在不同的risc-v硬件平台上运行pyperformance测试集并进行对比

为了测试u540和u740的性能差异，采用在两个平台上分别运行pyperformance的方法进行测试。

操作系统用的是ubuntu官方系统下载自：

Ubuntu 20.04.3 LTS (Focal Fossa)

https://cdimage.ubuntu.com/releases/20.04.3/release/ubuntu-20.04.3-preinstalled-server-riscv64+unmatched.img.xz

一、测试环境

操作系统：Ubuntu-20.04.3

内核版本：Linux-5.11.0-1017-generic

Python版本：3.8.10 (64-bit)

python_compiler: GCC 9.3.0

Pyperformance版本：1.0.2

安装命令：python3 -m pip install pyperformance

测试命令：python3 -m pyperformance run -o 540ubu.json

                  python3 -m pyperformance run -o 740ubu.json

二、测试结果对比

对比命令：

 python3 -m pyperf compare_to u540.json u740.json –-table > out.txt

+-------------------------+----------+------------------------+
| Benchmark               | u540     | u740                   |
+=========================+==========+========================+
| 2to3                    | 7.36 sec | 8.76 sec: 1.19x slower |
+-------------------------+----------+------------------------+
| chameleon               | 280 ms   | 369 ms: 1.32x slower   |
+-------------------------+----------+------------------------+
| chaos                   | 2.72 sec | 3.18 sec: 1.17x slower |
+-------------------------+----------+------------------------+
| crypto_pyaes            | 2.47 sec | 2.85 sec: 1.15x slower |
+-------------------------+----------+------------------------+
| deltablue               | 186 ms   | 222 ms: 1.19x slower   |
+-------------------------+----------+------------------------+
| django_template         | 1.62 sec | 1.86 sec: 1.15x slower |
+-------------------------+----------+------------------------+
| dulwich_log             | 1.48 sec | 1.63 sec: 1.10x slower |
+-------------------------+----------+------------------------+
| fannkuch                | 9.46 sec | 10.0 sec: 1.06x slower |
+-------------------------+----------+------------------------+
| float                   | 2.81 sec | 3.82 sec: 1.36x slower |
+-------------------------+----------+------------------------+
| go                      | 6.18 sec | 7.28 sec: 1.18x slower |
+-------------------------+----------+------------------------+
| hexiom                  | 230 ms   | 302 ms: 1.31x slower   |
+-------------------------+----------+------------------------+
| json_dumps              | 335 ms   | 397 ms: 1.18x slower   |
+-------------------------+----------+------------------------+
| json_loads              | 594 us   | 581 us: 1.02x faster   |
+-------------------------+----------+------------------------+
| logging_format          | 263 us   | 343 us: 1.30x slower   |
+-------------------------+----------+------------------------+
| logging_silent          | 4.78 us  | 6.65 us: 1.39x slower  |
+-------------------------+----------+------------------------+
| logging_simple          | 243 us   | 282 us: 1.16x slower   |
+-------------------------+----------+------------------------+
| mako                    | 450 ms   | 581 ms: 1.29x slower   |
+-------------------------+----------+------------------------+
| meteor_contest          | 1.85 sec | 2.15 sec: 1.16x slower |
+-------------------------+----------+------------------------+
| nbody                   | 2.59 sec | 2.96 sec: 1.14x slower |
+-------------------------+----------+------------------------+
| nqueens                 | 2.39 sec | 2.73 sec: 1.14x slower |
+-------------------------+----------+------------------------+
| pathlib                 | 517 ms   | 591 ms: 1.14x slower   |
+-------------------------+----------+------------------------+
| pickle                  | 217 us   | 200 us: 1.09x faster   |
+-------------------------+----------+------------------------+
| pickle_dict             | 483 us   | 400 us: 1.21x faster   |
+-------------------------+----------+------------------------+
| pickle_list             | 65.6 us  | 59.0 us: 1.11x faster  |
+-------------------------+----------+------------------------+
| pickle_pure_python      | 12.9 ms  | 17.3 ms: 1.35x slower  |
+-------------------------+----------+------------------------+
| pidigits                | 3.53 sec | 2.36 sec: 1.50x faster |
+-------------------------+----------+------------------------+
| pyflate                 | 16.6 sec | 19.9 sec: 1.20x slower |
+-------------------------+----------+------------------------+
| python_startup          | 121 ms   | 133 ms: 1.10x slower   |
+-------------------------+----------+------------------------+
| python_startup_no_site  | 79.0 ms  | 86.7 ms: 1.10x slower  |
+-------------------------+----------+------------------------+
| raytrace                | 13.0 sec | 15.1 sec: 1.17x slower |
+-------------------------+----------+------------------------+
| regex_compile           | 3.97 sec | 4.90 sec: 1.23x slower |
+-------------------------+----------+------------------------+
| regex_dna               | 3.15 sec | 2.69 sec: 1.17x faster |
+-------------------------+----------+------------------------+
| regex_effbot            | 62.7 ms  | 55.6 ms: 1.13x faster  |
+-------------------------+----------+------------------------+
| regex_v8                | 449 ms   | 432 ms: 1.04x faster   |
+-------------------------+----------+------------------------+
| richards                | 1.80 sec | 2.10 sec: 1.17x slower |
+-------------------------+----------+------------------------+
| scimark_fft             | 7.24 sec | 8.18 sec: 1.13x slower |
+-------------------------+----------+------------------------+
| scimark_lu              | 3.98 sec | 4.96 sec: 1.25x slower |
+-------------------------+----------+------------------------+
| scimark_monte_carlo     | 2.52 sec | 2.94 sec: 1.17x slower |
+-------------------------+----------+------------------------+
| scimark_sor             | 4.72 sec | 5.39 sec: 1.14x slower |
+-------------------------+----------+------------------------+
| scimark_sparse_mat_mult | 101 ms   | 144 ms: 1.42x slower   |
+-------------------------+----------+------------------------+
| spectral_norm           | 3.18 sec | 3.57 sec: 1.12x slower |
+-------------------------+----------+------------------------+
| sqlalchemy_declarative  | 2.66 sec | 3.08 sec: 1.16x slower |
+-------------------------+----------+------------------------+
| sqlalchemy_imperative   | 366 ms   | 425 ms: 1.16x slower   |
+-------------------------+----------+------------------------+
| sqlite_synth            | 77.2 us  | 85.8 us: 1.11x slower  |
+-------------------------+----------+------------------------+
| sympy_expand            | 12.2 sec | 14.5 sec: 1.18x slower |
+-------------------------+----------+------------------------+
| sympy_integrate         | 459 ms   | 541 ms: 1.18x slower   |
+-------------------------+----------+------------------------+
| sympy_sum               | 3.55 sec | 4.07 sec: 1.15x slower |
+-------------------------+----------+------------------------+
| sympy_str               | 7.08 sec | 8.45 sec: 1.19x slower |
+-------------------------+----------+------------------------+
| telco                   | 189 ms   | 221 ms: 1.17x slower   |
+-------------------------+----------+------------------------+
| unpack_sequence         | 1.05 us  | 1.17 us: 1.12x slower  |
+-------------------------+----------+------------------------+
| unpickle                | 329 us   | 308 us: 1.07x faster   |
+-------------------------+----------+------------------------+
| unpickle_list           | 74.9 us  | 76.8 us: 1.03x slower  |
+-------------------------+----------+------------------------+
| unpickle_pure_python    | 9.11 ms  | 11.4 ms: 1.25x slower  |
+-------------------------+----------+------------------------+
| xml_etree_parse         | 3.41 sec | 3.79 sec: 1.11x slower |
+-------------------------+----------+------------------------+
| xml_etree_iterparse     | 2.48 sec | 3.04 sec: 1.23x slower |
+-------------------------+----------+------------------------+
| xml_etree_generate      | 2.58 sec | 3.12 sec: 1.21x slower |
+-------------------------+----------+------------------------+
| xml_etree_process       | 2.15 sec | 2.58 sec: 1.20x slower |
+-------------------------+----------+------------------------+
| Geometric mean          | (ref)    | 1.13x slower           |
+-------------------------+----------+------------------------+
Ignored benchmarks (1) of u740.json: tornado_http

可以看到u740的性能看起来并没有太大的优势。

在很多项目上都是slower，只有个别项目是faster。

参考：

Usage — Python Performance Benchmark Suite 1.0.3 documentation