ext/bcmath: Avoid unnecessary memset

[SakiTakamachi]

0 padding on the right is not required for calc. This is because when converting back to string, zend_string is padded with 0.

php-src/ext/bcmath/libbcmath/src/num2str.c

Line 68 in 0642855

*sptr++ = BCD_CHAR(0);

(I will follow up for add later.)

benchmark: #14132

before

// 1
Time (mean ± σ):     551.0 ms ±   8.4 ms    [User: 546.6 ms, System: 3.7 ms]
Range (min … max):   536.9 ms … 563.4 ms    10 runs

// 2
Time (mean ± σ):     610.8 ms ±  16.4 ms    [User: 607.4 ms, System: 2.6 ms]
Range (min … max):   599.4 ms … 653.7 ms    10 runs

// 3
Time (mean ± σ):     707.8 ms ±  21.3 ms    [User: 704.4 ms, System: 2.6 ms]
Range (min … max):   681.0 ms … 754.1 ms    10 runs

after

// 1
Time (mean ± σ):     523.3 ms ±   3.9 ms    [User: 519.3 ms, System: 3.2 ms]
Range (min … max):   518.5 ms … 529.4 ms    10 runs

// 2
Time (mean ± σ):     591.5 ms ±  13.4 ms    [User: 587.8 ms, System: 2.9 ms]
Range (min … max):   575.3 ms … 620.9 ms    10 runs

// 3
Time (mean ± σ):     685.9 ms ±  10.2 ms    [User: 681.8 ms, System: 3.3 ms]
Range (min … max):   668.6 ms … 696.7 ms    10 runs

[ndossche]

Intuitively, this speed gain feels strange because memset only takes up 0.35% of runtime on my machine for 1.php for example. Furthermore, I actually see a slowdown when applying this patch on current master:

Benchmark 1: ./sapi/cli/php 1.php
  Time (mean ± σ):     436.2 ms ±   6.2 ms    [User: 432.7 ms, System: 2.8 ms]
  Range (min … max):   430.3 ms … 450.7 ms    10 runs
 
Benchmark 2: ./sapi/cli/php_old 1.php
  Time (mean ± σ):     418.8 ms ±   8.3 ms    [User: 415.5 ms, System: 2.5 ms]
  Range (min … max):   412.3 ms … 440.9 ms    10 runs
 
Summary
  ./sapi/cli/php_old 1.php ran
    1.04 ± 0.03 times faster than ./sapi/cli/php 1.php


Benchmark 1: ./sapi/cli/php 2.php
  Time (mean ± σ):     499.6 ms ±   3.1 ms    [User: 496.0 ms, System: 2.7 ms]
  Range (min … max):   494.7 ms … 503.3 ms    10 runs
 
Benchmark 2: ./sapi/cli/php_old 2.php
  Time (mean ± σ):     475.8 ms ±   3.4 ms    [User: 473.5 ms, System: 1.4 ms]
  Range (min … max):   472.0 ms … 480.1 ms    10 runs
 
Summary
  ./sapi/cli/php_old 2.php ran
    1.05 ± 0.01 times faster than ./sapi/cli/php 2.php

I see a similar slowdown for your old benchmark script with microtime.
It doesn't make sense that the execution becomes slower, I think something deeper is going on especially as you see the opposite effect...

[SakiTakamachi]

Thanks. It seems the measurements on my machine are unreliable....

I will try to prepare a stable measurement environment such as EC2.

[SakiTakamachi]

@nielsdos
I merged the latest master and then compared master and this branch on EC2.

master:

hyperfine "php 1.php" --warmup 10
Time (mean ± σ):     654.7 ms ±   3.0 ms    [User: 650.5 ms, System: 2.6 ms]
Range (min … max):   650.1 ms … 659.2 ms    10 runs

hyperfine "php 2.php" --warmup 10
Time (mean ± σ):     769.4 ms ±   5.4 ms    [User: 765.8 ms, System: 2.0 ms]
Range (min … max):   762.6 ms … 781.8 ms    10 runs

hyperfine "php 3.php" --warmup 10
Time (mean ± σ):     910.3 ms ±  13.3 ms    [User: 905.6 ms, System: 2.8 ms]
Range (min … max):   896.8 ms … 934.6 ms    10 runs

php old.php // my old bench
1.6298861503601
1.9048039913177
2.2188358306885

this branch:

hyperfine "php 1.php" --warmup 10
Time (mean ± σ):     605.0 ms ±   2.5 ms    [User: 601.5 ms, System: 2.0 ms]
Range (min … max):   601.3 ms … 609.0 ms    10 runs

hyperfine "php 2.php" --warmup 10
Time (mean ± σ):     717.5 ms ±   5.0 ms    [User: 713.1 ms, System: 2.7 ms]
Range (min … max):   712.5 ms … 730.7 ms    10 runs

hyperfine "php 3.php" --warmup 10
Time (mean ± σ):     863.4 ms ±   9.3 ms    [User: 858.7 ms, System: 2.6 ms]
Range (min … max):   851.4 ms … 878.6 ms    10 runs

php old.php // my old bench
1.5056660175323
1.7816500663757
2.116672039032

[ndossche]

Okay, fine by me then.
Something influences my results and I'm not sure what. Might be caches or micro-architectural stuff.

[SakiTakamachi]

Is the compiler gcc?

I am compiling both with gcc on ubuntu. It might be worth checking if a different assembly is being generated.

[ndossche]

I'm using gcc 13

[SakiTakamachi]

My machine is 9, but the EC2 is 13.

Could you please send me the assembler code later when you have time? I'd like to compare it with mine.

[ndossche]

Could you please send me the assembler code later when you have time? I'd like to compare it with mine.

This is the assembly for this PR's doaddsub.s: https://gist.github.com/nielsdos/551b59b07a25df46b8416edc31f89a73
This is the assembly for master's doaddsub.s: https://gist.github.com/nielsdos/5a6200fdb5173c391c23a03c0f7a7dbc

However, I'm not so sure the problem is the assembly, I think it's some CPU magic (e.g. microarchitecture or cache state or something like that) that is coincidentally influenced with this change.