For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Mojo function
matmul_dispatch_sm100_fp8
def matmul_dispatch_sm100_fp8[c_type: DType, a_type: DType, b_type: DType, ComputeFnType: ElementwiseComputeFn, //, transpose_b: Bool = True, elementwise_lambda_fn: Optional[def[dtype: DType, width: SIMDLength, *, alignment: Int = Int(1)](IndexList[Int(2)], SIMD[dtype, width]) capturing thin -> None] = None, pdl_level: PDLLevel = PDLLevel(), has_compute_fn: Bool = True](c: TileTensor[c_type, Engine=c.Engine, address_space=c.address_space, linear_idx_type=c.linear_idx_type], a: TileTensor[a_type, Engine=a.Engine, address_space=a.address_space, linear_idx_type=a.linear_idx_type], b: TileTensor[b_type, Engine=b.Engine, address_space=b.address_space, linear_idx_type=b.linear_idx_type], compute_fn: ComputeFnType, ctx: DeviceContext) -> Int
Dispatches an FP8-input SM100 matmul to a tuned or heuristic config.
For M <= 128, tries the heuristic outlier dispatch first and falls back to
the default SM100 config on a miss. For larger M, searches the FP8 tuning
table by static M bucket, then falls through to the heuristic outlier
dispatch for untuned (N, K) shapes. Only bfloat16 output is supported.
compute_fn is applied to each output value when has_compute_fn is
set and ignored otherwise.
Returns: