Update semi_structured_sparse.py

[catswe]

In my environment with an A100, I only see the speedup when enabling the following line

SparseSemiStructuredTensor._FORCE_CUTLASS = True

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3608

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[meta-cla]

Hi @catswe!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at [email protected]. Thanks!

[catswe]

If I run without forcing CUTLASS and use CUSPARSELT instead, the speedup can't be observed.

import torch
from torch.sparse import to_sparse_semi_structured, SparseSemiStructuredTensor
from torch.utils.benchmark import Timer
import torch
print("torch:", torch.__version__, "cuda:", torch.version.cuda)
print("gpu:", torch.cuda.get_device_name(0), "cc:", torch.cuda.get_device_capability(0))

# mask Linear weight to be 2:4 sparse
mask = torch.Tensor([0, 0, 1, 1]).tile((3072, 2560)).cuda().bool()
linear = torch.nn.Linear(10240, 3072).half().cuda().eval()
linear.weight = torch.nn.Parameter(mask * linear.weight)
print(linear.weight)

x = torch.rand(3072, 10240).half().cuda()

with torch.inference_mode():
    dense_output = linear(x)
    dense_t = Timer(stmt="linear(x)",
                    globals={"linear": linear,
                             "x": x}).blocked_autorange().median * 1e3

    # accelerate via SparseSemiStructuredTensor
    linear.weight = torch.nn.Parameter(to_sparse_semi_structured(linear.weight))
    print(linear.weight)

    sparse_output = linear(x)
    sparse_t = Timer(stmt="linear(x)",
                    globals={"linear": linear,
                             "x": x}).blocked_autorange().median * 1e3

    # sparse and dense matmul are numerically equivalent
    # On an A100 80GB, we see: `Dense: 0.870ms Sparse: 0.630ms | Speedup: 1.382x`
    assert torch.allclose(sparse_output, dense_output, atol=1e-3)
    print(f"Dense: {dense_t:.3f}ms Sparse: {sparse_t:.3f}ms | Speedup: {(dense_t / sparse_t):.3f}x")

torch: 2.8.0+cu126 cuda: 12.6
gpu: NVIDIA A100-SXM4-40GB cc: (8, 0)
Parameter containing:
tensor([[-0.0000,  0.0000,  0.0036,  ..., -0.0000,  0.0042,  0.0057],
        [-0.0000,  0.0000, -0.0010,  ...,  0.0000,  0.0083, -0.0043],
        [-0.0000, -0.0000, -0.0054,  ...,  0.0000, -0.0007, -0.0020],
        ...,
        [ 0.0000, -0.0000,  0.0007,  ..., -0.0000,  0.0033,  0.0074],
        [ 0.0000, -0.0000, -0.0079,  ..., -0.0000,  0.0071, -0.0024],
        [-0.0000, -0.0000, -0.0018,  ..., -0.0000, -0.0079, -0.0042]],
       device='cuda:0', dtype=torch.float16, requires_grad=True)
SparseSemiStructuredTensorCUSPARSELT(shape=torch.Size([3072, 10240]))
Dense: 0.803ms Sparse: 1.057ms | Speedup: 0.760x