Skip to content
KernelIndex
Search⌘K

torch.compile (inductor)

PyTorch · python · MIT

Use it

Vendorable · source mirrored · MITView source →

No package. Vendor the mirrored source: 26 lines, MIT.

99_Matmul_GELU_Softmax.py
curl "https://kernelindex.com/api/v1/implementations/kernelbench-l2-99-matmul-gelu-softmax-torch-compile-inductor?include=source"
interfacepython · torch_compile_inductor
symbolModel.forward
Compatibility
measured onNVIDIA H100
declared hardwaredeclared only
architectures—
dtypes

Benchmark evidence

2 measurements across 1 GPU, fastest first.

Operation / workload
Hardware
Latency
Rank
Observed
Matmul GELU Softmaxfp32 · [1024, 8192]
NVIDIA H100
2.78ms±0.00
#2 of 2
2026-03-05
Matmul GELU Softmaxfp32 · [1024, 8192]
NVIDIA H100
5.16ms±0.02
#2 of 2
2026-03-05

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:98f32d9bae95f803914b665f51aa215dccbd7fadc31c6280dd2b0ca5b19bdff9
license declaredMIT
license concludedMIT
imported2026-08-26

Kernel source

99_Matmul_GELU_Softmax.py26 lines
import torch
import torch.nn as nn

class Model(nn.Module):
    """
    Simple model that performs a matrix multiplication, applies GELU, and then applies Softmax.
    """
    def __init__(self, in_features, out_features):
        super(Model, self).__init__()
        self.linear = nn.Linear(in_features, out_features)

    def forward(self, x):
        x = self.linear(x)
        x = torch.nn.functional.gelu(x)
        x = torch.nn.functional.softmax(x, dim=1)
        return x

batch_size = 1024
in_features = 8192
out_features = 8192

def get_inputs():
    return [torch.rand(batch_size, in_features)]

def get_init_inputs():
    return [in_features, out_features]

Source code from KernelBench, © 2023 Anne Ouyang, Simon Guo, Azalia Mirhoseini (Scaling Intelligence Lab, Stanford University), MIT License · MIT

Best evidence level for this revision: reported

JSON