PHP web framework comparison: any two, head to head
workerman is quicker than swoole on 9 of the 16 profiles they both run, and behind on 7. On the HTTP/1.1 composite they score 707 and 576.
Pick any two of the 13 PHP entries. Requests per second is the best of three runs and the delta is the first column against the second; every row is the same profile at the same connection count, on the same machine. What each profile measures.
| Profile | Conns | swoole req/sec | workerman req/sec | Delta | swoole p99 | workerman p99 |
|---|---|---|---|---|---|---|
| Baseline | 512 | 2,035,856 | 2,844,479 | -28% | 3.31ms | 488us |
| Baseline | 4,096 | 2,034,436 | 2,848,413 | -29% | 8.16ms | 3.05ms |
| Pipelined | 512 | 2,833,444 | 3,637,410 | -22% | 8.99ms | 4.04ms |
| Pipelined | 4,096 | 2,928,585 | 3,936,670 | -26% | 45.20ms | 21.20ms |
| Short-lived | 512 | 1,511,351 | 1,967,016 | -23% | 3.59ms | 1.97ms |
| Short-lived | 4,096 | 1,796,637 | 2,215,363 | -19% | 9.69ms | 21.40ms |
| JSON | 4,096 | 946,517 | 1,039,295 | -9% | 16.40ms | 70.40ms |
| JSON TLS | 4,096 | 776,894 | 734,683 | +6% | - | 5.54ms |
| Upload | 32 | 279 | 596 | -53% | 848.30ms | 306.60ms |
| Upload | 256 | 343 | 470 | -27% | 3.23s | 2.13s |
| Static | 1,024 | 1,124,608 | 292,011 | +285% | - | 3.52ms |
| Static | 4,096 | 1,111,770 | 287,254 | +287% | - | 16.02ms |
| Static | 6,800 | 1,112,711 | 285,448 | +290% | - | 46.60ms |
| Async DB | 1,024 | 243,187 | 200,350 | +21% | 17.70ms | 24.00ms |
| API-4 | 256 | 46,045 | 39,709 | +16% | 27.50ms | 80.00ms |
| API-16 | 1,024 | 131,689 | 123,488 | +7% | 42.90ms | 105.50ms |
Common comparisons
swoole vs workerman · swoole vs php-fpm · swoole vs hyperf · workerman vs php-fpm · workerman vs hyperf · php-fpm vs hyperf
Questions
Is swoole faster than workerman?
workerman is quicker than swoole on 9 of the 16 profiles they both run, and behind on 7. On the HTTP/1.1 composite they score 707 and 576. Both run on the same machine, in the same round, against the same profiles.
Is swoole faster than php-fpm?
swoole is quicker than php-fpm on 18 of the 22 profiles they both run, and behind on 4. On the HTTP/1.1 composite they score 707 and 278. Both run on the same machine, in the same round, against the same profiles.
Is swoole faster than hyperf?
swoole is quicker than hyperf on 21 of the 21 profiles they both run. On the HTTP/1.1 composite they score 707 and 158. Both run on the same machine, in the same round, against the same profiles.
Is workerman faster than php-fpm?
workerman is quicker than php-fpm on 10 of the 15 profiles they both run, and behind on 5. On the HTTP/1.1 composite they score 576 and 278. Both run on the same machine, in the same round, against the same profiles.
Which PHP framework is fastest overall?
true-async-server holds the highest HTTP/1.1 composite of the PHP entries, at 880. The composite is throughput over the whole suite; if a service is one shape of workload, compare that profile on its own in the table above, because the order changes.
Every PHP entry
- true-async-server
- swoole
- ngx-php
- mark
- workerman
- php-fpm
- symfony-spawn-tas
- frankenphp-trueasync
- hyperf
- symfony-spawn-franken
- laravel
- slim
- workerman-websocket
Results updated · PHP composite ranking · Other languages · Open the leaderboard