Orange Pi 5 Ultra: mini placa, máximo rendimiento - Página 11 de 15

Tabla de contenidos

Benchmarks

Y como no, también hemos utilizado el famoso SBC de Thomas Kaiser que se descarga a instala así …

root@orangepi5ultra:/home/orangepi# wget https://raw.githubusercontent.com/ThomasKaiser/sbc-bench/master/sbc-bench.sh

--2025-05-28 17:02:19--  https://raw.githubusercontent.com/ThomasKaiser/sbc-bench/master/sbc-bench.sh

Resolviendo raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.109.133, 185.199.111.133, 185.199.108.133, ...

Conectando con raw.githubusercontent.com (raw.githubusercontent.com)[185.199.109.133]:443... conectado.

Petición HTTP enviada, esperando respuesta... 200 OK

Longitud: 451249 (441K) [text/plain]

Guardando como: ‘sbc-bench.sh’



sbc-bench.sh                        100%[=================================================================&gt;] 440,67K  --.-KB/s    en 0,02s   



2025-05-28 17:02:19 (18,8 MB/s) - ‘sbc-bench.sh’ guardado [451249/451249]

root@orangepi5ultra:/home/orangepi# wget https://raw.githubusercontent.com/ThomasKaiser/sbc-bench/master/sbc-bench.sh

--2025-05-28 17:02:19-- https://raw.githubusercontent.com/ThomasKaiser/sbc-bench/master/sbc-bench.sh

Resolviendo raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.109.133, 185.199.111.133, 185.199.108.133, ...

Conectando con raw.githubusercontent.com (raw.githubusercontent.com)[185.199.109.133]:443... conectado.

Petición HTTP enviada, esperando respuesta... 200 OK

Longitud: 451249 (441K) [text/plain]

Guardando como: ‘sbc-bench.sh’

sbc-bench.sh 100%[=================================================================>] 440,67K --.-KB/s en 0,02s

2025-05-28 17:02:19 (18,8 MB/s) - ‘sbc-bench.sh’ guardado [451249/451249]

y cuyos resultados os dejamos aquí.

root@orangepi5ultra:/home/orangepi# sudo /bin/bash ./sbc-bench.sh -r

Starting to examine hardware/software for review purposes...



Average load and/or CPU utilization too high (too much background activity). Waiting...



Too busy for benchmarking: 17:27:17 up  1:11,  6 users,  load average: 1.42, 3.19, 2.74,  cpu: 0%

Too busy for benchmarking: 17:27:22 up  1:11,  6 users,  load average: 1.39, 3.15, 2.73,  cpu: 0%

Too busy for benchmarking: 17:27:27 up  1:11,  6 users,  load average: 1.36, 3.12, 2.72,  cpu: 0%

Too busy for benchmarking: 17:27:32 up  1:12,  6 users,  load average: 1.33, 3.08, 2.71,  cpu: 0%

Too busy for benchmarking: 17:27:37 up  1:12,  6 users,  load average: 1.30, 3.05, 2.71,  cpu: 0%

Too busy for benchmarking: 17:27:42 up  1:12,  6 users,  load average: 1.28, 3.01, 2.70,  cpu: 0%



sbc-bench v0.9.71



Installing needed tools: apt-get -f -qq -y install dmidecode lshw links stress-ng. Updating (can't build cpuminer) Done.

Checking cpufreq OPP. Done.

Executing tinymembench. Done.

Executing RAM latency tester. Done.

Executing OpenSSL benchmark. Done.

Executing 7-zip benchmark. Done.

Throttling test: heating up the device, 5 more minutes to wait. Done.

Checking cpufreq OPP again. Done (15 minutes elapsed).



Results validation:



  * Advertised vs. measured max CPU clockspeed: -3.7% before, -4.3% after -&gt; https://tinyurl.com/32w9rr94

  * No swapping

  * Background activity (%system) OK

  * No throttling





Full results uploaded to https://0x0.st/8xFU.bin







# RK3588 OPi 5 Ultra



Tested with sbc-bench v0.9.71 on Wed, 28 May 2025 17:43:07 +0200. Full info: [https://0x0.st/8xFU.bin](http://0x0.st/8xFU.bin)



### General information:



    Information courtesy of cpufetch:

    

    SoC:                 Rockchip RK3588

    Technology:          8nm

    CPU 1:

      Microarchitecture: Cortex-A55

      Max Frequency:     1.800 GHz

      Cores:             4 cores

      Features:          NEON,SHA1,SHA2,AES,CRC32

    CPU 2:

      Microarchitecture: Cortex-A76

      Max Frequency:     2.352 GHz

      Cores:             4 cores

      Features:          NEON,SHA1,SHA2,AES,CRC32

    

The CPU features 3 clusters consisting of 2 different core types:



    Rockchip RK3588 (35881000 / 35 88 12 fe 21 4e  5a 58 59 38 00 00 00 00), Kernel: aarch64, Userland: arm64

    

    CPU sysfs topology (clusters, cpufreq members, clockspeeds)

                     cpufreq   min    max

     CPU    cluster  policy   speed  speed   core type

      0        0        0      408    1800   Cortex-A55 / r2p0

      1        0        0      408    1800   Cortex-A55 / r2p0

      2        0        0      408    1800   Cortex-A55 / r2p0

      3        0        0      408    1800   Cortex-A55 / r2p0

      4        0        4      408    2352   Cortex-A76 / r4p0

      5        0        4      408    2352   Cortex-A76 / r4p0

      6        0        6      408    2352   Cortex-A76 / r4p0

      7        0        6      408    2352   Cortex-A76 / r4p0



7934 KB available RAM



### Governors/policies (performance vs. idle consumption):



Original governor settings:



    cpufreq-policy0: ondemand / 1008 MHz (interactive conservative ondemand userspace powersave performance schedutil / 408 600 816 1008 1200 1416 1608 1800)

    cpufreq-policy4: ondemand / 1200 MHz (interactive conservative ondemand userspace powersave performance schedutil / 408 600 816 1008 1200 1416 1608 1800 2016 2208 2352)

    cpufreq-policy6: ondemand / 600 MHz (interactive conservative ondemand userspace powersave performance schedutil / 408 600 816 1008 1200 1416 1608 1800 2016 2208 2352)

    dmc: dmc_ondemand / 2400 MHz (rknpu_ondemand dmc_ondemand userspace powersave performance simple_ondemand / 534 1320 1968 2400)

    fb000000.gpu: simple_ondemand / 300 MHz (rknpu_ondemand dmc_ondemand userspace powersave performance simple_ondemand / 300 400 500 600 700 800 900 1000)

    fdab0000.npu: rknpu_ondemand / 300 MHz (rknpu_ondemand dmc_ondemand userspace powersave performance simple_ondemand / 300 400 500 600 700 800 900 1000)



Tuned governor settings:



    cpufreq-policy0: performance / 1800 MHz

    cpufreq-policy4: performance / 2352 MHz

    cpufreq-policy6: performance / 2352 MHz

    dmc: performance / 2400 MHz

    fb000000.gpu: performance / 1000 MHz

    fdab0000.npu: performance / 1000 MHz



Status of performance related policies found below /sys:



    /sys/devices/platform/fb000000.gpu/power_policy: [coarse_demand] always_on

    /sys/module/pcie_aspm/parameters/policy: default [performance] powersave powersupersave



### Clockspeeds (idle vs. heated up):



Before at 56.4°C:



    cpu0-cpu3 (Cortex-A55): OPP: 1800, Measured: 1781      (-1.1%)

    cpu4-cpu5 (Cortex-A76): OPP: 2352, Measured: 2269      (-3.5%)

    cpu6-cpu7 (Cortex-A76): OPP: 2352, Measured: 2265      (-3.7%)



After at 74.8°C:



    cpu0-cpu3 (Cortex-A55): OPP: 1800, Measured: 1770      (-1.7%)

    cpu4-cpu5 (Cortex-A76): OPP: 2352, Measured: 2254      (-4.2%)

    cpu6-cpu7 (Cortex-A76): OPP: 2352, Measured: 2251      (-4.3%)



### Performance baseline



  * cpu0 (Cortex-A55): memcpy: 6539.3 MB/s, memchr: 2738.9 MB/s, memset: 21632.3 MB/s

  * cpu4 (Cortex-A76): memcpy: 12178.7 MB/s, memchr: 15863.9 MB/s, memset: 27540.4 MB/s

  * cpu6 (Cortex-A76): memcpy: 12151.2 MB/s, memchr: 15837.2 MB/s, memset: 27432.3 MB/s

  * cpu0 (Cortex-A55) 16M latency: 140.1 140.7 137.8 140.3 137.4 141.4 218.5 383.6 

  * cpu4 (Cortex-A76) 16M latency: 141.3 130.3 139.6 130.3 139.4 133.7 128.6 127.5 

  * cpu6 (Cortex-A76) 16M latency: 139.7 129.6 140.4 129.9 139.1 129.0 126.1 127.2 

  * cpu0 (Cortex-A55) 128M latency: 162.0 163.2 161.7 163.0 161.4 163.1 235.1 407.3 

  * cpu4 (Cortex-A76) 128M latency: 153.8 152.3 153.3 152.1 153.5 152.2 152.9 159.4 

  * cpu6 (Cortex-A76) 128M latency: 155.6 155.0 155.4 155.0 155.6 154.6 154.7 162.4 

  * 7-zip MIPS (3 consecutive runs): 15590, 15688, 15678 (15650 avg), single-threaded: 2955

  * `aes-256-cbc     145614.48k   382933.21k   645486.08k   778974.89k   829057.71k   833049.94k (Cortex-A55)`

  * `aes-256-cbc     574930.81k   996768.64k  1204175.79k  1265561.94k  1289377.11k  1291709.10k (Cortex-A76)`

  * `aes-256-cbc     574853.17k   993951.47k  1201168.04k  1263089.32k  1286845.78k  1289136.81k (Cortex-A76)`



### PCIe and storage devices:



  * Realtek RTL8125 2.5GbE: Speed 5GT/s (ok), Width x1 (ok), driver in use: r8169, ASPM Disabled

  * 238.5GB "BR 256GB" SSD as /dev/nvme0: Speed 8GT/s (ok), Width x4 (ok), 0% worn out, drive temp: 40°C, ASPM Disabled

  * 58.3GB "ISOCOM" HS400 Enhanced strobe eMMC 5.1 card as /dev/mmcblk0: date 11/2024, manfid/oemid: 0x000004/0x0101, hw/fw rev: 0x0/0x1000000000000000

  * 14.5GB "SanDisk SD16G" UHS SDR50 SD card as /dev/mmcblk1: date 05/2014, manfid/oemid: 0x000003/0x5344, hw/fw rev: 0x3/0x0

  * Xmc XM25QU128C 16MB SPI NOR flash, drivers in use: spi-nor/rockchip-sfc



### Swap configuration:



  * /dev/zram0: 3.9G (0K used, lzo-rle, 8 streams, 4K data, 74B compressed, 12K total)



### Software versions:



  * Ubuntu 22.04.5 LTS (jammy)

  * Build scripts: https://github.com/orangepi-xunlong/orangepi-build, 1.0.0, Orange Pi 5 Ultra, rockchip-rk3588, rockchip-rk3588

  * Compiler: /usr/bin/gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 / aarch64-linux-gnu

  * OpenSSL 3.0.2, built on 15 Mar 2022 (Library: OpenSSL 3.0.2 15 Mar 2022)    



### Kernel info:



  * `/proc/cmdline: root=UUID=69cb8be7-f4f7-4515-a53b-ca446b668190 rootwait rootfstype=ext4 splash plymouth.ignore-serial-consoles console=ttyS2,1500000 console=tty1 consoleblank=0 loglevel=1 ubootpart=6b53f235-db5e-c54f-b2d2-9edf86801bd8 usb-storage.quirks= cma=128M  cgroup_enable=cpuset cgroup_memory=1 cgroup_enable=memory swapaccount=1`

  * Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl

  * Vulnerability Spectre v1:        Mitigation; __user pointer sanitization

  * Vulnerability Spectre v2:        Vulnerable: Unprivileged eBPF enabled

  * Kernel 6.1.43-rockchip-rk3588 / CONFIG_HZ=300



All known settings adjusted for performance. Device now ready for benchmarking.

Once finished stop with [ctrl]-[c] to get info about throttling, frequency cap

and too high background activity all potentially invalidating benchmark scores.

All changes with storage and PCIe devices as well as suspicious dmesg contents

will be reported too.



Time       cpu0/cpu4/cpu6    load %cpu %sys %usr %nice %io %irq   Temp

17:43:09: 1800/2352/2352MHz  6.95  11%   0%   9%   0%   1%   0%  66.5°C  

17:44:09: 1800/2352/2352MHz  3.18   0%   0%   0%   0%   0%   0%  60.1°C  

17:45:09: 1800/2352/2352MHz  1.80   0%   0%   0%   0%   0%   0%  58.2°C  

17:46:09: 1800/2352/2352MHz  1.36   0%   0%   0%   0%   0%   0%  58.2°C  

17:47:09: 1800/2352/2352MHz  1.13   0%   0%   0%   0%   0%   0%  58.2°C  

17:48:09: 1800/2352/2352MHz  1.05   0%   0%   0%   0%   0%   0%  57.3°C  

17:49:09: 1800/2352/2352MHz  1.02   0%   0%   0%   0%   0%   0%  57.3°C  

17:50:09: 1800/2352/2352MHz  1.11   0%   0%   0%   0%   0%   0%  57.3°C  

17:51:09: 1800/2352/2352MHz  1.04   0%   0%   0%   0%   0%   0%  57.3°C  

17:52:09: 1800/2352/2352MHz  1.01   0%   0%   0%   0%   0%   0%  56.4°C  

17:53:09: 1800/2352/2352MHz  1.07   0%   0%   0%   0%   0%   0%  57.3°C  

^C



Cleaning up. Done.

Checking cpufreq OPP again. Done.



Clockspeeds now at 58.2°C:



    cpu0-cpu3 (Cortex-A55): OPP: 1800, Measured: 1781      (-1.1%)

    cpu4-cpu5 (Cortex-A76): OPP: 2352, Measured: 2268      (-3.6%)

    cpu6-cpu7 (Cortex-A76): OPP: 2352, Measured: 2264      (-3.7%)



ATTENTION: some noise in kernel ring buffer since start of monitoring:



[ 5279.478653] [dhd] [wlan0] wl_run_escan : LEGACY_SCAN sync ID: 51, bssidx: 0

[ 5400.476177] [dhd] [wlan0] wl_run_escan : LEGACY_SCAN sync ID: 52, bssidx: 0

[ 5521.479663] [dhd] [wlan0] wl_run_escan : LEGACY_SCAN sync ID: 53, bssidx: 0

[ 5642.481037] [dhd] [wlan0] wl_run_escan : LEGACY_SCAN sync ID: 54, bssidx: 0

[ 5763.488804] [dhd] [wlan0] wl_run_escan : LEGACY_SCAN sync ID: 55, bssidx: 0

[ 5884.495641] [dhd] [wlan0] wl_run_escan : LEGACY_SCAN sync ID: 56, bssidx: 0



Results validation:



  * Advertised vs. measured max CPU clockspeed: -3.7% before, -4.3% after -&gt; https://tinyurl.com/32w9rr94

  * No swapping

  * Background activity (%system) OK

  * No throttling

  * ondemand cpufreq governor used by distro but io_is_busy not set to 1 on all cores -&gt; http://tinyurl.com/44pbmw79

Nos quedamos con que la velocidad máxima en ningún momento alcanzó los 75ºC (y el ruido del ventilador al máximo sin molestar lo más mínimo) y sin necesidad de thermal throttling para mantener funcionando sin problemas el procesador.

7022

4 Comments

Admin

Pharizna

1 año antes

Resulta cuando menos curioso que haya sido publicar nuestra revisión para que ahora el RPi5 Ultra … se niegue a pararse.

Lo peor del caso es que parece no somos los únicos a los que no funciona el HALT en Debían 12 desde la terminal.

En cuanto lo tengamos resuelto os lo contamos en la revisión.

En respuesta a Pharizna

RESUELTO!

En lugar de utilizar halt … con shutdown funciona perfectamente. Una cosa más que hemos aprendido

Una cosa «negativa» que se nos ha olvidado incluir en la revisión es la FALTA DE PATITAS en la caja metálica para evitar «roces» contra el mueble en el que lo colocas y que no se te enfade la parienta

11 meses antes

Si todo va de acuerdo de lo previsto en unos días empezaremos a revisar el nuevo Orange Pi 6 Plus. Además de nuestras típicas pruebas intentaremos probar el desarrollo de aplicaciones de IA aprovechando la potencia del nuevo SBC.