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 571–586

sqScoreGradient

Full file
P = 2^(t - L) and dS = P (dP - D), into S's registers (the products have finished: the first operation waits for them)
571def sqScoreGradient = (lambda unrestricted tail : (family SM86Program) .
572  (saFor 32 (lambda unrestricted i : Nat .
573    (let unrestricted nt = (naturalDivideUnchecked i 4) in (let unrestricted h = (naturalDivideUnchecked (naturalModuloUnchecked i 4) 2) in
574      (let unrestricted b = (naturalModuloUnchecked i 2) in
575      (let unrestricted e = (saElem (saS nt) h b) in
576      (lambda unrestricted rest : (family SM86Program) .
577        (saFfma e e saScale (sqNegL h)
578          (saEx2 e e saWaitNone rest))))))))
579  (saFor 32 (lambda unrestricted i : Nat .
580    (let unrestricted nt = (naturalDivideUnchecked i 4) in (let unrestricted h = (naturalDivideUnchecked (naturalModuloUnchecked i 4) 2) in
581      (let unrestricted b = (naturalModuloUnchecked i 2) in
582      (let unrestricted e = (saElem (saS nt) h b) in (let unrestricted g = (saElem (sqDP nt) h b) in
583      (lambda unrestricted rest : (family SM86Program) .
584        (saFadd g g (sqNegD h) saWaitNone
585          (saFmul e e g (naturalSelect (naturalIsZero i) saWait4 saWaitNone) rest)))))))))
586    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.