rk3588/rk3588s benchmark sbc bench 性能测试 OpenGL跑分
glmark2 直接运行显示没有驱动 分数113 glmark2-es2运行 有arm驱动支持opengl es 3.2
跑分最多740（打开多个窗口运行一定的负载，最好进行散热）网络文章显示有rk3588
在默认驱动下跑到900左右。
关于opengles跑分，目前可知的rk3588分数一般是raspberry pi 4的2倍多，通过使用开源驱动和wayland等各种优化，最高可以达到4000到5000分，游戏性能是raspberry pi 4的4倍以上
关于opengles如何刷到最高，请在youtub搜索 rk3588 相关keyword
国外评测数据汇总
orangepi@orangepi5:~$ glmark2-es2
arm_release_ver of this libmali is 'g6p0-01eac0', rk_so_ver is '5'.
=======================================================
glmark2 2021.02
=======================================================
OpenGL Information
GL_VENDOR: ARM
GL_RENDERER: Mali-LODX
GL_VERSION: OpenGL ES 3.2 v1.g6p0-01eac0.efb75e2978d783a80fe78be1bfb0efc1
=======================================================
[build] use-vbo=false: FPS: 959 FrameTime: 1.043 ms
[build] use-vbo=true: FPS: 805 FrameTime: 1.242 ms
[texture] texture-filter=nearest: FPS: 856 FrameTime: 1.168 ms
[texture] texture-filter=linear: FPS: 960 FrameTime: 1.042 ms
[texture] texture-filter=mipmap: FPS: 871 FrameTime: 1.148 ms
[shading] shading=gouraud: FPS: 764 FrameTime: 1.309 ms
[shading] shading=blinn-phong-inf: FPS: 778 FrameTime: 1.285 ms
[shading] shading=phong: FPS: 758 FrameTime: 1.319 ms
[shading] shading=cel: FPS: 773 FrameTime: 1.294 ms
[bump] bump-render=high-poly: FPS: 549 FrameTime: 1.821 ms
[bump] bump-render=normals: FPS: 1036 FrameTime: 0.965 ms
[bump] bump-render=height: FPS: 928 FrameTime: 1.078 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 976 FrameTime: 1.025 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 591 FrameTime: 1.692 ms
[pulsar] light=false:quads=5:texture=false: FPS: 886 FrameTime: 1.129 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 502 FrameTime: 1.992 ms
[desktop] effect=shadow:windows=4: FPS: 718 FrameTime: 1.393 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 218 FrameTime: 4.587 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 194 FrameTime: 5.155 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 322 FrameTime: 3.106 ms
[ideas] speed=duration: FPS: 487 FrameTime: 2.053 ms
[jellyfish] <default>: FPS: 749 FrameTime: 1.335 ms
[terrain] <default>: FPS: 187 FrameTime: 5.348 ms
[shadow] <default>: FPS: 688 FrameTime: 1.453 ms
[refract] <default>: FPS: 305 FrameTime: 3.279 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 810 FrameTime: 1.235 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 807 FrameTime: 1.239 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 839 FrameTime: 1.192 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 831 FrameTime: 1.203 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 876 FrameTime: 1.142 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 781 FrameTime: 1.280 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 764 FrameTime: 1.309 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 802 FrameTime: 1.247 ms
=======================================================
glmark2 Score: 708
=======================================================
sbc bench跑分
sbc-bench v0.7.3 Orange Pi 5 (Sun, 01 Jan 2023 12:53:45 +0800)
Distributor ID: Debian
Description: Debian GNU/Linux 11 (bullseye)
Release: 11
Codename: bullseye
Architecture: arm64
/usr/bin/gcc (Debian 10.2.1-6) 10.2.1 20210110
Uptime: 12:53:45 up 44 min, 4 users, load average: 0.15, 0.13, 0.40
Linux 5.10.110-rockchip-rk3588 (orangepi5) 01/01/2023 _aarch64_ (8 CPU)
Device tps kB_read/s kB_wrtn/s kB_dscd/s kB_read kB_wrtn kB_dscd
mmcblk1 5.20 263.80 2.46 0.00 712017 6652 0
mtdblock0 0.01 0.12 0.00 0.00 336 0 0
nvme0n1 0.11 3.20 0.00 0.00 8648 0 0
zram0 0.21 0.83 0.00 0.00 2248 4 0
zram1 0.36 0.42 11.63 0.00 1128 31400 0
total used free shared buff/cache available
Mem: 7.5Gi 911Mi 5.8Gi 110Mi 855Mi 6.4Gi
Swap: 3.8Gi 0B 3.8Gi
Filename Type Size Used Priority
/dev/zram0 partition 3937088 0 5
##########################################################################
Checking cpufreq OPP for cpu0-cpu3:
Cpufreq OPP: 1800 Measured: 1836.743/1837.110/1836.906
Cpufreq OPP: 1608 Measured: 1620.401/1619.925/1620.560
Cpufreq OPP: 1416 Measured: 1425.485/1424.994/1425.454
Cpufreq OPP: 1200 Measured: 1199.891/1199.701/1200.245
Cpufreq OPP: 1008 Measured: 1038.560/1039.146/1039.146
Cpufreq OPP: 816 Measured: 828.895/829.098/828.996
Cpufreq OPP: 600 Measured: 590.927/591.133/591.159
Cpufreq OPP: 408 Measured: 393.373/393.203/393.328
Checking cpufreq OPP for cpu4-cpu7:
Cpufreq OPP: 2400 Measured: 2332.054/2331.686/2331.475
Cpufreq OPP: 2352 Measured: 2331.475/2331.370/2331.370
Cpufreq OPP: 2304 Measured: 2331.159/2331.054/2330.896
Cpufreq OPP: 2256 Measured: 2330.791/2330.844/2330.791
Cpufreq OPP: 2208 Measured: 2186.029/2185.844/2185.937
Cpufreq OPP: 2016 Measured: 2013.802/2013.606/2013.851
Cpufreq OPP: 1800 Measured: 1810.824/1810.824/1810.785
Cpufreq OPP: 1608 Measured: 1576.282/1576.094/1576.094
Cpufreq OPP: 1416 Measured: 1391.639/1391.610/1391.610
Cpufreq OPP: 1200 Measured: 1222.522/1222.692/1222.409
Cpufreq OPP: 1008 Measured: 1025.422/1025.347/1025.347
Cpufreq OPP: 816 Measured: 821.831/821.911/821.871
Cpufreq OPP: 600 Measured: 592.763/592.789/592.854
Cpufreq OPP: 408 Measured: 394.806/394.851/394.914
##########################################################################
7-Zip (a) [64] 16.02 : Copyright (c) 1999-2016 Igor Pavlov : 2016-05-21
p7zip Version 16.02 (locale=en_US.UTF-8,Utf16=on,HugeFiles=on,64 bits,8 CPUs LE)
Compression: 13361,12543,11561
Decompression: 15968,14649,14785
Total: 14664,13596,13173
==========================================================================
== Memory bandwidth tests ==
== ==
== Note 1: 1MB = 1000000 bytes ==
== Note 2: Results for 'copy' tests show how many bytes can be ==
== copied per second (adding together read and writen ==
== bytes would have provided twice higher numbers) ==
== Note 3: 2-pass copy means that we are using a small temporary buffer ==
== to first fetch data into it, and only then write it to the ==
== destination (source -> L1 cache, L1 cache -> destination) ==
== Note 4: If sample standard deviation exceeds 0.1%, it is shown in ==
== brackets ==
==========================================================================
C copy backwards : 9529.9 MB/s (20.1%)
C copy backwards (32 byte blocks) : 10513.4 MB/s (0.6%)
C copy backwards (64 byte blocks) : 10322.3 MB/s
C copy : 10792.7 MB/s (0.2%)
C copy prefetched (32 bytes step) : 10935.3 MB/s
C copy prefetched (64 bytes step) : 10970.8 MB/s (0.2%)
C 2-pass copy : 5026.7 MB/s (0.2%)
C 2-pass copy prefetched (32 bytes step) : 7664.6 MB/s
C 2-pass copy prefetched (64 bytes step) : 8120.3 MB/s
C fill : 28703.3 MB/s (0.2%)
C fill (shuffle within 16 byte blocks) : 28689.5 MB/s (0.2%)
C fill (shuffle within 32 byte blocks) : 28713.4 MB/s (0.3%)
C fill (shuffle within 64 byte blocks) : 28635.6 MB/s (0.5%)
---
standard memcpy : 11305.6 MB/s (0.3%)
standard memset : 27851.9 MB/s (0.4%)
##########################################################################
OpenSSL 1.1.1n, built on 15 Mar 2022
type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 16384 bytes
aes-128-cbc 165173.55k 486658.35k 949545.13k 1237372.59k 1359560.70k 1382372.69k
aes-128-cbc 646869.79k 1318039.30k 1685556.05k 1797646.68k 1842525.53k 1848377.34k
aes-192-cbc 157161.81k 432656.90k 769975.21k 958941.87k 1032353.11k 1038822.06k
aes-192-cbc 606593.90k 1153387.11k 1423940.52k 1483209.73k 1537425.41k 1541313.88k
aes-256-cbc 153394.23k 398090.79k 664333.82k 801074.52k 852028.07k 856517.29k
aes-256-cbc 605940.47k 1024541.63k 1234562.13k 1294262.27k 1319398.06k 1322090.50k
##########################################################################
System health while running OpenSSL benchmark:
Time big.LITTLE load %cpu %sys %usr %nice %io %irq Temp
13:06:06: 2400/1800MHz 1.00 6% 1% 4% 0% 0% 0% 53.6°C
13:06:16: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 51.8°C
13:06:26: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 54.5°C
13:06:36: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 53.6°C
13:06:46: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 51.8°C
13:06:56: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 50.8°C
13:07:06: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 54.5°C
13:07:16: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 53.6°C
13:07:26: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 50.8°C
13:07:36: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 51.8°C
13:07:46: 2400/1800MHz 1.00 12% 0% 12% 0% 0% 0% 53.6°C
完整的sbc bench地址
如何进行优化？
参考
https://github.com/ThomasKaiser/Knowledge/blob/master/articles/Quick_Preview_of_ROCK_5B.md#performance
已知的问题
并未发现使用nvme做系统盘能提高opengles跑分
使用debian性能比ubuntu好，debian下支持rk3588的4k60p硬件解码，也就是AV1解码，实际使用可以不用带风扇播放4k视频
本板 orange pi 5 仍未有m.2接口的wifi6 无线网卡，
如果能有硬件上对 mini pcie接口的支持（转接），可以实现对4g模组 ，5g模组，nbiot，ai加速硬件，2.5g网卡，fpga加速器。。。。。。的支持
更新 rk3588 7zip b 跑分
