swift - Swift Beta性能：排序数组(Swift Beta performance: sorting arrays)

Question

I was implementing an algorithm in Swift Beta and noticed that the performance was very poor.

(我在Swift Beta中实现了一个算法，并注意到性能非常差。)

After digging deeper I realized that one of the bottlenecks was something as simple as sorting arrays.

(在深入挖掘之后，我意识到其中一个瓶颈就像排序数组一样简单。)

The relevant part is here:

(相关部分在这里：)

let n = 1000000
var x =  [Int](repeating: 0, count: n)
for i in 0..<n {
    x[i] = random()
}
// start clock here
let y = sort(x)
// stop clock here

In C++, a similar operation takes 0.06s on my computer.

(在C ++中，类似的操作在我的计算机上需要0.06秒 。)

In Python, it takes 0.6s (no tricks, just y = sorted(x) for a list of integers).

(在Python中，它需要0.6秒 （没有技巧，只有y =排序（x）表示整数列表）。)

In Swift it takes 6s if I compile it with the following command:

(在Swift中，如果我使用以下命令编译它需要6秒 ：)

xcrun swift -O3 -sdk `xcrun --show-sdk-path --sdk macosx`

And it takes as much as 88s if I compile it with the following command:

(如果我使用以下命令编译它需要多达88秒 ：)

xcrun swift -O0 -sdk `xcrun --show-sdk-path --sdk macosx`

Timings in Xcode with "Release" vs. "Debug" builds are similar.

(Xcode中的“释放”与“调试”版本的计时相似。)

What is wrong here?

(这有什么不对？)

I could understand some performance loss in comparison with C++, but not a 10-fold slowdown in comparison with pure Python.

(与C ++相比，我可以理解一些性能损失，但与纯Python相比，速度没有降低10倍。)

---

Edit: weather noticed that changing -O3 to -Ofast makes this code run almost as fast as the C++ version!

(编辑：天气注意到将-O3更改为-Ofast使得此代码的运行速度几乎与C ++版本一样快！)

However, -Ofast changes the semantics of the language a lot — in my testing, it disabled the checks for integer overflows and array indexing overflows .

(但是， -Ofast改变了语言的语义 - 在我的测试中，它禁止检查整数溢出和数组索引溢出 。)

For example, with -Ofast the following Swift code runs silently without crashing (and prints out some garbage):

(例如，使用-Ofast ，以下Swift代码以静默方式运行而不会崩溃（并打印出一些垃圾）：)

let n = 10000000
print(n*n*n*n*n)
let x =  [Int](repeating: 10, count: n)
print(x[n])

So -Ofast is not what we want;

(所以-Ofast不是我们想要的;)

the whole point of Swift is that we have the safety nets in place.

(斯威夫特的全部意义在于我们有安全网。)

Of course, the safety nets have some impact on the performance, but they should not make the programs 100 times slower.

(当然，安全网对性能有一些影响，但它们不应该使程序慢100倍。)

Remember that Java already checks for array bounds, and in typical cases, the slowdown is by a factor much less than 2. And in Clang and GCC we have got -ftrapv for checking (signed) integer overflows, and it is not that slow, either.

(请记住，Java已经检查了数组边界，并且在典型情况下，减速是一个远小于2的因素。在Clang和GCC中，我们有-ftrapv用于检查（签名）整数溢出，并且它不是那么慢，无论是。)

Hence the question: how can we get reasonable performance in Swift without losing the safety nets?

(因此，问题是：如何在不丢失安全网的情况下在Swift中获得合理的性能？)

---

Edit 2: I did some more benchmarking, with very simple loops along the lines of

(编辑2：我做了一些基准测试，非常简单的循环)

for i in 0..<n {
    x[i] = x[i] ^ 12345678
}

(Here the xor operation is there just so that I can more easily find the relevant loop in the assembly code. I tried to pick an operation that is easy to spot but also "harmless" in the sense that it should not require any checks related to integer overflows.)

(（这里的xor操作只是为了让我可以更容易地在汇编代码中找到相关的循环。我试图选择一个易于发现但也“无害”的操作，因为它不需要任何相关的检查到整数溢出。）)

Again, there was a huge difference in the performance between -O3 and -Ofast .

(同样， -O3和-Ofast之间的性能差异-Ofast 。)

So I had a look at the assembly code:

(所以我看了一下汇编代码：)

• With -Ofast I get pretty much what I would expect.

  (随着-Ofast我得到了我所期望的。)

  The relevant part is a loop with 5 machine language instructions.

  (相关部分是一个包含5个机器语言指令的循环。)

• With -O3 I get something that was beyond my wildest imagination.

  (有了-O3我得到的东西超出了我的想象力。)

  The inner loop spans 88 lines of assembly code.

  (内环跨越88行汇编代码。)

  I did not try to understand all of it, but the most suspicious parts are 13 invocations of "callq _swift_retain" and another 13 invocations of "callq _swift_release".

  (我没有尝试理解所有这些，但最可疑的部分是13次调用“callq _swift_retain”和另外13次调用“callq _swift_release”。)

  That is, 26 subroutine calls in the inner loop !

  (也就是说， 内循环中有26个子程序调用 ！)

---

Edit 3: In comments, Ferruccio asked for benchmarks that are fair in the sense that they do not rely on built-in functions (eg sort).

(编辑3：在评论中，Ferruccio要求提供公平的基准，因为他们不依赖于内置函数（例如排序）。)

I think the following program is a fairly good example:

(我认为以下程序是一个相当好的例子：)

let n = 10000
var x = [Int](repeating: 1, count: n)
for i in 0..<n {
    for j in 0..<n {
        x[i] = x[j]
    }
}

There is no arithmetic, so we do not need to worry about integer overflows.

(没有算术，所以我们不需要担心整数溢出。)

The only thing that we do is just lots of array references.

(我们唯一做的就是大量的数组引用。)

And the results are here—Swift -O3 loses by a factor almost 500 in comparison with -Ofast:

(结果在这里 - 与-Ofast相比，Swift -O3损失了近500倍：)

• C++ -O3: 0.05 s

  (C ++ -O3： 0.05秒)

• C++ -O0: 0.4 s

  (C ++ -O0：0.4秒)

• Java: 0.2 s

  (Java： 0.2秒)

• Python with PyPy: 0.5 s

  (使用PyPy的Python：0.5秒)

• Python: 12 s

  (Python： 12秒)

• Swift -Ofast: 0.05 s

  (Swift -Ofast：0.05秒)

• Swift -O3: 23 s

  (Swift -O3： 23秒)

• Swift -O0: 443 s

  (Swift -O0：443秒)

(If you are concerned that the compiler might optimize out the pointless loops entirely, you can change it to eg x[i] ^= x[j] , and add a print statement that outputs x[0] . This does not change anything; the timings will be very similar.)

(（如果您担心编译器可能会完全优化无意义循环，您可以将其更改为例如x[i] ^= x[j] ，并添加一个输出x[0]的print语句。这不会改变任何内容;时间将非常相似。）)

And yes, here the Python implementation was a stupid pure Python implementation with a list of ints and nested for loops.

(是的，这里的Python实现是一个愚蠢的纯Python实现，带有一个int列表和嵌套for循环。)

It should be much slower than unoptimized Swift.

(它应该比未优化雨燕慢得多 。)

Something seems to be seriously broken with Swift and array indexing.

(使用Swift和数组索引似乎严重破坏了某些东西。)

---

Edit 4: These issues (as well as some other performance issues) seems to have been fixed in Xcode 6 beta 5.

(编辑4：这些问题（以及一些其他性能问题）似乎已在Xcode 6 beta 5中得到修复。)

For sorting, I now have the following timings:

(为了排序，我现在有以下时间：)

• clang++ -O3: 0.06 s

  (clang ++ -O3：0.06 s)

• swiftc -Ofast: 0.1 s

  (swiftc -Ofast：0.1秒)

• swiftc -O: 0.1 s

  (swiftc -O：0.1秒)

• swiftc: 4 s

  (swiftc：4秒)

For nested loops:

(对于嵌套循环：)

• clang++ -O3: 0.06 s

  (clang ++ -O3：0.06 s)

• swiftc -Ofast: 0.3 s

  (swiftc -Ofast：0.3秒)

• swiftc -O: 0.4 s

  (swiftc -O：0.4 s)

• swiftc: 540 s

  (swiftc：540秒)

It seems that there is no reason anymore to use the unsafe -Ofast (aka -Ounchecked );

(似乎没有理由再使用unsafe -Ofast （aka -Ounchecked ）;)

plain -O produces equally good code.

(plain -O产生同样好的代码。)

  ask by Jukka Suomela translate from so

Answer 1

tl;dr Swift 1.0 is now as fast as C by this benchmark using the default release optimisation level [-O].

(tl; Dr Swift 1.0现在使用默认版本优化级别[-O]通过此基准测试与C一样快。)

---

Here is an in-place quicksort in Swift Beta:

(这是Swift Beta中的就地快速排序：)

func quicksort_swift(inout a:CInt[], start:Int, end:Int) {
    if (end - start < 2){
        return
    }
    var p = a[start + (end - start)/2]
    var l = start
    var r = end - 1
    while (l <= r){
        if (a[l] < p){
            l += 1
            continue
        }
        if (a[r] > p){
            r -= 1
            continue
        }
        var t = a[l]
        a[l] = a[r]
        a[r] = t
        l += 1
        r -= 1
    }
    quicksort_swift(&a, start, r + 1)
    quicksort_swift(&a, r + 1, end)
}

And the same in C:

(在C中也一样：)

void quicksort_c(int *a, int n) {
    if (n < 2)
        return;
    int p = a[n / 2];
    int *l = a;
    int *r = a + n - 1;
    while (l <= r) {
        if (*l < p) {
            l++;
            continue;
        }
        if (*r > p) {
            r--;
            continue;
        }
        int t = *l;
        *l++ = *r;
        *r-- = t;
    }
    quicksort_c(a, r - a + 1);
    quicksort_c(l, a + n - l);
}

Both work:

(两者都有效：)

var a_swift:CInt[] = [0,5,2,8,1234,-1,2]
var a_c:CInt[] = [0,5,2,8,1234,-1,2]

quicksort_swift(&a_swift, 0, a_swift.count)
quicksort_c(&a_c, CInt(a_c.count))

// [-1, 0, 2, 2, 5, 8, 1234]
// [-1, 0, 2, 2, 5, 8, 1234]

Both are called in the same program as written.

(两者都在与编写的程序中调用。)

var x_swift = CInt[](count: n, repeatedValue: 0)
var x_c = CInt[](count: n, repeatedValue: 0)
for var i = 0; i < n; ++i {
    x_swift[i] = CInt(random())
    x_c[i] = CInt(random())
}

let swift_start:UInt64 = mach_absolute_time();
quicksort_swift(&x_swift, 0, x_swift.count)
let swift_stop:UInt64 = mach_absolute_time();

let c_start:UInt64 = mach_absolute_time();
quicksort_c(&x_c, CInt(x_c.count))
let c_stop:UInt64 = mach_absolute_time();

This converts the absolute times to seconds:

(这会将绝对时间转换为秒：)

static const uint64_t NANOS_PER_USEC = 1000ULL;
static const uint64_t NANOS_PER_MSEC = 1000ULL * NANOS_PER_USEC;
static const uint64_t NANOS_PER_SEC = 1000ULL * NANOS_PER_MSEC;

mach_timebase_info_data_t timebase_info;

uint64_t abs_to_nanos(uint64_t abs) {
    if ( timebase_info.denom == 0 ) {
        (void)mach_timebase_info(&timebase_info);
    }
    return abs * timebase_info.numer  / timebase_info.denom;
}

double abs_to_seconds(uint64_t abs) {
    return abs_to_nanos(abs) / (double)NANOS_PER_SEC;
}

Here is a summary of the compiler's optimazation levels:

(以下是编译器优化级别的摘要：)

[-Onone] no optimizations, the default for debug.
[-O]     perform optimizations, the default for release.
[-Ofast] perform optimizations and disable runtime overflow checks and runtime type checks.

Time in seconds with [-Onone] for n=10_000 :

(对于n = 10_000 ， [-Onone]的时间以秒为单位 ：)

Swift:            0.895296452
C:                0.001223848

Here is Swift's builtin sort() for n=10_000 :

(这是Swift的内置排序（），用于n = 10_000 ：)

Swift_builtin:    0.77865783

Here is [-O] for n=10_000 :

(对于n = 10_000，这是[-O] ：)

Swift:            0.045478346
C:                0.000784666
Swift_builtin:    0.032513488

As you can see, Swift's performance improved by a factor of 20.

(如您所见，Swift的性能提高了20倍。)

As per mweathers' answer , setting [-Ofast] makes the real difference, resulting in these times for n=10_000 :

(根据mweathers的回答 ，设置[-Ofast]会产生真正的差异，导致n = 10_000的这些时间：)

Swift:            0.000706745
C:                0.000742374
Swift_builtin:    0.000603576

And for n=1_000_000 :

(对于n = 1_000_000 ：)

Swift:            0.107111846
C:                0.114957179
Swift_sort:       0.092688548

For comparison, this is with [-Onone] for n=1_000_000 :

(为了比较，对于n = 1_000_000 ，这是[-Onone] ：)

Swift:            142.659763258
C:                0.162065333
Swift_sort:       114.095478272

So Swift with no optimizations was almost 1000x slower than C in this benchmark, at this stage in its development.

(因此，在这个基准测试中，没有优化的Swift在开发的这个阶段比C慢了近1000倍。)

On the other hand with both compilers set to [-Ofast] Swift actually performed at least as well if not slightly better than C.

(另一方面，两个编译器都设置为[-Ofast] Swift实际上至少表现得好，如果不是比C略好一点。)

It has been pointed out that [-Ofast] changes the semantics of the language, making it potentially unsafe.

(已经指出[-Ofast]改变了语言的语义，使其可能不安全。)

This is what Apple states in the Xcode 5.0 release notes:

(这就是Apple在Xcode 5.0发行说明中所说的：)

A new optimization level -Ofast, available in LLVM, enables aggressive optimizations.

(LLVM中提供的新优化级别-Ofast可实现积极的优化。)

-Ofast relaxes some conservative restrictions, mostly for floating-point operations, that are safe for most code.

(-Ofast放松了一些保守的限制，主要用于浮点运算，对大多数代码都是安全的。)

It can yield significant high-performance wins from the compiler.

(它可以从编译器中获得显着的高性能胜利。)

They all but advocate it.

(他们都提倡它。)

Whether that's wise or not I couldn't say, but from what I can tell it seems reasonable enough to use [-Ofast] in a release if you're not doing high-precision floating point arithmetic and you're confident no integer or array overflows are possible in your program.

(这是否明智我不能说，但从我可以说的是，如果你没有进行高精度浮点运算并且你确信没有整数或者一个版本，那么在一个版本中使用[-Ofast]似乎是合理的。您的程序中可能存在数组溢出。)

If you do need high performance and overflow checks / precise arithmetic then choose another language for now.

(如果您确实需要高性能和溢出检查/精确算术，那么现在就选择另一种语言。)

BETA 3 UPDATE:

(BETA 3更新：)

n=10_000 with [-O] :

(n = 10_000，带[ - O] ：)

Swift:            0.019697268
C:                0.000718064
Swift_sort:       0.002094721

Swift in general is a bit faster and it looks like Swift's built-in sort has changed quite significantly.

(一般来说Swift有点快，看起来Swift的内置排序已经发生了很大变化。)

FINAL UPDATE:

(最终更新：)

[-Onone] :

([-Onone] ：)

Swift:   0.678056695
C:       0.000973914

[-O] :

([-O] ：)

Swift:   0.001158492
C:       0.001192406

[-Ounchecked] :

([-Ounchecked] ：)

Swift:   0.000827764
C:       0.001078914