ClickHouse/base/glibc-compatibility/memcpy/memcpy.h

217 lines
11 KiB
C++
Raw Normal View History

2021-10-11 03:58:08 +00:00
#include <stddef.h>
2021-03-08 10:19:25 +00:00
#include <emmintrin.h>
2021-03-08 22:29:32 +00:00
2021-03-13 21:46:47 +00:00
/** Custom memcpy implementation for ClickHouse.
* It has the following benefits over using glibc's implementation:
* 1. Avoiding dependency on specific version of glibc's symbol, like memcpy@@GLIBC_2.14 for portability.
* 2. Avoiding indirect call via PLT due to shared linking, that can be less efficient.
* 3. It's possible to include this header and call inline_memcpy directly for better inlining or interprocedural analysis.
* 4. Better results on our performance tests on current CPUs: up to 25% on some queries and up to 0.7%..1% in average across all queries.
*
* Writing our own memcpy is extremely difficult for the following reasons:
* 1. The optimal variant depends on the specific CPU model.
* 2. The optimal variant depends on the distribution of size arguments.
* 3. It depends on the number of threads copying data concurrently.
* 4. It also depends on how the calling code is using the copied data and how the different memcpy calls are related to each other.
* Due to vast range of scenarios it makes proper testing especially difficult.
* When writing our own memcpy there is a risk to overoptimize it
* on non-representative microbenchmarks while making real-world use cases actually worse.
*
* Most of the benchmarks for memcpy on the internet are wrong.
*
* Let's look at the details:
*
* For small size, the order of branches in code is important.
* There are variants with specific order of branches (like here or in glibc)
* or with jump table (in asm code see example from Cosmopolitan libc:
* https://github.com/jart/cosmopolitan/blob/de09bec215675e9b0beb722df89c6f794da74f3f/libc/nexgen32e/memcpy.S#L61)
* or with Duff device in C (see https://github.com/skywind3000/FastMemcpy/)
*
* It's also important how to copy uneven sizes.
* Almost every implementation, including this, is using two overlapping movs.
*
* It is important to disable -ftree-loop-distribute-patterns when compiling memcpy implementation,
* otherwise the compiler can replace internal loops to a call to memcpy that will lead to infinite recursion.
*
* For larger sizes it's important to choose the instructions used:
* - SSE or AVX or AVX-512;
* - rep movsb;
* Performance will depend on the size threshold, on the CPU model, on the "erms" flag
* ("Enhansed Rep MovS" - it indicates that performance of "rep movsb" is decent for large sizes)
2021-03-13 21:51:13 +00:00
* https://stackoverflow.com/questions/43343231/enhanced-rep-movsb-for-memcpy
2021-03-13 21:46:47 +00:00
*
* Using AVX-512 can be bad due to throttling.
* Using AVX can be bad if most code is using SSE due to switching penalty
* (it also depends on the usage of "vzeroupper" instruction).
* But in some cases AVX gives a win.
*
* It also depends on how many times the loop will be unrolled.
* We are unrolling the loop 8 times (by the number of available registers), but it not always the best.
*
* It also depends on the usage of aligned or unaligned loads/stores.
* We are using unaligned loads and aligned stores.
*
* It also depends on the usage of prefetch instructions. It makes sense on some Intel CPUs but can slow down performance on AMD.
* Setting up correct offset for prefetching is non-obvious.
*
* Non-temporary (cache bypassing) stores can be used for very large sizes (more than a half of L3 cache).
* But the exact threshold is unclear - when doing memcpy from multiple threads the optimal threshold can be lower,
* because L3 cache is shared (and L2 cache is partially shared).
*
* Very large size of memcpy typically indicates suboptimal (not cache friendly) algorithms in code or unrealistic scenarios,
2021-03-13 21:53:00 +00:00
* so we don't pay attention to using non-temporary stores.
2021-03-13 21:46:47 +00:00
*
2022-09-02 08:54:48 +00:00
* On recent Intel CPUs, the presence of "erms" makes "rep movsb" the most beneficial,
2021-03-13 21:46:47 +00:00
* even comparing to non-temporary aligned unrolled stores even with the most wide registers.
*
* memcpy can be written in asm, C or C++. The latter can also use inline asm.
* The asm implementation can be better to make sure that compiler won't make the code worse,
* to ensure the order of branches, the code layout, the usage of all required registers.
* But if it is located in separate translation unit, inlining will not be possible
* (inline asm can be used to overcome this limitation).
* Sometimes C or C++ code can be further optimized by compiler.
* For example, clang is capable replacing SSE intrinsics to AVX code if -mavx is used.
*
* Please note that compiler can replace plain code to memcpy and vice versa.
* - memcpy with compile-time known small size is replaced to simple instructions without a call to memcpy;
* it is controlled by -fbuiltin-memcpy and can be manually ensured by calling __builtin_memcpy.
* This is often used to implement unaligned load/store without undefined behaviour in C++.
* - a loop with copying bytes can be recognized and replaced by a call to memcpy;
* it is controlled by -ftree-loop-distribute-patterns.
* - also note that a loop with copying bytes can be unrolled, peeled and vectorized that will give you
* inline code somewhat similar to a decent implementation of memcpy.
*
* This description is up to date as of Mar 2021.
*
* How to test the memcpy implementation for performance:
* 1. Test on real production workload.
* 2. For synthetic test, see utils/memcpy-bench, but make sure you will do the best to exhaust the wide range of scenarios.
2021-03-13 21:48:08 +00:00
*
* TODO: Add self-tuning memcpy with bayesian bandits algorithm for large sizes.
* See https://habr.com/en/company/yandex/blog/457612/
2021-03-13 21:46:47 +00:00
*/
2021-03-13 07:38:35 +00:00
static inline void * inline_memcpy(void * __restrict dst_, const void * __restrict src_, size_t size)
2021-03-08 10:19:25 +00:00
{
2021-03-13 21:46:47 +00:00
/// We will use pointer arithmetic, so char pointer will be used.
/// Note that __restrict makes sense (otherwise compiler will reload data from memory
/// instead of using the value of registers due to possible aliasing).
2021-03-13 07:38:35 +00:00
char * __restrict dst = reinterpret_cast<char * __restrict>(dst_);
const char * __restrict src = reinterpret_cast<const char * __restrict>(src_);
2021-03-13 21:46:47 +00:00
/// Standard memcpy returns the original value of dst. It is rarely used but we have to do it.
/// If you use memcpy with small but non-constant sizes, you can call inline_memcpy directly
/// for inlining and removing this single instruction.
2021-03-13 07:38:35 +00:00
void * ret = dst;
2021-03-08 10:19:25 +00:00
tail:
2021-03-13 21:46:47 +00:00
/// Small sizes and tails after the loop for large sizes.
/// The order of branches is important but in fact the optimal order depends on the distribution of sizes in your application.
/// This order of branches is from the disassembly of glibc's code.
/// We copy chunks of possibly uneven size with two overlapping movs.
/// Example: to copy 5 bytes [0, 1, 2, 3, 4] we will copy tail [1, 2, 3, 4] first and then head [0, 1, 2, 3].
2021-03-08 10:19:25 +00:00
if (size <= 16)
{
if (size >= 8)
{
2021-03-13 21:46:47 +00:00
/// Chunks of 8..16 bytes.
2021-03-08 10:19:25 +00:00
__builtin_memcpy(dst + size - 8, src + size - 8, 8);
__builtin_memcpy(dst, src, 8);
}
else if (size >= 4)
{
2021-03-13 21:46:47 +00:00
/// Chunks of 4..7 bytes.
2021-03-08 10:19:25 +00:00
__builtin_memcpy(dst + size - 4, src + size - 4, 4);
__builtin_memcpy(dst, src, 4);
}
else if (size >= 2)
{
2021-03-13 21:46:47 +00:00
/// Chunks of 2..3 bytes.
2021-03-08 10:19:25 +00:00
__builtin_memcpy(dst + size - 2, src + size - 2, 2);
__builtin_memcpy(dst, src, 2);
}
else if (size >= 1)
{
2021-03-13 21:46:47 +00:00
/// A single byte.
2021-03-08 10:19:25 +00:00
*dst = *src;
}
2021-03-13 21:48:44 +00:00
/// No bytes remaining.
2021-03-08 10:19:25 +00:00
}
else
{
2021-03-13 21:46:47 +00:00
/// Medium and large sizes.
2021-03-08 10:19:25 +00:00
if (size <= 128)
{
2021-03-13 21:46:47 +00:00
/// Medium size, not enough for full loop unrolling.
/// We will copy the last 16 bytes.
2021-03-08 10:19:25 +00:00
_mm_storeu_si128(reinterpret_cast<__m128i *>(dst + size - 16), _mm_loadu_si128(reinterpret_cast<const __m128i *>(src + size - 16)));
2021-03-13 21:46:47 +00:00
/// Then we will copy every 16 bytes from the beginning in a loop.
/// The last loop iteration will possibly overwrite some part of already copied last 16 bytes.
/// This is Ok, similar to the code for small sizes above.
2021-03-08 10:19:25 +00:00
while (size > 16)
{
_mm_storeu_si128(reinterpret_cast<__m128i *>(dst), _mm_loadu_si128(reinterpret_cast<const __m128i *>(src)));
dst += 16;
src += 16;
size -= 16;
}
}
else
{
2021-03-13 21:46:47 +00:00
/// Large size with fully unrolled loop.
2021-03-08 10:19:25 +00:00
/// Align destination to 16 bytes boundary.
size_t padding = (16 - (reinterpret_cast<size_t>(dst) & 15)) & 15;
2021-03-13 21:46:47 +00:00
/// If not aligned - we will copy first 16 bytes with unaligned stores.
2021-03-08 10:19:25 +00:00
if (padding > 0)
{
__m128i head = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src));
_mm_storeu_si128(reinterpret_cast<__m128i*>(dst), head);
dst += padding;
src += padding;
size -= padding;
}
/// Aligned unrolled copy. We will use half of available SSE registers.
2021-03-13 21:46:47 +00:00
/// It's not possible to have both src and dst aligned.
/// So, we will use aligned stores and unaligned loads.
2021-03-08 10:19:25 +00:00
__m128i c0, c1, c2, c3, c4, c5, c6, c7;
while (size >= 128)
{
c0 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 0);
c1 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 1);
c2 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 2);
c3 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 3);
c4 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 4);
c5 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 5);
c6 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 6);
c7 = _mm_loadu_si128(reinterpret_cast<const __m128i*>(src) + 7);
src += 128;
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 0), c0);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 1), c1);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 2), c2);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 3), c3);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 4), c4);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 5), c5);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 6), c6);
_mm_store_si128((reinterpret_cast<__m128i*>(dst) + 7), c7);
dst += 128;
size -= 128;
}
2021-03-13 21:46:47 +00:00
/// The latest remaining 0..127 bytes will be processed as usual.
2021-03-08 10:19:25 +00:00
goto tail;
}
}
return ret;
}