benwtrent opened a new issue, #12621:
URL: https://github.com/apache/lucene/issues/12621
### Description
While testing and digging around, I noticed that our float comparisons are
way faster than byte on my Macbook (M1) and pretty much the same as our byte
comparisons on a GCP Intel Sapphire Rapids CPU.
This seems counter-intuitive to me. I would expect Panama to be able to do
more `byte` operations per cycle than `float`. My guess is the intrinsics are
weird? Panama Vector just doesn't support or detect the required operations?
Here are two benchmark results using @rmuir's helpful vectorbench project:
MacBook (Apple Silicon [128bits], JDK21):
```
FloatDotProductBenchmark.dotProductNew 768 thrpt 5 21.781 ± 0.254
ops/us
FloatDotProductBenchmark.dotProductNew 1024 thrpt 5 15.091 ± 0.217
ops/us
BinaryDotProductBenchmark.dotProductNew 768 thrpt 5 8.041 ± 0.108
ops/us
BinaryDotProductBenchmark.dotProductNew 1024 thrpt 5 6.085 ± 0.133
ops/us
```
GCP (Intel Sapphire Rapids [avx512], JDK21):
```
FloatDotProductBenchmark.dotProductNew 768 thrpt 5 20.169 ± 0.385
ops/us
FloatDotProductBenchmark.dotProductNew 1024 thrpt 5 18.334 ± 0.180
ops/us
BinaryDotProductBenchmark.dotProductNew 768 thrpt 5 19.686 ± 0.050
ops/us
BinaryDotProductBenchmark.dotProductNew 1024 thrpt 5 14.934 ± 0.014
ops/us
```
<details>
<summary>cpu-flags</summary>
```
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep
mtrr pge mca cmov pat pse36 clflus
h mmx fxsr sse sse2 ss ht syscall nx
pdpe1gb rdtscp lm constant_tsc rep_good n
opl xtopology nonstop_tsc cpuid
tsc_known_freq pni pclmulqdq ssse3 fma cx16 pc
id sse4_1 sse4_2 x2apic movbe popcnt aes
xsave avx f16c rdrand hypervisor lahf
_lm abm 3dnowprefetch invpcid_single ssbd
ibrs ibpb stibp ibrs_enhanced fsgsba
se tsc_adjust bmi1 avx2 smep bmi2 erms
invpcid rtm avx512f avx512dq rdseed adx
smap avx512ifma clflushopt clwb avx512cd
sha_ni avx512bw avx512vl xsaveopt xs
avec xgetbv1 xsaves avx512_bf16 arat
avx512vbmi umip avx512_vbmi2 gfni vaes vp
clmulqdq avx512_vnni avx512_bitalg
avx512_vpopcntdq rdpid cldemote movdiri mov
dir64b fsrm md_clear serialize
arch_capabilities
```
</details>
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
