Optimize PFCOUNT, PFMERGE command by SIMD acceleration

[Nugine]

WIP: unit tests

---

This PR optimizes the performance of HyperLogLog commands (PFCOUNT, PFMERGE) by adding AVX2 fast paths.

Two AVX2 functions are added for conversion between raw representation and dense representation. They are 15 ~ 30 times faster than scalar implementaion. Note that sparse representation is not accelerated.

AVX2 fast paths are enabled when the CPU supports AVX2 (checked at runtime) and the hyperloglog configuration is default (HLL_REGISTERS == 16384 && HLL_BITS == 6).

When merging 3 dense hll structures, the benchmark shows a 12x speedup compared to the scalar version.

pfcount key1 key2 key3
pfmerge keyall key1 key2 key3

======================================================================================================
Type             Ops/sec    Avg. Latency     p50 Latency     p99 Latency   p99.9 Latency       KB/sec 
------------------------------------------------------------------------------------------------------
PFCOUNT-scalar    5665.56        35.29839        32.25500        63.99900        67.58300       608.60
PFCOUNT-avx2     72377.83         2.75834         2.67100         5.34300         6.81500      7774.96
------------------------------------------------------------------------------------------------------
PFMERGE-scalar    9851.29        20.28806        20.09500        36.86300        39.16700       615.71
PFMERGE-avx2    125621.89         1.59126         1.55100         3.11900         4.70300     15702.74
------------------------------------------------------------------------------------------------------

scalar: valkey:unstable  2df56d87c0ebe802f38e8922bb2ea1e4ca9cfa76
avx2:   Nugine:hll-simd  8f9adc34021080d96e60bd0abe06b043f3ed0275

CPU:    13th Gen Intel® Core™ i9-13900H × 20
Memory: 32.0 GiB
OS:     Ubuntu 22.04.5 LTS

Experiment repo: https://github.com/Nugine/redis-hyperloglog
Benchmark script: https://github.com/Nugine/redis-hyperloglog/blob/main/scripts/memtier.sh
Algorithm: https://github.com/Nugine/redis-hyperloglog/blob/main/cpp/bench.cpp

[codecov]

Codecov Report

Attention: Patch coverage is 52.80899% with 42 lines in your changes missing coverage. Please review.

Project coverage is 70.70%. Comparing base (2df56d8) to head (a6136ca).
Report is 10 commits behind head on unstable.

Files with missing lines	Patch %	Lines
src/hyperloglog.c	52.80%	42 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##           unstable    #1293      +/-   ##
============================================
+ Coverage     70.69%   70.70%   +0.01%     
============================================
  Files           114      115       +1     
  Lines         63161    63233      +72     
============================================
+ Hits          44650    44710      +60     
- Misses        18511    18523      +12     

Files with missing lines	Coverage Δ
src/hyperloglog.c	86.35% <52.80%> (-4.89%)	⬇️

... and 22 files with indirect coverage changes

[xbasel]

How was this change tested, particularly to confirm that the non-AVX2 and AVX2 implementations produce the same results?

[Nugine]

How was this change tested, particularly to confirm that the non-AVX2 and AVX2 implementations produce the same results?

The algorithms are verified by comparing the results between scalar code and simd code with random input.
https://github.com/Nugine/redis-hyperloglog/blob/main/cpp/bench.cpp

[zuiderkwast]

This is so cool! 😎

How was this change tested, particularly to confirm that the non-AVX2 and AVX2 implementations produce the same results?

The algorithms are verified by comparing the results between scalar code and simd code with random input. https://github.com/Nugine/redis-hyperloglog/blob/main/cpp/bench.cpp

I think we need to test this in our repo in some way. The binary representation can't change, because things like replicas will not understand it, so we should verify the binary representation.

A hyperloglog key is actually a string so we can use the GET command to get the binary representation. In a TCL test case, we can use GET and compare the reply to the binary data we have stored earlier. Alternatively we can use DUMP. Can you add it?

[zuiderkwast]

@lipzhu You're the performance expert. Do you want to review this?

[lipzhu]

@Nugine Great job, the performance is impressive.

Echo @zuiderkwast @xbasel .
Maybe a unit test needed to verify the results is totally same W/ AVX2 instructions and make sure binary file will not changed when RDB presentence is enabled?

[lipzhu]

Nit: Missing indentation.

[Nugine]

clang-format-18 removes the indentation 😂

[zuiderkwast]

Yes, case and labels for goto are indented one step less than other code. This is correct indentation, a common style in C.

[lipzhu]

Will the last 4 bytes padding bytes impact the binary dump files which caused the inconsistent with origin one?

[Nugine]

The 4 padding bytes produced by AVX2 STORE are overwritten by the last scalar computation (32 registers).

The two AVX2 functions keep the memory format unchanged. They just accelerate the computation.

[Nugine]

WIP: unit tests for verifying the results produced by non-AVX2 and AVX2 implementations.

I have other things to do these days. So it may take a while.