pub const MVMUL_PER_TILE_MATMUL_PER_PASS: u64 = 16;Expand description
MVMUL issues one 32x32 tile-matmul retires PER FIDELITY PASS. The matrix unit walks the tile in 16 steps, and math fidelity is literally how many times it walks it: LoFi takes the operands’ top mantissa bits in one pass, HiFi4 makes four. So the fidelity knob is not a fitted scale factor on the compute term — it is an exact integer multiplier on the work, and the machine says so.
MEASURED at 640^3, same program, three fidelities: LoFi 128,000 / HiFi2 256,000 / HiFi4 512,000 retired, against 8,000 tile-matmuls. 16, 32 and 64 per tile-matmul exactly.