PHP Benchmarks

Benchmarks by their nature are artificial:

We all know that so there is no need to prattle on about it. The results of these tests indeed need to be taken with a grain or salt, or a shot of Bourbon. The point is however, that some things are simply better practices and knowing those is a way forward and not an excuse to look back and change existing code. Performance follows the classic 80/20 Rule, if test a vs test b does not have > 20% difference, you are very likely optimized.

How it works:

The total number of test loops is set to try run the longest test for ~1 second. This total is then sub divided in to 20 test runs to get an average and margin of error. The test order is random in each group to attempt to minimize any preference or impact of server load on one test group vs another. Completed tests that have a low margin of error are recorded in to the historical statistical record.

Benchmark Group: addslashes

21 hist addslashes1: addslashes() vs. addcslashes()
12 hist addslashes2: addslashes() vs. mysql_escape_string()

Benchmark Group: array

7 hist array1: integer vs. string keys
6 hist array2: arrays vs. variables
13 hist array3: array_unique() vs. !isset() (integers)
6 hist array4: array_push() vs. [] =
4 hist array5: list () vs. [0]
5 hist array6: set duplicates vs. !isset()
3 hist array7: set unique vs. !isset()
11 hist array8: array_diff() vs. ==
8 hist array9: set immediately or later
3 hist array10: in_array() vs. key
12 hist array11: array_keys(array_flip()) vs. array_unique()

Benchmark Group: array_map

9 hist array_map1: array_map() vs. foreach
6 hist array_map2: array_map() vs. array_walk()
9 hist array_map3: array_walk() vs. array_walk_recursive()

Benchmark Group: array_merge

5 hist array_merge1: array_merge() vs. + (string keys)
4 hist array_merge2: array_merge() vs. for (int keys)

Benchmark Group: cast

8 hist cast1: (int) vs. intval()
9 hist cast2: (int) vs. settype()
5 hist cast3: (bool) vs. !empty()
5 hist cast4: (string) vs. interpolate
5 hist cast5: (string) vs. concat
5 hist cast6: concat (string) vs. not
3 hist cast7: (array) vs. array()
3 hist cast8: (array) vs. array() w/is_array()

Benchmark Group: chr

6 hist chr1: chr() v. literal
8 hist chr2: chr() v. pack()
2 hist chr3: ord() v. unpack() (str keys)
4 hist chr4: ord() v. unpack() (int keys)
3 hist chr5: pack() v. sprintf()

Benchmark Group: classes

8 hist classes1: single class vs. associative arrays
8 hist classes2: many classes vs. associative arrays
10 hist classes3: passing classes vs. arrays
6 hist classes4: function vs. class method

Benchmark Group: constants

6 hist constants1: define constants vs. variables
9 hist constants2: using constants vs. variables
5 hist constants3: defined() vs. isset()
6 hist constants4: redefine constant

Benchmark Group: create_function

3 hist create_function1: create_function() vs. eval()
3 hist create_function2: many create_function() vs. eval()
6 hist create_function3: create_function() vs. func callback
5 hist create_function4: create_function() vs. closure inside loop
9 hist create_function5: create_function() vs. closure outside loop

Benchmark Group: date

3 hist date1: date() vs. strftime()
16 hist date2: time() vs. strtotime('now')
6 hist date3: time() vs. microtime(true)
2 hist date4: strftime() vs. localtime()
6 hist date5: strtotime('today') vs. mktime()

Benchmark Group: dir

10 hist dir1: glob() vs. glob no sort files only
58 hist dir2: glob(no_sort) vs scandir() files only
17 hist dir3: glob(no_sort) vs opendir/readdir files only
9 hist dir4: glob(no_sort) vs DirectoryIterator()
2 hist dir5: glob(no_sort) vs DirectoryIterator()

Benchmark Group: echo

7 hist echo1: commas vs. concat on long strings
5 hist echo2: commas vs. concat on short strings
3 hist echo3: commas vs. interpolation on long strings
7 hist echo4: commas vs. interpolation on short strings
4 hist echo5: echo vs. ?>...<?php (static)
7 hist echo6: echo vs. ?>...<?php (dynamic)
5 hist echo7: commas vs. sprintf() on long strings
3 hist echo8: commas vs. sprintf() on short strings

Benchmark Group: equal

4 hist equal1: == vs. === (fail, different types)
6 hist equal2: == vs. === (different types with cast)
3 hist equal3: == vs. === (fail, same types)
8 hist equal4: == vs. === (pass, same types)
1 hist equal5: === vs. strcmp() (fail)
7 hist equal6: === vs. strcmp() (pass)
6 hist equal7: === null vs. is_null()
6 hist equal8: === null vs. == null

Benchmark Group: errors

4 hist errors1: generating notices vs. not
4 hist errors2: error_reporting(0) vs. @
4 hist errors3: Exceptions vs. return code (fail)
7 hist errors4: Exceptions vs. return code (pass)
6 hist errors5: ignoring notices vs. not
4 hist errors6: @ (no error)

Benchmark Group: exec

8 hist exec1: exec() vs. shell_exec()
exec2: system() vs. shell_exec()

Benchmark Group: file

4 hist file1: file() vs. file_get_contents() (str)
7 hist file2: file() vs. file_get_contents() (array)
1 hist file3: file() vs. fgets() (array)

Benchmark Group: foreach

6 hist foreach1: foreach reference reading
2 hist foreach2: foreach reference writing
6 hist foreach3: foreach $k => $dummy vs. array_keys()
4 hist foreach4: foreach vs. while (list)
11 hist foreach5: foreach/range() vs. for
1 hist foreach6: foreach writing ref vs. key
4 hist foreach7: foreach key vs. no key
8 hist foreach8: foreach vs. while (next)
8 hist foreach9: foreach (fcall) vs. foreach ($temp)
2 hist foreach10: foreach vs. for

Benchmark Group: func

3 hist func1: fcall vs. not (non-trivial)
3 hist func2: fcall vs. call_user_func()
3 hist func3: fcall vs. call_user_func_array()
5 hist func4: fcall vs. callback
6 hist func5: callback vs. if
3 hist func6: callback vs. wrapped
5 hist func7: callback vs. call_user_func()
9 hist func8: callback vs. wrapped

Benchmark Group: global

2 hist global1: global vs. $GLOBALS
4 hist global2: global vs. local
3 hist global3: global vs. static

Benchmark Group: hash

14 hist hash1: md5() vs. sha1()
2 hist hash2: sha1() short vs. long

Benchmark Group: if

4 hist if1: else if vs. elseif
4 hist if2: if vs. ternary
if3: nested ifs
5 hist if4: if vs. switch
6 hist if5: if vs. ternary large string

Benchmark Group: ifor

20 hist ifor1: array_keys(array_flip()) vs. array_unique()

Benchmark Group: include

3 hist include1: include vs. require
4 hist include2: include vs. eval()
2 hist include3: include vs. include_once
7 hist include4: file_exists() vs. require
4 hist include5: include_once vs. function_exists()

Benchmark Group: index

index1 (MySQL): performance penalty of one index

Benchmark Group: is

3 hist is1: is_float vs. Float Casting
3 hist is2: is_int vs. Casting
5 hist is3: is_string vs. Casting

Benchmark Group: keycreation

3 hist keycreation: strtr() vs. str_replace() on long strings

Benchmark Group: last

5 hist last1: Last Character

Benchmark Group: math

2 hist math1: float vs. integer addition
2 hist math2: float vs. integer multiplication
1 hist math3: multiply vs. add
3 hist math4: min()/max() vs. if
8 hist math5: abs() vs. ternary

Benchmark Group: ob_start

5 hist ob_start1: ob_start()/ob_get_contents() vs. concat
3 hist ob_start2: ob_start() vs. not (no output)

Benchmark Group: preg

16 hist preg1: preg_replace() vs. str_replace()
8 hist preg2: preg_replace() vs. stri_replace()
2 hist preg3: preg_match() vs. strpos() (simple)
10 hist preg4: preg_match() vs. strpos() (complex)
5 hist preg5: preg_split() vs. explode()/str_replace()
5 hist preg6: preg_replace() vs. preg_replace_callback()

Benchmark Group: rand

2 hist rand1: rand() vs. mt_rand()
1 hist rand2: rand() more vs. less
4 hist rand3: rand() vs. array_rand() (odd keys)
2 hist rand4: rand() vs. array_rand() (seq keys)

Benchmark Group: recurse

5 hist recurse1: recurison vs. not (trivial)
14 hist recurse2: tail recurison

Benchmark Group: socket

36 hist socket1: stream_socket_client() vs. fsockopen()
1 hist socket2: sockets blocking vs. not

Benchmark Group: strpios

10 hist strpios1: stripos() vs. strtolower(substr())
11 hist strpios2: stripos() vs. strtolower(strpos())

Benchmark Group: strpos

6 hist strpos1: strstr() vs. strpos()
10 hist strpos2: strncmp() vs. strpos() (long)
19 hist strpos3: strncmp() vs. strpos() (short)
56 hist strpos4: substr() vs. strpos() (long)
3 hist strpos5: substr() vs. strpos() (short)
2 hist strpos6: strpos()/strtolower() vs. stripos()
3 hist strpos7: substr_compare vs strpos

Benchmark Group: strtr

10 hist strtr1: strtr() vs. str_replace() on long strings
5 hist strtr2: strtr() vs. str_replace() on short strings
5 hist strtr3: str_ireplace vs. strtolower + str_replace
2 hist strtr4: strtr() vs. str_replace()/array_keys() on long strings

Benchmark Group: substr

3 hist substr1: substr($var, 5, 1) vs. {5}
6 hist substr2: substr() negative vs. strlen()
10 hist substr3: substr() on short vs. long strings
9 hist substr4: substr($var, 5, 1) vs. isset && {5}
7 hist substr5: substr($var, 25, 1) vs. isset && {25}

Benchmark Group: urlencode

3 hist urlencode1: urlencode() vs. rawurlencode()
7 hist urlencode2: urlencode() vs. base64_encode()
3 hist urlencode3: urlencode() vs. urldecode()