Skip to content
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

Fix cudagraph mem #112

Closed
wants to merge 3 commits into from
Closed

Fix cudagraph mem #112

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9102236
fix cudagraph mem
FindHao Dec 11, 2024
1aa0c89
Revert "fix cudagraph mem"
FindHao Dec 11, 2024
7fd089f
fix do_bench_cudagraph_mem
FindHao Dec 11, 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
43 changes: 20 additions & 23 deletions tritonbench/utils/triton_op.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1265,31 +1265,28 @@ def _init_extra_metrics() -> Dict[str, Any]:
def do_bench_cudagraph_mem(
self, fn, n_repeat=2, grad_to_none=None, device_type="cuda"
):
if torch.cuda.current_stream() == torch.cuda.default_stream():
raise RuntimeError(
"Cannot capture graph in default stream. Please use side stream in benchmark code."
)
# warmup
fn()
if grad_to_none is not None:
for x in grad_to_none:
x.detach_()
x.requires_grad_(True)
x.grad = None
g = torch.cuda.CUDAGraph()
with torch.cuda.graph(g):
with torch.cuda.stream(torch.cuda.Stream()):
# warmup
fn()
torch.cuda.synchronize()
g.replay()
torch.cuda.synchronize()
g = torch.cuda.CUDAGraph()
with torch.cuda.graph(g):
for _ in range(n_repeat):
if grad_to_none is not None:
for x in grad_to_none:
x.grad = None
if grad_to_none is not None:
for x in grad_to_none:
x.detach_()
x.requires_grad_(True)
x.grad = None
g = torch.cuda.CUDAGraph()
with torch.cuda.graph(g):
fn()
torch.cuda.synchronize()
torch.cuda.synchronize()
g.replay()
torch.cuda.synchronize()
g = torch.cuda.CUDAGraph()
with torch.cuda.graph(g):
for _ in range(n_repeat):
if grad_to_none is not None:
for x in grad_to_none:
x.grad = None
fn()
torch.cuda.synchronize()

def do_bench_mem(self, fn, n_repeat=2, grad_to_none=None, device_type="cuda"):
di = torch._dynamo.device_interface.get_interface_for_device(device_type)
Expand Down
Loading