我正在處理一些代碼挑戰，發現自定義排序（排序接口的實現）比僅針對切片的原始結構工作得更快。這是為什么？切片轉換為類型是否有一些魔力（比如轉換為結構指針的切片）？我做了一些代碼來測試我的 hipotesispackage sortingexampleimport (    "sort"    "testing")// Example of struct we going to sort.type Point struct {    X, Y int}// --- Struct / Raw Datavar TestCases = []Point{    {10, 3},    {10, 4},    {10, 35},    {10, 5},    {10, 51},    {10, 25},    {10, 59},    {10, 15},    {10, 22},    {10, 91},}// Example One - Sorting Slice Directly// somehow - slowest way to sort it.func SortSlice(points []Point) {    sort.Slice(points, func(i, j int) bool {        return points[i].Y < points[j].Y    })}func BenchmarkSlice(b *testing.B) {    tmp := make([]Point, len(TestCases))    for i := 0; i < b.N; i++ {        copy(tmp, TestCases)        SortSlice(tmp)    }}// Example Two - Sorting Slice Directly// much faster performancetype Points []Point// Sort interface implementationfunc (p Points) Less(i, j int) bool { return p[i].Y < p[j].Y }func (p Points) Len() int           { return len(p) }func (p Points) Swap(i, j int)      { p[i], p[j] = p[j], p[i] }func SortStruct(points []Point) {    sort.Sort(Points(points))}func BenchmarkStruct(b *testing.B) {    tmp := make([]Point, len(TestCases))    for i := 0; i < b.N; i++ {        copy(tmp, TestCases)        SortStruct(tmp)    }}// --- Pointersvar TestCasesPoints = []*Point{    &Point{10, 3},    &Point{10, 4},    &Point{10, 35},    &Point{10, 5},    &Point{10, 51},    &Point{10, 25},    &Point{10, 59},    &Point{10, 15},    &Point{10, 22},    &Point{10, 91},}// Example Three - Sorting Slice of Pointersfunc SortSlicePointers(points []*Point) {    sort.Slice(points, func(i, j int) bool {        return points[i].Y < points[j].Y    })}func BenchmarkSlicePointers(b *testing.B) {    tmp := make([]*Point, len(TestCasesPoints))    for i := 0; i < b.N; i++ {        copy(tmp, TestCasesPoints)        SortSlicePointers(tmp)    }}很明顯，對指針切片進行排序會更快，但是為什么自定義排序實現會更快呢？有什么我可以閱讀的資源嗎？