curve25519-dalek中field域内和scalar域内的运算性能对比

通过修改lib.rs文件，将field和backendmodule改为pub。
通过性能对比发现在curve25519-dalek库的代码实现，实际field域内计算速度优于scalar域内的计算性能。

针对field域内的加减乘除和scalar域内的加减乘除运算对应的bench代码如下：

mod scalar_benches {use super::*;fn scalar_inversion(c: &mut Criterion) {c.bench_function("Scalar inversion", |b| {let s = Scalar::from(897987897u64).invert();b.iter(|| s.invert());});}fn scalar_mul(c: &mut Criterion) {c.bench_function("Scalar multiplication", |b| {let s = Scalar::from(897987897u64).invert();b.iter(|| s*s);});}fn scalar_add(c: &mut Criterion) {c.bench_function("Scalar add", |b| {let s = Scalar::from(897987897u64).invert();b.iter(|| s+s);});}fn scalar_sub(c: &mut Criterion) {c.bench_function("Scalar sub", |b| {let s = Scalar::from(897987897u64).invert();b.iter(|| s-s);});}fn batch_scalar_inversion(c: &mut Criterion) {c.bench_function_over_inputs("Batch scalar inversion",|b, &&size| {let mut rng = OsRng::new().unwrap();let scalars: Vec<Scalar> = (0..size).map(|_| Scalar::random(&mut rng)).collect();b.iter(|| {let mut s = scalars.clone();Scalar::batch_invert(&mut s);});},&BATCH_SIZES,);}criterion_group! {name = scalar_benches;config = Criterion::default();targets =scalar_inversion,scalar_mul,scalar_add,scalar_sub,//batch_scalar_inversion,}
}mod field_benches {use super::*;fn field_inversion(c: &mut Criterion) {c.bench_function("field inversion", |b| {let a: [u8; 32] = [ //0x35863539  as 897987897u640x39, 0x35, 0x86, 0x35, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00];let s = FieldElement::from_bytes(&a).invert();b.iter(|| s.invert());});}fn field_mul(c: &mut Criterion) {c.bench_function("field multiplication", |b| {let a: [u8; 32] = [ //0x35863539  as 897987897u640x39, 0x35, 0x86, 0x35, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00];let s = FieldElement::from_bytes(&a).invert();b.iter(|| &s * &s );});}fn field_add(c: &mut Criterion) {c.bench_function("field add", |b| {let a: [u8; 32] = [ //0x35863539  as 897987897u640x39, 0x35, 0x86, 0x35, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00];let s = FieldElement::from_bytes(&a).invert();b.iter(|| &s + &s );});}fn field_sub(c: &mut Criterion) {c.bench_function("field sub", |b| {let a: [u8; 32] = [ //0x35863539  as 897987897u640x39, 0x35, 0x86, 0x35, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00];let s = FieldElement::from_bytes(&a).invert();b.iter(|| &s - &s );});}criterion_group! {name = field_benches;config = Criterion::default();targets =field_inversion,field_mul,field_sub,field_add,}
}

在1核4G内存Ubuntu16.04系统下运行性能如下：

unning target/release/deps/dalek_benchmarks-53fcb1faec6cb376
Scalar inversion        time:   [11.761 us 11.819 us 11.893 us]change: [-13.746% +0.2914% +17.272%] (p = 0.97 > 0.05)No change in performance detected.
Found 17 outliers among 100 measurements (17.00%)1 (1.00%) high mild16 (16.00%) high severeScalar multiplication   time:   [170.81 ns 193.62 ns 218.47 ns]change: [+2.0752% +17.017% +34.861%] (p = 0.03 < 0.05)Performance has regressed.
Found 5 outliers among 100 measurements (5.00%)3 (3.00%) high mild2 (2.00%) high severeScalar add              time:   [63.678 ns 64.160 ns 64.790 ns]
Found 18 outliers among 100 measurements (18.00%)18 (18.00%) high severeScalar sub              time:   [63.023 ns 63.360 ns 63.790 ns]
Found 17 outliers among 100 measurements (17.00%)1 (1.00%) high mild16 (16.00%) high severefield inversion         time:   [3.6348 us 3.6528 us 3.6763 us]change: [-9.9310% +2.1958% +16.053%] (p = 0.74 > 0.05)No change in performance detected.
Found 17 outliers among 100 measurements (17.00%)17 (17.00%) high severefield multiplication    time:   [27.161 ns 27.300 ns 27.479 ns]change: [-11.422% +1.2224% +15.799%] (p = 0.86 > 0.05)No change in performance detected.
Found 19 outliers among 100 measurements (19.00%)2 (2.00%) high mild17 (17.00%) high severefield sub               time:   [11.746 ns 11.810 ns 11.889 ns]
Found 17 outliers among 100 measurements (17.00%)17 (17.00%) high severefield add               time:   [11.541 ns 11.729 ns 11.965 ns]
Found 22 outliers among 100 measurements (22.00%)1 (1.00%) high mild21 (21.00%) high severe

curve25519-dalek中field域内和scalar域内的运算性能对比相关推荐

property field java_Java 中 field 和 variable 区别及相关术语解释
原标题:Java 中 field 和 variable 区别及相关术语解释先说一下 field 和 variable 之间的区别: class variables and instance vari ...
Solr中Field常用属性
FieldType 实例:<fieldType name="text_ik" class="solr.TextField"></fieldTy ...
js中几种实用的跨域方法原理详解
这里说的js跨域是指通过js在不同的域之间进行数据传输或通信,比如用ajax向一个不同的域请求数据,或者通过js获取页面中不同域的框架中(iframe)的数据.只要协议.域名.端口有任何一个不同,都被 ...
内网渗透-域内有网和无网
文章目录前言域内主机有网攻击流程域内主机无网操作流程前言在渗透测试的过程中,拿下域内一台主机以后,有时候内网的其他主机有网,有时候没网,此时,如果有网我们就按常规的渗透测试即可,如果没网 ...
内网渗透-域渗透简单思路
文章目录 1.权限提升 2.渗透姿势爆hash,爆aes key 爆hash 爆aes key 3.远程登录注册表修改连接命令 4.连接操作 PTH攻击(利用明文或hash连接) IPC(明文连 ...
内网渗透-域内信息收集
文章目录域内基础信息收集查询域查询域内所有计算机查询所有域成员计算机列表获取域信任查询域控器查域控器的主机名查询域控组获取域内用户和管理员信息查询所有域用户列表向控制器进行查询 ...
sqlserver2000给账户授予所有的权限_内网渗透 | 域内权限解读
域内权限解读目录域本地组全局组通用组 A-G-DL-P策略内置组几个比较重要的域本地组几个比较重要的全局组.通用组的权限域本地组多域用户访问单域资源(访问同一个域) 可以从任何域添加 ...
企业级内网的域控环境搭建教程
所谓域控就是一台或多台域控制器能够控制域内的其他服务器,可实现统一更换电脑桌面,统一修改密码等诸多操作,就像网吧里的网络结构一样,只不过更复杂,更庞大,下面是详细部署教程说明:要做这个实验需要模拟内 ...
内网安全-域横向PTHPTKPTT哈希票据传递
PTH(pass the hash) #利用 lm 或 ntlm 的值进行的渗透测试 PTT(pass the ticket) #利用的票据凭证 TGT 进行的渗透测试 PTK(pass the ke ...

curve25519-dalek中field域内和scalar域内的运算性能对比

curve25519-dalek中field域内和scalar域内的运算性能对比相关推荐

最新文章

热门文章