc1764
Added comment for map entries
iomaganaris Oct 27, 2025
ad794c9
Merge branch 'main' into nvtx_markers
iomaganaris Dec 4, 2025
71a18bf
Merge branch 'main' into nvtx_markers
iomaganaris Dec 4, 2025
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
1 change: 1 addition & 0 deletions src/gt4py/next/metrics.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,7 @@
# Metric collection levels
DISABLED: Final[int] = 0
MINIMAL: Final[int] = 1
GPU_TX_MARKERS: Final[int] = 2
PERFORMANCE: Final[int] = 10
INFO: Final[int] = 30
VERBOSE: Final[int] = 50
Expand Down
8 changes: 8 additions & 0 deletions src/gt4py/next/program_processors/runners/dace/workflow/translation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -267,6 +267,14 @@ def add_instrumentation(sdfg: dace.SDFG, gpu: bool) -> None:
dace.Memlet(f"{output}[0]"),
)

if (config.COLLECT_METRICS_LEVEL == metrics.GPU_TX_MARKERS) and gpu:
Copy link
Contributor

@edopao edopao Oct 27, 2025

Choose a reason for hiding this comment

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

Your current check is at compile time. The check in the instrumentation above is at runtime. I propose to always add NVTX instrumentation, because the config.COLLECT_METRICS_LEVEL value is meant to be a runtime setting. It is still possible to skip SDFG instrumentation by setting use_metrics=False on the dace backend, at compile-time.

Copy link
Contributor Author

@iomaganaris iomaganaris Oct 27, 2025

Choose a reason for hiding this comment

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

I am not sure I would be up to enabling NVTX ranges by default since they will add some (really minimal) overhead to the CPU side. If use_metrics=False can be set by default then it would be fine but it would increase some more complexity for production and profiling runs

sdfg.instrument = dace.dtypes.InstrumentationType.GPU_TX_MARKERS
for node, _ in sdfg.all_nodes_recursive():
if isinstance(node, dace.nodes.MapEntry):
node.instrument = dace.dtypes.InstrumentationType.GPU_TX_MARKERS
elif isinstance(node, dace.sdfg.state.SDFGState):
node.instrument = dace.dtypes.InstrumentationType.GPU_TX_MARKERS

# Check SDFG validity after applying the above changes.
sdfg.validate()

Expand Down