Skip to main content

Module op

Module op 

Source
Expand description

Canonical op names. Free-form &'static str is accepted everywhere, but sticking to these keeps the report groupable.

Constantsยง

ENGINE_PREFILL_BATCH
An engine-level capability rather than a per-tensor kernel: does this engine have a batched prefill path at all?
FFN_SWIGLU
Whole gate/up/down SwiGLU fused into one dispatch.
GEMM_PREFILL
A real batched GEMM over batch activations (prefill).
MATVEC
One activation against a whole weight matrix (decode).
MATVEC_MULTI
Several independent matvecs sharing one activation (fused q/k/v).