Source/Packages

Accelerator.SM86.Operands

packages/hardware/architectures/nvidia-sm86/src/Accelerator/SM86/Operands.alpha

699 lines54 declarations58.4 KiBSHA-256 ac21d61cced1

def · lines 542–589

sm86MinimumStall

Full file
542def sm86MinimumStall =
543  (lambda unrestricted body : (family SM86InstructionBody) .
544    (eliminate SM86InstructionBody (lambda unrestricted current : (family SM86InstructionBody) . Nat) body
545      (branch SM86MoveConstant destination bank offset control . 1)
546      (branch SM86SpecialToRegister destination source control . 1)
547      (branch SM86MoveImmediate destination immediate control . 1)
548      (branch SM86IntegerMultiplyAddConstant destination left bank offset addend control . 1)
549      (branch SM86IntegerMultiplyAddImmediate destination left immediate addend control . 1)
550      (branch SM86IntegerMultiplyAddWideConstant destination left right bank offset control . 1)
551      (branch SM86IntegerAddThreeImmediate destination left immediate control . 1)
552      (branch SM86IntegerAddThreeRegister destination left right control . 1)
553      (branch SM86ShiftRightImmediate destination source amount control . 1)
554      (branch SM86LogicThreeInputTruthTable destination left right truthTable control . 1)
555      (branch SM86IntegerToFloat destination source control . 1)
556      (branch SM86FloatAdd destination left right control . 1)
557      (branch SM86FloatMultiply destination left right control . 1)
558      (branch SM86Float64Binary operation destination left right control . 1)
559      (branch SM86FloatConvert conversion control . 1)
560      (branch SM86FloatFusedMultiplyAdd destination left right addend control . 1)
561      (branch SM86MultiFunctionUnitApproximation destination source operation control . 1)
562      (branch SM86FloatMinimumOrMaximum destination left right mode control . 1)
563      (branch SM86FloatNegate destination source control . 1)
564      (branch SM86FloatPairToPackedHalfPair destination high low control . 1)
565      (branch SM86FloatPairToPackedBFloat16Pair destination high low control . 1)
566      (branch SM86HalfToFloat destination source selector control . 1)
567      (branch SM86BFloat16ToFloat destination source selector control . 1)
568      (branch SM86TensorCoreHalfMatrixMultiplyAccumulate16x8x16Float32 destination a b accumulator control .
569        (naturalSelect (sm86ControlWritesBarrier control) 1 sm86TensorIssueStall))
570      (branch SM86TensorCoreBFloat16MatrixMultiplyAccumulate16x8x16Float32 destination a b accumulator control .
571        (naturalSelect (sm86ControlWritesBarrier control) 1 sm86TensorIssueStall))
572      (branch SM86LoadGlobal destination address offset control . 1)
573      (branch SM86LoadGlobalWide destination address offset control . 1)
574      (branch SM86WarpShuffle destination source lane segment mode control . 1)
575      (branch SM86LoadShared destination address offset control . 1)
576      (branch SM86LoadSharedMatrix destination address offset count transpose control . 1)
577      (branch SM86StoreShared address value offset control . 1)
578      (branch SM86LoadGlobalToShared address offset source sourceOffset control . 1)
579      (branch SM86CommitAsyncGroup control . 1)
580      (branch SM86WaitAsyncGroups count control . sm86WaitAsyncGroupsStall)
581      (branch SM86BarrierSynchronize control . sm86BarrierSynchronizeStall)
582      (branch SM86BarrierSynchronizeCount barrier threads control . sm86BarrierSynchronizeStall)
583      (branch SM86StoreGlobal address value offset control . 1)
584      (branch SM86StoreGlobalWide address value offset control . 1)
585      (branch SM86StoreGlobal64 address value offset control . 1)
586      (branch SM86ReduceGlobalAddFloat32 address value offset control . 1)
587      (branch SM86PredicateGreaterThanImmediate predicate source immediate control . 1)
588      (branch SM86Branch offset descriptor control . 1)
589      (branch SM86Exit control . 1)))

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.