List your company

ResearchFormerly WizWand

Matrix Multiplication

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

KernelBench FP4 Matmul

AutoKernel (Triton)

2,898Throughput (TF/s)

14

Mar 24, 2026

Synthetic Transformer Shapes Query-Key Q ⊗ K⊤

5.41Latency (µs)

9

Apr 7, 2026

Synthetic Transformer Shapes Attention-Value Att ⊗ V

6.38Latency (µs)

9

Apr 7, 2026

CUDA-LLM kernels task suite Matrix Multiplication

4.4Execution Time (s)

9

Feb 27, 2026

CUDA-LLM kernels task suite 1.0 (test)

4.4Execution Time

9

Feb 27, 2026

Asymmetric Matrix Multiplication S1-S5 on H200

0.211Indicator S1

5

Apr 23, 2026

Simulated Matrix Multiplication Sequence Lengths 1K-32K

27.63Latency (ms)

3

Apr 27, 2026

General Matrices

AHAT-based architecture

3Layers (L)

2

Apr 23, 2026

Matrix Multiplication 3x3, 4x4, 5x5

4Circuit Depth

2

Feb 26, 2026

CUDA-LLM kernels task suite (test)

—

—Execution Time (s)

0

Feb 26, 2026