RFR: 8347009: Speed ​​up parseInt and parseLong [v3]

[Shaojin Wen]

On Sat, 4 Jan 2025 12:09:04 GMT, Shaojin Wen <swen at openjdk.org> wrote:

>> This is an optimization for decimal Integer.parseInt and Long.parseLong, which improves performance by about 10%. The optimization includes:
>> 1. Improve performance by parsing 2 numbers at a time, which has performance improvements for numbers with length >= 3.
>> 2. It uses charAt(0) for the first number. Assuming that the optimization can eliminate boundary checks, this will be more friendly to parsing numbers with length 1.
>> 3. It removes the reliance on the Character.digit method and eliminates the reliance on the CharacterDataLatin1#DIGITS cache array, which avoids performance degradation caused by cache misses.
>
> Shaojin Wen has updated the pull request incrementally with one additional commit since the last revision:
> 
>   use String::value

Below are the performance numbers on each platform

## 1. Script

git remote add wenshao git at github.com:wenshao/jdk.git
git fetch wenshao

#master
git checkout 5bb9644202a8b961faad02439c95a81ca6862168
make test TEST="micro:java.lang.Longs.parseLong"
make test TEST="micro:java.lang.Integers.parseInt" 

 # current
git checkout 9ea084555792fb3c870f6393ce53873e9bde73e6
make test TEST="micro:java.lang.Longs.parseLong"
make test TEST="micro:java.lang.Integers.parseInt" 


## 2. aliyun_ecs_c8a_x64 (CPU AMD EPYC™ Genoa)

-Benchmark          (size)  Mode  Cnt  Score   Error  Units
-Longs.parseLong       500  avgt   15  3.542 ± 0.022  us/op
-Integers.parseInt     500  avgt   15  3.269 ± 0.015  us/op

+Benchmark          (size)  Mode  Cnt  Score   Error  Units
+Longs.parseLong       500  avgt   15  3.240 ± 0.037  us/op +9.32%
+Integers.parseInt     500  avgt   15  3.208 ± 0.004  us/op +1.90%


## 3. aliyun_ecs_c8i_x64 (CPU Intel®Xeon®Emerald Rapids)

-Benchmark          (size)  Mode  Cnt  Score   Error  Units
-Longs.parseLong       500  avgt   15  3.477 ± 0.009  us/op
-Integers.parseInt     500  avgt   15  3.138 ± 0.006  us/op

+Benchmark          (size)  Mode  Cnt  Score   Error  Units
+Longs.parseLong       500  avgt   15  3.023 ± 0.028  us/op +15.01%
+Integers.parseInt     500  avgt   15  2.847 ± 0.005  us/op +10.22%


## 4. aliyun_ecs_c8y_aarch64 (CPU Aliyun Yitian 710 ARM)

-Benchmark          (size)  Mode  Cnt  Score   Error  Units
-Longs.parseLong       500  avgt   15  3.819 ± 0.044  us/op
-Integers.parseInt     500  avgt   15  3.639 ± 0.075  us/op

+Benchmark          (size)  Mode  Cnt  Score   Error  Units
+Longs.parseLong       500  avgt   15  3.347 ± 0.015  us/op +14.10%
+Integers.parseInt     500  avgt   15  3.192 ± 0.010  us/op +14.03%


## 5. MacBook M1 Pro (aarch64)

-Benchmark          (size)  Mode  Cnt  Score   Error  Units
-Longs.parseLong       500  avgt   15  2.608 ? 0.035  us/op
-Integers.parseInt     500  avgt   15  2.325 ? 0.005  us/op

+Benchmark          (size)  Mode  Cnt  Score   Error  Units
+Longs.parseLong       500  avgt   15  2.291 ? 0.018  us/op +13.83%
+Integers.parseInt     500  avgt   15  2.079 ? 0.011  us/op +11.83%

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

PR Comment: https://git.openjdk.org/jdk/pull/22919#issuecomment-2571272942