Pipeline Ready — Select a workload or click Tune on My GPU
Matrix Multiply (SGEMM): 268.4M FLOPs • 3.1 MB
Compute Throughput Comparison (GFLOPS)
9-run hardware median — higher is faster • Matrix Multiply (SGEMM)
WinnerVerifiedDropped
IEEE-754 FP32 Throughput • 5 Warmup + 9 Timed Dispatches • Dawn/wgpu Architecture
Awaiting Kernel Benchmark
Run benchmark to generate, compile, verify, and crown the fastest kernel on your hardware.
Kernel Candidates
0 Variants TestedBaselineDROPPED
Naive Baseline
Not Dispatched
Dropped: Variant not dispatched or pipeline completed without return.
Quarantined in Failure Graveyard
Variant ADROPPED
Variant A: Workgroup Geometry
Not Dispatched
Dropped: Variant not dispatched or pipeline completed without return.
Quarantined in Failure Graveyard
Variant BDROPPED
Variant B: Shared Memory Tiling
Not Dispatched
Dropped: Variant not dispatched or pipeline completed without return.
Quarantined in Failure Graveyard
Variant CDROPPED
Variant C: Vec4 Vectorized
Not Dispatched
Dropped: Variant not dispatched or pipeline completed without return.
Quarantined in Failure Graveyard
Guardrail Status: Not yet run — awaiting kernel benchmark.
Not yet runShader & Artifact Inspector
1// No WGSL shader loaded yet.