Function zend_hash_find_bucket consumes much time in WordPress workload #12873
Description
Description
Dear, maintainers:
I am utilizing PHP-FPM and have profiled performance data for a WordPress workload. Within this benchmark, I've identified certain hotspots. The performance analysis from 'perf' highlights that functions such as zend_hash_find and zend_hash_find_known_hash are consuming a considerable amount of time. Upon investigation, I suspect that the core issue lies within the zend_hash_find_bucket function.
Upon inspecting the assembly code and annotating it, I noticed that certain 'mov' instructions are causing significant delays. To address this, I attempted to optimize the zend_hash_find_bucket function. I experimented with optimizations such as utilizing prefetch instructions and employing struct alignment. However, these optimizations did not seem to have a discernible impact on performance.
Percent│
│
│
│ Disassembly of section .text:
│
│ 0000000000532ae0 <zend_hash_find>:
│ zend_hash_find():
0.84 │ push %r12
0.04 │ push %rbp
0.01 │ mov %rsi,%rbp
2.52 │ push %rbx
│ zend_string_hash_val():
8.01 │ mov 0x8(%rsi),%rax
│ zend_hash_find():
0.02 │ mov %rdi,%rbx
│ zend_string_hash_val():
0.12 │ test %rax,%rax
0.14 │ ↓ je 78
│ zend_hash_find_bucket():
5.63 │13: or 0xc(%rbx),%eax
0.28 │ mov 0x10(%rbx),%r12
0.16 │ cltq
25.03 │ mov (%r12,%rax,4),%ebx
1.98 │ cmp $0xffffffff,%ebx
0.56 │ ↓ je a0
0.85 │ shl $0x5,%rbx
0.50 │ add %r12,%rbx
15.20 │ cmp 0x18(%rbx),%rbp
12.97 │ ↓ jne 68
│ zend_hash_find():
1.41 │32: mov %rbx,%rax
0.60 │ pop %rbx
0.03 │ pop %rbp
1.25 │ pop %r12
1.83 │ ← ret
│ nop
│ zend_hash_find_bucket():
0.26 │40: mov 0x18(%rbx),%rdi
0.04 │ test %rdi,%rdi
│ ↓ je 53
│ zend_string_equal_content():
0.01 │ mov 0x10(%rbp),%rax
0.01 │ cmp %rax,0x10(%rdi)
5.03 │ ↓ je 90
│ zend_hash_find_bucket():
1.06 │53: mov 0xc(%rbx),%ebx
0.11 │ cmp $0xffffffff,%ebx
0.15 │ ↓ je a0
0.33 │ shl $0x5,%rbx
0.24 │ add %r12,%rbx
1.90 │ cmp 0x18(%rbx),%rbp
2.69 │ ↑ je 32
2.00 │68: mov 0x8(%rbp),%rax
0.24 │ cmp %rax,0x10(%rbx)
Is this a known performance issue within the PHP community? I'm seeking suggestions or insights on potential avenues for optimization. Your feedback and suggestions would be greatly appreciated.
PHP Version
master
Operating System
No response