Source/Packages

Realization.Nvidia.SM86.StreamingAttentionSM86

packages/realizations/cooperative/nvidia-sm86/src/Realization/Nvidia/SM86/StreamingAttentionSM86.alpha

1,129 lines205 declarations66.1 KiBSHA-256 aefcbc3b00ca

def · lines 550–557

sqLoadRows

Full file
a tile's rows (K's or V's) as B fragments, the first load waiting on `wait`
550def sqLoadRows = (lambda unrestricted pointer : Nat . (lambda unrestricted wait : Nat . (lambda unrestricted tail : (family SM86Program) .
551  (saFor 32 (lambda unrestricted i : Nat .
552    (let unrestricted k = (naturalDivideUnchecked i 8) in (let unrestricted nt = (naturalModuloUnchecked i 8) in
553      (lambda unrestricted rest : (family SM86Program) .
554        (saLoad (saB k nt) pointer (naturalAdd (naturalMultiply nt 1024) (naturalMultiply k 32)) saSB0 (naturalSelect (naturalIsZero i) wait saWaitNone)
555        (saLoad (succ (saB k nt)) pointer (naturalAdd (naturalMultiply nt 1024) (naturalAdd (naturalMultiply k 32) 16)) saSB0 saWaitNone
556          rest))))))
557    tail))))

The compiler supplied declaration spans and resolved links from this source snapshot. This page does not assert that this file belongs to a checked closure.