Run TypeNN in your browser
Tensor Playground
Change the matrix values, choose an operation and backend, then run it through the TypeNN runtime bundled with this site.
TypeNN Runtime
Inputs
Browser support varies. A software renderer is not a physical GPU benchmark.
Result
ReLU(A × B)
Run an operation to see its output.
The result is read by an explicit data() call. Intermediate tensors are disposed after the read.
Equivalent TypeScriptTypeScript · TypeNN API
const device = await nn.device({ device: "auto" });
const a = device.tensor([1, 2, 3, 4], { shape: [2, 2] });
const b = device.tensor([5, 6, 7, 8], { shape: [2, 2] });
const output = a.matmul(b).relu();
const values = await output.data();What this runs: the shipped TypeNN build and its TypeShade runtime. It is not a JavaScript reimplementation or a mock result.
The example is intentionally small. Runtime startup includes backend preparation; this page does not report its duration as a benchmark. Automatic mode tries WebGPU and falls back to CPU. WebGL2 must be selected explicitly.