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

sum

def sum[dtype: DType, BN: Int, //, *, width: Int = Int(8)](x: TileTensor[dtype, Layout[TypeList[ComptimeInt[BN]](), TypeList[ComptimeInt[Int(1)]]()], MutUntrackedOrigin, Engine=x.Engine, address_space=AddressSpace.LOCAL]) -> SIMD[dtype, 2]

Reduces a BN-element local tensor into a width-2 SIMD vector via vectorized accumulation.

Parameters:

• ​dtype (DType): Element dtype of the input tensor (inferred).
• ​BN (Int): Number of elements in the input tensor; must be divisible by width (inferred).
• ​width (Int): Vectorization width for the accumulation (defaults to 8).

Args:

• ​x (TileTensor[dtype, Layout[TypeList[ComptimeInt[BN]](), TypeList[ComptimeInt[Int(1)]]()], MutUntrackedOrigin, Engine=x.Engine, address_space=AddressSpace.LOCAL]): Local tensor of BN elements to reduce.

Returns:

SIMD[dtype, 2]