github.com/apache/arrow/go/v10@v10.0.1/internal/utils/min_max_amd64.go (about)

     1  // Licensed to the Apache Software Foundation (ASF) under one
     2  // or more contributor license agreements.  See the NOTICE file
     3  // distributed with this work for additional information
     4  // regarding copyright ownership.  The ASF licenses this file
     5  // to you under the Apache License, Version 2.0 (the
     6  // "License"); you may not use this file except in compliance
     7  // with the License.  You may obtain a copy of the License at
     8  //
     9  // http://www.apache.org/licenses/LICENSE-2.0
    10  //
    11  // Unless required by applicable law or agreed to in writing, software
    12  // distributed under the License is distributed on an "AS IS" BASIS,
    13  // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    14  // See the License for the specific language governing permissions and
    15  // limitations under the License.
    16  
    17  //go:build !noasm
    18  // +build !noasm
    19  
    20  package utils
    21  
    22  import "golang.org/x/sys/cpu"
    23  
    24  func init() {
    25  	// if the CPU supports AVX2 or SSE4 then let's use those to benefit from SIMD
    26  	// to accelerate the performance for finding the min and max for an integral slice.
    27  	// otherwise fallback to a pure go implementation if the cpu doesn't have these features.
    28  	if cpu.X86.HasAVX2 {
    29  		minmaxFuncs.i8 = int8MaxMinAVX2
    30  		minmaxFuncs.ui8 = uint8MaxMinAVX2
    31  		minmaxFuncs.i16 = int16MaxMinAVX2
    32  		minmaxFuncs.ui16 = uint16MaxMinAVX2
    33  		minmaxFuncs.i32 = int32MaxMinAVX2
    34  		minmaxFuncs.ui32 = uint32MaxMinAVX2
    35  		minmaxFuncs.i64 = int64MaxMinAVX2
    36  		minmaxFuncs.ui64 = uint64MaxMinAVX2
    37  	} else if cpu.X86.HasSSE42 {
    38  		minmaxFuncs.i8 = int8MaxMinSSE4
    39  		minmaxFuncs.ui8 = uint8MaxMinSSE4
    40  		minmaxFuncs.i16 = int16MaxMinSSE4
    41  		minmaxFuncs.ui16 = uint16MaxMinSSE4
    42  		minmaxFuncs.i32 = int32MaxMinSSE4
    43  		minmaxFuncs.ui32 = uint32MaxMinSSE4
    44  		minmaxFuncs.i64 = int64MaxMinSSE4
    45  		minmaxFuncs.ui64 = uint64MaxMinSSE4
    46  	} else {
    47  		minmaxFuncs.i8 = int8MinMax
    48  		minmaxFuncs.ui8 = uint8MinMax
    49  		minmaxFuncs.i16 = int16MinMax
    50  		minmaxFuncs.ui16 = uint16MinMax
    51  		minmaxFuncs.i32 = int32MinMax
    52  		minmaxFuncs.ui32 = uint32MinMax
    53  		minmaxFuncs.i64 = int64MinMax
    54  		minmaxFuncs.ui64 = uint64MinMax
    55  	}
    56  }