fastjson_benchmark

更多测试数据看这里 https://github.com/alibaba/fastjson2/tree/main/docs/benchmark

1. 测试环境

1.1 各个库的版本

fastjson2 2.0.34
fastjson1 1.1.83
jackson 2.15.1
gson 2.10.1
kryo 5.5.0
hessian 4.0.66

都是截止测试时间（2023-06-18）各个库的最新版本

1.2 服务器环境

测试的服务器基于阿里云最新代ECS，包括x64架构的Intel和阿里云平头哥的倚天处理器。

ecs.c8i.xlarge
ecs.c8y.xlarge
Orange Pi 5 Plus

推荐阿里云平头哥Yitian 170 CPU的机器，性能快20%，价格便宜40%

ecs.c8i.xlarge

这个是阿里云当前代标准型ECS，处理器型号 Intel Xeon(Sapphire Rapids) Platinum 8475B，4核，8G内存

按量付费每小时0.8562元

ecs.c8y.xlarge

这个是阿里云上售卖ARM处理器，处理器型号 Yitian 710，4核，8G内存。这个是阿里云平头哥的倚天710处理器。

按量付费每个小时0.533866元

Orange Pi 5 Plus

这个是国产的树莓派的高性能替代 http://www.orangepi.cn/html/hardWare/computerAndMicrocontrollers/details/Orange-Pi-5-plus.html 8核16G内存天猫价949元

1.3 JDK版本

基于Oracle最新版本的Linux x64/aarch64的JDK版本，下载地址 https://www.oracle.com/java/technologies/

oracle-jdk1.8.0_371
oracle-jdk-11.0.19
oracle-jdk-17.0.7
graalvm-17.0.7

1.4 测试代码以及运行方式

代码路径 https://github.com/alibaba/fastjson2/tree/2.0.34/benchmark/src/main/java/com/alibaba/fastjson2/benchmark/eishay
运行方式测试代码方式需要准备响应的JDK版本到Install目录下

git clone https://github.com/alibaba/fastjson2
cd fastjson2
git checkout 2.0.34
mvn clean install -Dmaven.test.skip
~/Install/jdk8/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay
~/Install/jdk11/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay
~/Install/jdk17/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay
~/Install/graalvm-17/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay

2. 测试数据&结果分析

原始数据 https://github.com/alibaba/fastjson2/blob/2.0.34/docs/benchmark/benchmark_2.0.34_raw.md

2.1 EishayParseString

这个是最常用的场景，将JSON格式字符串反序列化为Java对象，这个场景在fastjson中的代码如下：

String str = "...";
Bean bean = JSON.parseJSONObject(str, Bean.class);

这个场景是fastjson1最强的场景；

在ecs.c8i.large-jdk11.0.19环境中，fastjson1只相当于只相当于fastjson2的71.55% ;jackson则只相当于fastjson2的35.21%; gson是fastjson2的35.36%，也就是说这个最常见的反序列化场景，fastjson2的性能差不多是jackson/gson的三倍。

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2 > fastjson1 > jackson > gson
jdk8  100%        85%         41.67%    37.08%
jdk11 100%        71.55%      35.21%    35.68%
jdk17 100%        91.91%      35.91%    36.29%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	2582.255	2194.912 (85%)	1076.041 (41.67%)	957.543 (37.08%)
	jdk-11.0.19	2753.434	1970.107 (71.55%)	969.615 (35.21%)	982.558 (35.68%)
	jdk-17.0.7	2879.11	2646.103 (91.91%)	1033.83 (35.91%)	1044.744 (36.29%)
	graalvm_17.0.7+8.1	2877.882	3490.823 (121.3%)	1074.694 (37.34%)	1045.023 (36.31%)
aliyun_ecs.g8y.large	jdk1.8.0_371	4013.19	3080.664 (76.76%)	1387.022 (34.56%)	1414.971 (35.26%)
	jdk-11.0.19	4150.442	3027.893 (72.95%)	1352.156 (32.58%)	1423.13 (34.29%)
	jdk-17.0.7	4247.923	4114.055 (96.85%)	1420.088 (33.43%)	1425.616 (33.56%)
	graalvm_17.0.7+8.1	4564.054	4711.368 (103.23%)	1456.355 (31.91%)	1570.283 (34.41%)
orangepi5p	jdk1.8.0_371	3163.978	2663.744 (84.19%)	1190.413 (37.62%)	1067.005 (33.72%)
	jdk-11.0.19	3245.9	2561.117 (78.9%)	1105.771 (34.07%)	1072.671 (33.05%)
	jdk-17.0.7	3270.943	3567.236 (109.06%)	1204.83 (36.83%)	1087.251 (33.24%)
	graalvm_17.0.7+8.1	3596.944	4217.008 (117.24%)	1162.206 (32.31%)	1148.028 (31.92%)

2.2 EishayParseStringPretty

这个场景是将格式化过的JSON格式字符串反序列化为Java对象，这个场景fastjson1的parse算法不擅长的。这个场景fastjson2表现仍然非常好，性能是jackson/gson两倍以上。这个场景在ecs.c7.xlarge-jdk1.8.0_361环境下，fastjson1的性能是fastjson2的27.83%，jackson的性能是fastjson2的50.51%，gson是fastjson2的42.48%%.

这个场景在fastjson中的代码如下：

// 这里输入的是格式化过后的json字符串
String str = "{\n" +
        "\t\"id\":123\n" +
        "}";
Bean bean = JSON.parseJSONObject(str, Bean.class);

性能排序分别如下：

// aliyun_ecs.g8i.large

      fastjson2 > jackson > gson  > fastjson1
jdk8  100%        50.31%    47.97%  30.03%
jdk11 100%        45.21%    42.53%  29.13%
jdk17 100%        46.78%    45.19%  31.89%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	1951.123	586.008 (30.03%)	981.529 (50.31%)	935.943 (47.97%)
	jdk-11.0.19	1980.14	576.837 (29.13%)	895.231 (45.21%)	842.114 (42.53%)
	jdk-17.0.7	1936.139	617.388 (31.89%)	905.749 (46.78%)	874.874 (45.19%)
	graalvm_17.0.7+8.1	2074.951	860.298 (41.46%)	962.607 (46.39%)	983.036 (47.38%)
aliyun_ecs.g8y.large	jdk1.8.0_371	1816.268	863.233 (47.53%)	1266.37 (69.72%)	1303.309 (71.76%)
	jdk-11.0.19	3198.409	904.486 (28.28%)	1249.957 (39.08%)	1338.467 (41.85%)
	jdk-17.0.7	2997.572	959.907 (32.02%)	1299.603 (43.36%)	1322.768 (44.13%)
	graalvm_17.0.7+8.1	3277.775	1298.414 (39.61%)	1325.719 (40.45%)	1430.954 (43.66%)
orangepi5p	jdk1.8.0_371	2579.831	713.589 (27.66%)	1111.452 (43.08%)	964.844 (37.4%)
	jdk-11.0.19	2593.822	718.589 (27.7%)	1053.424 (40.61%)	990.794 (38.2%)
	jdk-17.0.7	2577.714	829.781 (32.19%)	1106.165 (42.91%)	1001.714 (38.86%)
	graalvm_17.0.7+8.1	2712.799	980.767 (36.15%)	1050.036 (38.71%)	1039.594 (38.32%)

2.3 EishayParseTreeString

这个场景是将JSON格式字符串反序列化为JSONObject，这也是一个最常见的场景之一，这个场景在fastjson中的代码如下：

String str = "...";
JSONObject jsonObject = JSON.parseJSONObject(str);

fastjson2在这个场景也表现了远超fastjson1/jackson/gson的性能。在不同的环境中，fastjson1和jackson都只能相当于fastjson2性能的40%~60%，gson则更差一些。

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2 > fastjson1 > jackson  > gson
jdk8  100%        54.31%      53.78%     38.68%
jdk11 100%        51.64%      51.53%     40.37%
jdk17 100%        50.91%      47.02%     33.00%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	2157.731	1171.904 (54.31%)	1160.387 (53.78%)	834.626 (38.68%)
	jdk-11.0.19	1967.972	1016.21 (51.64%)	1014.106 (51.53%)	794.442 (40.37%)
	jdk-17.0.7	2460.387	1252.679 (50.91%)	1156.897 (47.02%)	811.927 (33%)
	graalvm_17.0.7+8.1	2494.678	1493.373 (59.86%)	1213.447 (48.64%)	1074.36 (43.07%)
aliyun_ecs.g8y.large	jdk1.8.0_371	2544.851	1335.47 (52.48%)	1169.094 (45.94%)	1075.553 (42.26%)
	jdk-11.0.19	2831.14	1352.26 (47.76%)	1259.093 (44.47%)	1127.018 (39.81%)
	jdk-17.0.7	2821.527	1434.44 (50.84%)	1369.809 (48.55%)	1129.4 (40.03%)
	graalvm_17.0.7+8.1	3211.594	2145.957 (66.82%)	1553.492 (48.37%)	1381.737 (43.02%)
orangepi5p	jdk1.8.0_371	2081.1	1066.732 (51.26%)	1150.45 (55.28%)	892.438 (42.88%)
	jdk-11.0.19	2174.288	1061.327 (48.81%)	1103.669 (50.76%)	883.381 (40.63%)
	jdk-17.0.7	2228.951	1304.163 (58.51%)	1198.572 (53.77%)	874.087 (39.22%)
	graalvm_17.0.7+8.1	2394.753	1683.816 (70.31%)	1270.827 (53.07%)	999.579 (41.74%)

2.4 EishayParseUTF8Bytes

这个是将UTF8格式的byte数组反序列化为Java对象，这个场景在缓存和RPC场景中常用。

这个场景在fastjson中的代码如下：

byte[] utf8Bytes = ...;
Bean bean = JSON.parseJSONObject(utf8Bytes, Bean.class);

这个场景fastjson2同样表现出了卓越的性能；在JDK11下，fastjson1和jackson的性能分别之后fastjson2的72.39%和45.97%; gson不直接支持输入utf8Bytes，需要先构造字符串，性能只有fastjson2的34.14%。

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2 > fastjson1 > jackson  > gson
jdk8  100%        76.55%      49.58%     37.17%
jdk11 100%        72.39%      45.97%     41.59%
jdk17 100%        70.86%      36.05%     33.66%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	2416.504	1849.718 (76.55%)	1198.099 (49.58%)	898.149 (37.17%)
	jdk-11.0.19	2295.677	1661.76 (72.39%)	1055.356 (45.97%)	954.814 (41.59%)
	jdk-17.0.7	2927.294	2074.57 (70.87%)	1055.151 (36.05%)	985.241 (33.66%)
	graalvm_17.0.7+8.1	2982.08	2482.913 (83.26%)	1184.176 (39.71%)	1018.019 (34.14%)
aliyun_ecs.g8y.large	jdk1.8.0_371	3566.348	2660.834 (74.61%)	1544.212 (43.3%)	1298.701 (36.42%)
	jdk-11.0.19	4219.927	2479.671 (58.76%)	1507.776 (35.73%)	1342.897 (31.82%)
	jdk-17.0.7	4249.983	3111.088 (73.2%)	1475.474 (34.72%)	1338.74 (31.5%)
	graalvm_17.0.7+8.1	4686.092	3145.142 (67.12%)	1366.381 (29.16%)	1484.494 (31.68%)
orangepi5p	jdk1.8.0_371	2798.083	2052.691 (73.36%)	1386.798 (49.56%)	932.633 (33.33%)
	jdk-11.0.19	3164.784	2041.79 (64.52%)	1294.929 (40.92%)	1010.805 (31.94%)
	jdk-17.0.7	3278.378	2659.782 (81.13%)	1323.835 (40.38%)	1043.902 (31.84%)
	graalvm_17.0.7+8.1	3652.372	2285.648 (62.58%)	1218.417 (33.36%)	1096.793 (30.03%)

2.5 EishayWriteString

这个是Java对象序列化成字符串，这个是序列化最常用的场景之一。这个场景在fastjson中的代码如下：

Bean bean = ...;
String str = JSON.toJSONString(bean);

在这个最常用的序列化场景，fastjson2表现出远超fastjson1/jackson/gson的性能；这个场景gson表现很差，只有fastjson2的20~30%。

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2 > jackson >   fastjson1  > gson
jdk8  100%        66.86%      46.91%       29.33%
jdk11 100%        64.27%      46.63%       25.89%
jdk17 100%        58.23%      41.18%       19.61%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	3351.555	1572.199 (46.91%)	2241.006 (66.86%)	982.929 (29.33%)
	jdk-11.0.19	3050.54	1422.379 (46.63%)	1960.731 (64.27%)	789.742 (25.89%)
	jdk-17.0.7	3663.698	1508.58 (41.18%)	2133.552 (58.23%)	718.56 (19.61%)
	graalvm_17.0.7+8.1	5020.114	1745.964 (34.78%)	2506.563 (49.93%)	687.009 (13.69%)
aliyun_ecs.g8y.large	jdk1.8.0_371	4600.757	2073.222 (45.06%)	2471.014 (53.71%)	1338.269 (29.09%)
	jdk-11.0.19	5150.534	1932.703 (37.52%)	2547.575 (49.46%)	1105.371 (21.46%)
	jdk-17.0.7	5344.878	2149.421 (40.21%)	2765.983 (51.75%)	844.873 (15.81%)
	graalvm_17.0.7+8.1	6792.175	3506.425 (51.62%)	3331.736 (49.05%)	875.97 (12.9%)
orangepi5p	jdk1.8.0_371	3914.563	3192.798 (81.56%)	1996.57 (51%)	955.556 (24.41%)
	jdk-11.0.19	4420.479	1460.266 (33.03%)	2093.505 (47.36%)	920.034 (20.81%)
	jdk-17.0.7	4504.858	1640.45 (36.42%)	2146.788 (47.65%)	832.334 (18.48%)
	graalvm_17.0.7+8.1	4853.059	1952.938 (40.24%)	2304.794 (47.49%)	798.729 (16.46%)

2.6 EishayWriteUTF8Bytes

这个是Java对象序列化成UTF格式的byte数组，这个是最常用的场景之一，在缓存和RPC场景常用。gson不直接支持，需要先序列化为String再转换为UTF8格式的byte数组。

这个场景在fastjson中的代码如下：

Bean bean = ...;
byte[] utf8Bytes = JSON.toJSONBytes(bean);

在这个最常用的序列化场景，fastjson2表现出远超fastjson1/jackson/gson的性能；这个场景gson表现很差。

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2 > jackson >   fastjson1  > gson
jdk8  100%        61.94%      44.83%       28.00%
jdk11 100%        55.59%      47.40%       22.91%
jdk17 100%        47.99%      32.03%       17.20%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	3117.898	1397.729 (44.83%)	1931.227 (61.94%)	872.899 (28%)
	jdk-11.0.19	3290.208	1559.448 (47.4%)	1829.172 (55.59%)	753.726 (22.91%)
	jdk-17.0.7	4094.515	1311.418 (32.03%)	1964.922 (47.99%)	704.437 (17.2%)
	graalvm_17.0.7+8.1	5293.463	1574.744 (29.75%)	2496.69 (47.17%)	692.987 (13.09%)
aliyun_ecs.g8y.large	jdk1.8.0_371	5066.945	1877.169 (37.05%)	2531.139 (49.95%)	1165.513 (23%)
	jdk-11.0.19	5395.098	2641.487 (48.96%)	2402.697 (44.53%)	1128.482 (20.92%)
	jdk-17.0.7	5484.846	1934.638 (35.27%)	2702.184 (49.27%)	816.215 (14.88%)
	graalvm_17.0.7+8.1	7678.186	2230.843 (29.05%)	3437.981 (44.78%)	832.257 (10.84%)
orangepi5p	jdk1.8.0_371	4335.683	1421.801 (32.79%)	2099.996 (48.44%)	805.302 (18.57%)
	jdk-11.0.19	4675.018	1355.289 (28.99%)	1887.46 (40.37%)	900.066 (19.25%)
	jdk-17.0.7	4850.415	1419.481 (29.27%)	2137.183 (44.06%)	754.629 (15.56%)
	graalvm_17.0.7+8.1	5873.62	1975.705 (33.64%)	2479.081 (42.21%)	798.512 (13.59%)

2.7 EishayWriteStringTree

这个场景是JSONObject序列化成字符串，这个是序列化最常用的场景之一。这个场景在fastjson中的代码如下：

JSONObject jsonObject = ...;
String str = jsonObject.toJSONString();

在这个最常用的序列化场景，fastjson2表现出远超fastjson1/jackson/gson的性能；这个场景gson表现很差，只有fastjson2的1/3

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2 > jackson >   fastjson1  > gson
jdk8  100%        60.01%      60.00%       33.35%
jdk11 100%        71.79%      70.46%       32.87%
jdk17 100%        51.64%      59.24%       23.41%

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
aliyun_ecs.g8i.large	jdk1.8.0_371	3261.752	1957.175 (60%)	2120.432 (65.01%)	1087.704 (33.35%)
	jdk-11.0.19	2578.451	1816.697 (70.46%)	1850.786 (71.78%)	847.599 (32.87%)
	jdk-17.0.7	3121.17	1849.111 (59.24%)	1611.837 (51.64%)	730.516 (23.41%)
	graalvm_17.0.7+8.1	3445.437	2106.462 (61.14%)	2172.493 (63.05%)	731.296 (21.23%)
aliyun_ecs.g8y.large	jdk1.8.0_371	3734.306	2265.691 (60.67%)	2534.349 (67.87%)	1614.077 (43.22%)
	jdk-11.0.19	3813.869	2424.057 (63.56%)	2559.903 (67.12%)	1365.337 (35.8%)
	jdk-17.0.7	4778.902	2268.404 (47.47%)	2507.172 (52.46%)	877.351 (18.36%)
	graalvm_17.0.7+8.1	4529.66	3241.686 (71.57%)	3095.552 (68.34%)	902.854 (19.93%)
orangepi5p	jdk1.8.0_371	2686.429	1767.555 (65.8%)	2019.632 (75.18%)	1020.227 (37.98%)
	jdk-11.0.19	2689.171	1783.27 (66.31%)	2091.617 (77.78%)	1033.408 (38.43%)
	jdk-17.0.7	3132.018	1797.459 (57.39%)	2089.26 (66.71%)	861.928 (27.52%)
	graalvm_17.0.7+8.1	3264.53	2195.442 (67.25%)	2329.688 (71.36%)	849.392 (26.02%)

2.8 EishayWriteBinaryAutoType

这个场景是将JavaBean序列化成二进制格式并且带类型信息，这个是代替Java/Hessian序列化的场景，代码如下：

JSONB.toBytes(
        mc,
        JSONWriter.Feature.WriteClassName,
        JSONWriter.Feature.IgnoreNoneSerializable,
        JSONWriter.Feature.FieldBased,
        JSONWriter.Feature.ReferenceDetection,
        JSONWriter.Feature.WriteNulls,
        JSONWriter.Feature.NotWriteDefaultValue,
        JSONWriter.Feature.NotWriteHashMapArrayListClassName,
        JSONWriter.Feature.WriteNameAsSymbol
)

这个场景下，fastjson_jsonb的性能是hessian的4~5倍

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2_jsonb > hessian  > javaSerialize
jdk8  100%              25.87%     15.56%
jdk11 100%              28.08%     17.61%
jdk17 100%              24.30%     15.89%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是hessian/fastjson2_jsonb的百分数表示。

aliyun ecs spec	jdk version	fastjson2JSONB	hessian	javaSerialize
aliyun_ecs.g8i.large	jdk1.8.0_371	3361.737	869.574 (25.87%)	523.171 (15.56%)
	jdk-11.0.19	2939.23	825.242 (28.08%)	517.6 (17.61%)
	jdk-17.0.7	3369.112	818.714 (24.3%)	535.34 (15.89%)
	graalvm_17.0.7+8.1	4690.64	1256.73 (26.79%)	572.517 (12.21%)
aliyun_ecs.g8y.large	jdk1.8.0_371	4575.932	1189.073 (25.99%)	825.689 (18.04%)
	jdk-11.0.19	4540.034	1029.985 (22.69%)	820.761 (18.08%)
	jdk-17.0.7	4404.005	1243.94 (28.25%)	817.773 (18.57%)
	graalvm_17.0.7+8.1	6289.787	1302.702 (20.71%)	817.432 (13%)
orangepi5p	jdk1.8.0_371	3584.42	682.634 (19.04%)	626.146 (17.47%)
	jdk-11.0.19	3645.459	683.877 (18.76%)	651.808 (17.88%)
	jdk-17.0.7	3490.94	692.969 (19.85%)	650.79 (18.64%)
	graalvm_17.0.7+8.1	4493.808	704.132 (15.67%)	601.961 (13.4%)

2.9 EishayParseBinaryAutoType

这个场景是JavaBean二进制带类型信息反序列化的场景，这个是代替Java/Hessian反序列化的场景，代码如下：

JSONB.parseObject(
        fastjson2JSONBBytes,
        Object.class,
        JSONReader.Feature.SupportAutoType,
        JSONReader.Feature.IgnoreNoneSerializable,
        JSONReader.Feature.UseDefaultConstructorAsPossible,
        JSONReader.Feature.UseNativeObject,
        JSONReader.Feature.FieldBased)

这个场景下，fastjson2二进制格式性能是hessian的5~~8倍，是Java序列化的30~~40倍

性能排序如下：

// aliyun_ecs.g8i.large

      fastjson2_jsonb > hessian  > javaSerialize
jdk8  100%              17.82%     3.08%
jdk11 100%              14.25%     2.68%
jdk17 100%              13.54%     2.72%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是hessian/fastjson2_jsonb的百分数表示。

aliyun ecs spec	jdk version	fastjson2JSONB	hessian	javaSerialize
aliyun_ecs.g8i.large	jdk1.8.0_371	3794.594	676.044 (17.82%)	117.058 (3.08%)
	jdk-11.0.19	4339.484	618.233 (14.25%)	116.458 (2.68%)
	jdk-17.0.7	4779.001	647.288 (13.54%)	130.006 (2.72%)
	graalvm_17.0.7+8.1	5181.598	794.567 (15.33%)	141.942 (2.74%)
aliyun_ecs.g8y.large	jdk1.8.0_371	5430.68	331.514 (6.1%)	162.923 (3%)
	jdk-11.0.19	5729.123	441.793 (7.71%)	156.723 (2.74%)
	jdk-17.0.7	5907.579	393.633 (6.66%)	161.934 (2.74%)
	graalvm_17.0.7+8.1	6358.521	541.31 (8.51%)	170.87 (2.69%)
orangepi5p	jdk1.8.0_371	3795.223	182.861 (4.82%)	136.707 (3.6%)
	jdk-11.0.19	4189.609	184.741 (4.41%)	128.911 (3.08%)
	jdk-17.0.7	4261.107	192.88 (4.53%)	144.517 (3.39%)
	graalvm_17.0.7+8.1	4998.054	239.668 (4.8%)	145.492 (2.91%)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fastjson_benchmark

1. 测试环境

1.1 各个库的版本

1.2 服务器环境

ecs.c8i.xlarge

ecs.c8y.xlarge

Orange Pi 5 Plus

1.3 JDK版本

1.4 测试代码以及运行方式

2. 测试数据&结果分析

2.1 EishayParseString

2.2 EishayParseStringPretty

2.3 EishayParseTreeString

2.4 EishayParseUTF8Bytes

2.5 EishayWriteString

2.6 EishayWriteUTF8Bytes

2.7 EishayWriteStringTree

2.8 EishayWriteBinaryAutoType

2.9 EishayParseBinaryAutoType

Clone this wiki locally