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 w8a8 benchmark and add Llama-3-8B #5562

Merged
merged 3 commits into from
Jun 17, 2024
Merged

Fix w8a8 benchmark and add Llama-3-8B #5562

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
fc6c50f
Fix w8a8 benchmark and add Llama-3-8B
comaniac Jun 14, 2024
4a62a2f
format
comaniac Jun 14, 2024
c85abf8
trigger ci
comaniac Jun 16, 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
21 changes: 13 additions & 8 deletions benchmarks/cutlass_benchmarks/w8a8_benchmarks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ def make_rand_tensors(dtype: torch.dtype, m: int, n: int,
# impl


def pytorch_i8_impl(a: torch.tensor, b: torch.tensor, scale_a: torch.tensor,
def pytorch_mm_impl(a: torch.tensor, b: torch.tensor, scale_a: torch.tensor,
scale_b: torch.tensor,
out_dtype: torch.dtype) -> torch.tensor:
return torch.mm(a, b)
Expand Down Expand Up @@ -115,7 +115,7 @@ def bench_int8(dtype: torch.dtype, m: int, k: int, n: int, label: str,
timers.append(
bench_fn(a.to(dtype=torch.bfloat16, device="cuda"),
b.to(dtype=torch.bfloat16, device="cuda"), scale_a, scale_b,
torch.bfloat16, label, sub_label, pytorch_i8_impl,
torch.bfloat16, label, sub_label, pytorch_mm_impl,
"pytorch_bf16_bf16_bf16_matmul-no-scales"))

# cutlass impl
Expand All @@ -136,6 +136,13 @@ def bench_fp8(dtype: torch.dtype, m: int, k: int, n: int, label: str,

timers = []

# pytorch impl w. bf16
timers.append(
bench_fn(a.to(dtype=torch.bfloat16, device="cuda"),
b.to(dtype=torch.bfloat16, device="cuda"), scale_a, scale_b,
torch.bfloat16, label, sub_label, pytorch_mm_impl,
"pytorch_bf16_bf16_bf16_matmul-no-scales"))

# pytorch impl: bf16 output, without fp8 fast accum
timers.append(
bench_fn(a, b, scale_a, scale_b, torch.bfloat16, label, sub_label,
Expand All @@ -160,14 +167,12 @@ def bench_fp8(dtype: torch.dtype, m: int, k: int, n: int, label: str,

# cutlass impl: bf16 output
timers.append(
bench_fn(a, b, scale_a.to(device="cpu"), scale_b.to(device="cpu"),
torch.bfloat16, label, sub_label, cutlass_impl,
"cutlass_fp8_fp8_bf16_scaled_mm"))
bench_fn(a, b, scale_a, scale_b, torch.bfloat16, label, sub_label,
cutlass_impl, "cutlass_fp8_fp8_bf16_scaled_mm"))
# cutlass impl: fp16 output
timers.append(
bench_fn(a, b, scale_a.to(device="cpu"), scale_b.to(device="cpu"),
torch.float16, label, sub_label, cutlass_impl,
"cutlass_fp8_fp8_fp16_scaled_mm"))
bench_fn(a, b, scale_a, scale_b, torch.float16, label, sub_label,
cutlass_impl, "cutlass_fp8_fp8_fp16_scaled_mm"))
return timers


Expand Down
6 changes: 6 additions & 0 deletions benchmarks/cutlass_benchmarks/weight_shapes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,12 @@
([4096, 22016], 1),
([11008, 4096], 0),
],
"meta-llama/Llama-3-8b": [
([4096, 6144], 1),
([4096, 4096], 0),
([4096, 28672], 1),
([14336, 4096], 0),
],
"meta-llama/Llama-2-13b-hf": [
([5120, 15360], 1),
([5120, 5120], 0),
Expand Down
Loading