For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

matmul_dispatch_sm100

def matmul_dispatch_sm100[c_type: DType, a_type: DType, b_type: DType, transpose_b: Bool = False, use_tf32: 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, elementwise_lambda_wrapper: Optional[def[dtype: DType, width: SIMDLength, *, alignment: Int = Int(1)](IndexList[Int(2)], SIMD[dtype, width]) capturing thin -> None] = None, elementwise_compute_lambda_fn: Optional[def[dtype: DType, width: SIMDLength, *, alignment: Int = Int(1)](IndexList[Int(2)], SIMD[dtype, width]) capturing thin -> SIMD[dtype, width]] = None, pdl_level: PDLLevel = PDLLevel()](c: TileTensor[c_type, Storage=c.Storage, address_space=c.address_space, linear_idx_type=c.linear_idx_type], a: TileTensor[a_type, Storage=a.Storage, address_space=a.address_space, linear_idx_type=a.linear_idx_type], b: TileTensor[b_type, Storage=b.Storage, address_space=b.address_space, linear_idx_type=b.linear_idx_type], ctx: DeviceContext)

Dispatches a 2D matmul to the appropriate SM100 (B200+) kernel.

Routes the problem to GEMV for M=1 or N=1 shapes, to the IEEE-fp32 split-K GEMV for precise float32, or to the dtype-specific SM100 dispatcher (bf16, fp8, fp32) for general shapes, falling back to vendor BLAS when no Mojo SM100 config applies. In autotuning mode, launches a single compile-time-configured kernel from environment defines.

Parameters:

c_type (DType): Output element type.
a_type (DType): Element type of the LHS operand a.
b_type (DType): Element type of the RHS operand b.
transpose_b (Bool): Whether b is stored transposed (defaults to False).
use_tf32 (Bool): Whether to allow TF32 (truncated mantissa) multiplies for float32 instead of requiring IEEE-fp32 precision (defaults to True).
elementwise_lambda_fn (Optional[def[dtype: DType, width: SIMDLength, *, alignment: Int = Int(1)](IndexList[Int(2)], SIMD[dtype, width]) capturing thin -> None]): Optional epilogue applied to each output element, passed to the SM100 GEMM path (defaults to None).
elementwise_lambda_wrapper (Optional[def[dtype: DType, width: SIMDLength, *, alignment: Int = Int(1)](IndexList[Int(2)], SIMD[dtype, width]) capturing thin -> None]): Optional epilogue lambda for GEMV and vendor fallback paths, folding in the compute lambda (defaults to None).
elementwise_compute_lambda_fn (Optional[def[dtype: DType, width: SIMDLength, *, alignment: Int = Int(1)](IndexList[Int(2)], SIMD[dtype, width]) capturing thin -> SIMD[dtype, width]]): Optional compute epilogue lambda, for example a static scale, passed to the SM100 GEMM path (defaults to None).
pdl_level (PDLLevel): Programmatic dependent launch level for the dispatched kernel (defaults to PDLLevel()).