15 #ifndef __IPSDKUTIL_INSTRUCTIONSET_ARITHMETIC_DETAIL_AVX2_SCANPACK_H__ 16 #define __IPSDKUTIL_INSTRUCTIONSET_ARITHMETIC_DETAIL_AVX2_SCANPACK_H__ 38 static IPSDK_FORCEINLINE
45 static IPSDK_FORCEINLINE
57 static IPSDK_FORCEINLINE
67 static IPSDK_FORCEINLINE
70 __m256d regOffset = _mm256_setzero_pd();
78 for (; i < nbAlignedElts; i += simd::NbEltsPerReg<ePackType::ePT_Avx, ipReal64>::Value) {
83 regOut = _mm256_add_pd(regOut, regOffset);
85 regOffset = _mm256_permute4x64_pd(regOut, 0xff);
87 for (; i < nbElts; ++i) {
90 out[i] = out[i - 1] + in[i];
102 #endif // __IPSDKUTIL_INSTRUCTIONSET_ARITHMETIC_DETAIL_AVX2_SCANPACK_H__
Defines the IPSDK_FORCEINLINE.
Main namespace for IPSDK library.
Definition: AlgorithmFunctionEfficiency.h:22
template structure which is specialized to implement the cumulative sum on 2 registers, depending on the used instructionSet and on the types of the buffers loaded in the registers
Definition: ScanReg.h:36
uint64_t ipUInt64
Base types definition.
Definition: BaseTypes.h:55
double ipReal64
Base types definition.
Definition: BaseTypes.h:57
Definition: NbEltsPerReg.h:28
eInstructionSet
Enumerate for processor instruction set description.
Definition: InstructionSetTypes.h:31
Advanced Vector Extensions 2.
Definition: InstructionSetTypes.h:48
Definition: ScanPack.h:35
Definition of import/export macro for library.
structure containing intrinsic registers used to store vectorized data
Definition: BaseReg.h:29
Definition: UnloadReg.h:30
Definition: LoadRegDecl.h:30
Definition: ScanPack.h:30
float ipReal32
Base types definition.
Definition: BaseTypes.h:56
structure containing intrinsic registers used to store vectorized data
Definition: BasePackDecl.h:29