RFR: 8310929: Optimization for Integer.toString

[Glavo]

On Wed, 28 Jun 2023 17:06:56 GMT, 温绍锦 <duke at openjdk.org> wrote:

> Speed up Integer.toString with a precomputed cache array of length 1000, use division by 1000, so that each iteration can calculate three digits.  
> 
> this optimization can also be used in StringBuilder#appent(int) and Long#toString.
> 
> # Benchmark Result
> 
> 
> make test TEST="micro:java.lang.Integers.toString*"
> 
> 
> ## [aliyun_ecs_c8i.xlarge](https://help.aliyun.com/document_detail/25378.html#c8i)
> * cpu : intel xeon sapphire rapids (x64)
> ``` diff
> -Benchmark               (size)  Mode  Cnt  Score   Error  Units (baseline)
> -Integers.toStringBig       500  avgt   15  6.811 ± 0.016  us/op
> -Integers.toStringSmall     500  avgt   15  4.794 ± 0.011  us/op
> -Integers.toStringTiny      500  avgt   15  3.752 ± 0.065  us/op
> 
> +Benchmark               (size)  Mode  Cnt  Score   Error  Units (PR)
> +Integers.toStringBig       500  avgt   15  5.164 ± 0.052  us/op +31.89%
> +Integers.toStringSmall     500  avgt   15  3.844 ± 0.025  us/op +24.71%
> +Integers.toStringTiny      500  avgt   15  3.453 ± 0.027  us/op +8.65%
> 
> 
> ### [aliyun_ecs_c8a.xlarge](https://help.aliyun.com/document_detail/25378.html#c8a)
> * cpu : amd epc genoa (x64)
> ``` diff
> -Benchmark               (size)  Mode  Cnt  Score   Error  Units (baseline)
> -Integers.toStringBig       500  avgt   15  6.762 ± 0.018  us/op
> -Integers.toStringSmall     500  avgt   15  4.481 ± 0.004  us/op
> -Integers.toStringTiny      500  avgt   15  2.761 ± 0.012  us/op
> 
> +Benchmark               (size)  Mode  Cnt  Score   Error  Units (PR)
> +Integers.toStringBig       500  avgt   15  5.063 ± 0.175  us/op +33.55%
> +Integers.toStringSmall     500  avgt   15  3.498 ± 0.123  us/op +28.10%
> +Integers.toStringTiny      500  avgt   15  2.743 ± 0.147  us/op +0.65%
> 
> 
> ## [aliyun_ecs_c8y.xlarge](https://help.aliyun.com/document_detail/25378.html#c8y)
> * cpu : aliyun yitian 710 (aarch64)
> ``` diff
> -Benchmark               (size)  Mode  Cnt   Score   Error  Units (baseline)
> -Integers.toStringBig       500  avgt   15  12.081 ± 0.377  us/op
> -Integers.toStringSmall     500  avgt   15   7.119 ± 0.177  us/op
> -Integers.toStringTiny      500  avgt   15   6.323 ± 0.239  us/op
> 
> +Benchmark               (size)  Mode  Cnt  Score   Error  Units (PR)
> +Integers.toStringBig       500  avgt   15  8.157 ± 0.274  us/op +48.10%
> +Integers.toStringSmall     500  avgt   15  7.207 ± 0.330  us/op -1.22%
> +Integers.toStringTiny      500  avgt   15  6.912 ± 0.199  us/op -8.52%

Although this method is `IntrinsicCandidate`, in reality it will only be optimized in the fluent-chain of `StringBuilder`/`StringBuffer`, so optimizing its Java implementation is meaningful.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/14699#issuecomment-1611961052