• Slang Standard Library Reference
    • Interfaces
    • Types
    • Attributes
    • Global Declarations
      • Atomic functions
      • Memory and control barriers
      • Bit operation functions
      • Conversion functions
      • Derivative functions
      • Vertex Interpolation Functions
      • Math functions
      • Mesh shading
      • Ray-tracing
      • Tessellation functions
      • Wave and quad functions
        • QuadReadAcrossDiagonal
        • QuadReadAcrossX
        • QuadReadAcrossY
        • QuadReadLaneAt
        • WaveActiveAllEqual
        • WaveActiveAllTrue
        • WaveActiveAnyTrue
        • WaveActiveBallot
        • WaveActiveBitAnd
        • WaveActiveBitOr
        • WaveActiveBitXor
        • WaveActiveCountBits
        • WaveActiveMax
        • WaveActiveMin
        • WaveActiveProduct
        • WaveActiveSum
        • WaveBroadcastLaneAt
        • WaveGetActiveMulti
        • WaveGetConvergedMulti
        • WaveGetLaneCount
        • WaveGetLaneIndex
        • WaveIsFirstLane
        • WaveMatch
        • WaveMultiPrefixBitAnd
        • WaveMultiPrefixBitOr
        • WaveMultiPrefixBitXor
        • WaveMultiPrefixCountBits
        • WaveMultiPrefixProduct
        • WaveMultiPrefixSum
        • WavePrefixCountBits
        • WavePrefixProduct
        • WavePrefixSum
        • WaveReadLaneAt
        • WaveReadLaneFirst
        • WaveShuffle
        • _WaveCountBits
      • CheckAccessFullyMapped
      • D3DCOLORtoUBYTE4
      • GetAttributeAtVertex
      • GetRenderTargetSampleCount
      • GetRenderTargetSamplePosition
      • IsHelperLane
      • NonUniformResourceIndex
      • QuadAll
      • QuadAny
      • ReorderThread
      • WaveClusteredRotate
      • WaveRotate
      • WorkgroupSize
      • abort
      • all
      • any
      • bitfieldExtract
      • bitfieldInsert
      • clip
      • clock2x32ARB
      • clockARB
      • concat
      • coopVecLoad
      • coopVecLoadGroupshared
      • coopVecMatMul
      • coopVecMatMulAdd
      • coopVecMatMulAddPacked
      • coopVecMatMulPacked
      • coopVecOuterProductAccumulate
      • coopVecReduceSumAccumulate
      • createDynamicObject
      • cudaBlockDim
      • cudaBlockIdx
      • cudaThreadIdx
      • debugBreak
      • defaultGetDescriptorFromHandle
      • detach
      • diffPair
      • getDescriptorFromHandle
      • getRealtimeClock
      • getRealtimeClockLow
      • getStringHash
      • isDifferentialNull
      • loadAligned
      • makeArrayFromElement
      • makeTuple
      • nextafter
      • nonuniform
      • operator*
      • packHalf2x16
      • packInt4x8
      • packInt4x8Clamp
      • packSnorm2x16
      • packSnorm4x8
      • packUint4x8
      • packUint4x8Clamp
      • packUnorm2x16
      • packUnorm4x8
      • pack_clamp_s8
      • pack_clamp_u8
      • pack_s8
      • pack_u8
      • printf
      • select
      • static_assert
      • storeAligned
      • syncTorchCudaStream
      • unmodified
      • unpackHalf2x16ToFloat
      • unpackHalf2x16ToHalf
      • unpackInt4x8ToInt16
      • unpackInt4x8ToInt32
      • unpackSnorm2x16ToFloat
      • unpackSnorm2x16ToHalf
      • unpackSnorm4x8ToFloat
      • unpackSnorm4x8ToHalf
      • unpackUint4x8ToUint16
      • unpackUint4x8ToUint32
      • unpackUnorm2x16ToFloat
      • unpackUnorm2x16ToHalf
      • unpackUnorm4x8ToFloat
      • unpackUnorm4x8ToHalf
      • unpack_s8s16
      • unpack_s8s32
      • unpack_u8u16
      • unpack_u8u32
      • unused
      • updateDiff
      • updatePair
      • updatePrimal
      • workgroupUniformLoad

WaveMultiPrefixSum

Description

Signature

T WaveMultiPrefixSum<T>(
    T value,
    vector<uint, 4> mask)
    where T : __BuiltinArithmeticType;

vector<T, N> WaveMultiPrefixSum<T, N:int>(
    vector<T, N> value,
    vector<uint, 4> mask)
    where T : __BuiltinArithmeticType;

matrix<T, N, M> WaveMultiPrefixSum<T, N:int, M:int>(
    matrix<T, N, M> value,
    vector<uint, 4> mask)
    where T : __BuiltinArithmeticType;

Generic Parameters

T: __BuiltinArithmeticType

N : int

M : int

Parameters

value : T

mask : vector<uint, 4>

value : vector<T, N>

value : matrix<T, N, M>

Availability and Requirements

Defined for the following targets:

hlsl

Available in all stages.

glsl

Available in all stages.

cuda

Available in all stages.

spirv

Available in all stages.

Requires capabilities: spvGroupNonUniformBallot, spvGroupNonUniformArithmetic, spvGroupNonUniformPartitionedNV.