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
maximum
def maximum[BN: Int, //, *, width: Int = Int(4)](x: Array[Float32, BN]) -> StaticTuple[Float32, width]
Reduces BN float32 scores into width lane-maxima using FTZ max.
Returns:
def maximum[BN: Int, //, *, width: Int = Int(4)](x: Array[Float32, BN], init: StaticTuple[Float32, width]) -> StaticTuple[Float32, width]
Reduces BN float32 scores into width lane-maxima, seeded from init.
Returns:
def maximum(x: StaticTuple[Float32, Int(4)]) -> Float32
Returns the maximum of four float32 values packed in a StaticTuple.
Returns:
def maximum(x: StaticTuple[Float32, Int(4)], init: Float32) -> Float32
Returns the FTZ maximum of a StaticTuple[4] and an initial value.
Returns:
def maximum(x: StaticTuple[Float32, Int(8)]) -> Float32
Returns the maximum of eight float32 values packed in a StaticTuple.
Returns:
def maximum(x: StaticTuple[Float32, Int(8)], init: Float32) -> Float32
Returns the FTZ maximum of a StaticTuple[8] and an initial value.
Returns:
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!