A first look at the Raspberry Pi Development Kit for CM5

Raspberry Pi has just launched the Compute Module 5 (CM5) and the company sent me a “Raspberry Pi Development Kit for CM5” for review and to play around with the new Broadcom BCM2712 system-on-module succeeding the Raspberry Pi Compute Module 4 (CM4) introduced in 2020.

Today, I’ll have a first look at the development kit checking out the hardware, connecting accessories, booting it up to Raspberry Pi OS, and collecting some basic system information. I’ll do a more in-depth review in a few weeks.

Unboxing of the Raspberry Pi Development Kit for CM5

I received the kit in a package reading “Raspberry Pi Development Kit” and “For Raspberry Pi CM5”. That’s a good start…

The bottom side lists the package’s content which I will check out shortly, and a link to the documentation.

Let’s see what we have with the Raspberry Pi Development Kit: a Raspberry Pi IO Case, a Raspberry Pi Compute Module 5, a Raspberry Pi Cooler (not designed for use with the IO case lid), a 27W USB-C power adapter, two HDMI cables, a USB-A to USB-C cable, and a Raspberry Pi Antenna Kit. I can’t see the Raspberry Pi CM5 IO board, but I hope it will be inside the IO case package…

Let’s check out the Raspberry Pi CM5 first and compare it to the Raspberry Pi CM4 (Lite) since I don’t own a CM4 module.

There’s one more component on the top inside: the Raspberry Ri RP1 peripheral controller also found in the Raspberry Pi 5 SBC. The power chip is larger having replaced the Dialog DA9090 by a Renesas DA9091 PMIC with some merger and acquisition activity happening in the meantime. If we zoom in on the configuration resistors on the top right side of the CM5 module, we can see I have a CM5104032 model with 4GB RAM, 32GB flash, and a wireless module. We’ll also notice some unannounced options, so it looks like a cheaper Raspberry Pi Compute Module 5 module with 1GB RAM and 8GB eMMC flash may eventually become available, as well as higher-end ones with up to 128GB eMMC flash.

The CM4 modules have their eMMC flash placed on top, but due to the RP1 chip on the CM5, the eMMC flash had to be moved to the bottom side of the module. We still have the two board-to-board (B2B) connectors in the same position and with the same layout.

The Raspberry Pi Cooler comes with a screw set, and we can see an opening on the left for the antennas.

The bottom side has three thermal pads to make sure the Raspberry Pi CM5 keeps cool under heavy loads.

I’m glad to finally confirm my IO case did come with a Raspberry Pi CM5 IO board. More of the ports can be found on the rear panel with two HDMI ports, Power and Status LEDs, a gigabit Ethernet RJ45 jack, two USB 3.0 ports, a USB-C port for power, a microSD card for CM5 Lite only, an SSD activity LED, and an opening for the WIFi and Bluetooth antenna.

The front panel has some ventilation holes, two openings for MIPI camera/display cables, and a power button.

Raspberry Pi Development Kit assembly and hardware setup

Let’s open the case by loosening the four screws (two on each side) holding the two parts together. A fan is connected to the lid for cooling. Let’s remove four more screws and disconnect the fan cable to take out the Raspberry Pi CM5 IO board.

The board comes with a CR2032 battery holder for the RTC battery, a 40-pin Raspberry Pi GPIO header with mounting holes for HAT expansion boards, a 4-pin configuration header, a 4-pin PoE header, the board-to-board connectors for the CM5 module, and M.2 M-key socket for 2230, 2242, 2260, or 2280 NVMe SSD, and a 14-pin “Jumper” header to configure boot, EEPROM write-protect, USB OTG, etc… as well as connect a power push button.

There aren’t any components on the bottom side.

Let’s now start the assembly by adding the SMA connector to the IO case.

We can now put the Compute Module 5 IO board back inside the IO case, insert the Raspberry Pi CM5 module, connect the WiFi antenna to the module, and add an official 256GB Raspberry Pi NVMe SSD. Ideally, I would have added a CR2032 battery, but I don’t have any.

I also connected a Raspberry Pi Camera Module 3 to the CAM/DISP 0 connector passing the cable through the opening in the IO case. I planned to connect a Raspberry Pi Touch Display 2 too, but I skipped it for now because it requires 5V from the 40-pin Raspberry Pi header and I want to first test the Raspberry Pi Developer Kit with the IO case fully closed. However, it will work fine if we keep the IO case open since we have access to the 5V/GND on the GPIO header that way.

I did not use the Raspberry Pi cooler for the same reason, since Raspberry Pi mentions it’s not designed to be used with the IO Case lid. That means we’ll just use the PWM fan for cooling.

First boot to Raspberry Pi OS and system information

Let’s close the lid add the external WiFi antenna, and try to boot the system up after connecting an HDMI monitor and RF dongles for a keyboard and a mouse.

The system was up and running even though, I had not flashed Raspberry Pi OS to the eMMC flash. The explanation is simple: the Development Kit simply booted from the 256GB SSD preloaded with Raspberry Pi OS.

pi@raspberrypi:~ $ df -h
Filesystem      Size  Used Avail Use% Mounted on
udev            1.9G     0  1.9G   0% /dev
tmpfs           405M  6.4M  398M   2% /run
/dev/nvme0n1p2  235G  5.1G  218G   3% /
tmpfs           2.0G   34M  2.0G   2% /dev/shm
tmpfs           5.0M   48K  5.0M   1% /run/lock
/dev/nvme0n1p1  510M   76M  435M  15% /boot/firmware
tmpfs           405M  208K  404M   1% /run/user/1000

pi@raspberrypi:~ $ df -h

Filesystem Size Used Avail Use% Mounted on

udev 1.9G 0 1.9G 0% /dev

tmpfs 405M 6.4M 398M 2% /run

/dev/nvme0n1p2 235G 5.1G 218G 3% /

tmpfs 2.0G 34M 2.0G 2% /dev/shm

tmpfs 5.0M 48K 5.0M 1% /run/lock

/dev/nvme0n1p1 510M 76M 435M 15% /boot/firmware

tmpfs 405M 208K 404M 1% /run/user/1000

But inxi confirms we also have a 32GB eMMC flash (Samsung BJTD4R) and other system information:

pi@raspberrypi:~ $ sudo inxi -Fc0
System:
  Host: raspberrypi Kernel: 6.6.51+rpt-rpi-2712 arch: aarch64 bits: 64
    Console: pty pts/2 Distro: Debian GNU/Linux 12 (bookworm)
Machine:
  Type: ARM System: Raspberry Pi Compute Module 5 Rev 1.0 details: N/A
    rev: c04180 serial: 2332d973fae3ffd8
CPU:
  Info: quad core model: N/A variant: cortex-a76 bits: 64 type: MCP cache:
    L2: 2 MiB
  Speed (MHz): avg: 2400 min/max: 1500/2400 cores: 1: 2400 2: 2400 3: 2400
    4: 2400
Graphics:
  Device-1: bcm2712-hdmi0 driver: vc4_hdmi v: N/A
  Device-2: bcm2712-hdmi1 driver: vc4_hdmi v: N/A
  Display: wayland server: X.org v: 1.21.1.7 with: Xwayland v: 22.1.9
    compositor: LabWC driver:
    gpu: vc4-drm,vc4_crtc,vc4_dpi,vc4_dsi,vc4_firmware_kms,vc4_hdmi,vc4_hvs,vc4_txp,vc4_v3d,vc4_vec
    tty: 80x24 resolution: 1920x1080
  API: EGL/GBM Message: No known Wayland EGL/GBM data sources.
Audio:
  Device-1: bcm2712-hdmi0 driver: vc4_hdmi
  Device-2: bcm2712-hdmi1 driver: vc4_hdmi
  API: ALSA v: k6.6.51+rpt-rpi-2712 status: kernel-api
Network:
  Device-1: Raspberry Pi RP1 PCIe 2.0 South Bridge driver: rp1
  IF: wlan0 state: up mac: 2c:cf:67:b2:8d:ea
  IF-ID-1: eth0 state: down mac: 2c:cf:67:b2:8d:e8
Bluetooth:
  Device-1: bcm7271-uart driver: bcm7271_uart
  Report: hciconfig ID: hci0 state: up address: 2C:CF:67:B2:8D:EB bt-v: 3.0
  Device-2: bcm7271-uart driver: N/A
Drives:
  Local Storage: total: 267.6 GiB used: 5.17 GiB (1.9%)
  ID-1: /dev/mmcblk0 type: Removable vendor: Samsung model: BJTD4R
    size: 29.12 GiB
  ID-2: /dev/nvme0n1 vendor: Samsung model: MZ9LQ256HBJD-00BVL
    size: 238.47 GiB
Partition:
  ID-1: / size: 234.2 GiB used: 5.1 GiB (2.2%) fs: ext4 dev: /dev/nvme0n1p2
Swap:
  ID-1: swap-1 type: file size: 200 MiB used: 0 KiB (0.0%) file: /var/swap
Sensors:
  System Temperatures: cpu: 68.8 C mobo: N/A
  Fan Speeds (RPM): cpu: 3516
Info:
  Processes: 214 Uptime: 4h 54m Memory: 3.95 GiB used: 1.33 GiB (33.5%)
  gpu: 8 MiB Init: systemd target: graphical (5) Shell: Sudo inxi: 3.3.26

pi@raspberrypi:~ $ sudo inxi -Fc0

System:

Host: raspberrypi Kernel: 6.6.51+rpt-rpi-2712 arch: aarch64 bits: 64

Console: pty pts/2 Distro: Debian GNU/Linux 12 (bookworm)

Machine:

Type: ARM System: Raspberry Pi Compute Module 5 Rev 1.0 details: N/A

rev: c04180 serial: 2332d973fae3ffd8

CPU:

Info: quad core model: N/A variant: cortex-a76 bits: 64 type: MCP cache:

L2: 2 MiB

Speed (MHz): avg: 2400 min/max: 1500/2400 cores: 1: 2400 2: 2400 3: 2400

4: 2400

Graphics:

Device-1: bcm2712-hdmi0 driver: vc4_hdmi v: N/A

Device-2: bcm2712-hdmi1 driver: vc4_hdmi v: N/A

Display: wayland server: X.org v: 1.21.1.7 with: Xwayland v: 22.1.9

compositor: LabWC driver:

gpu: vc4-drm,vc4_crtc,vc4_dpi,vc4_dsi,vc4_firmware_kms,vc4_hdmi,vc4_hvs,vc4_txp,vc4_v3d,vc4_vec

tty: 80x24 resolution: 1920x1080

API: EGL/GBM Message: No known Wayland EGL/GBM data sources.

Audio:

Device-1: bcm2712-hdmi0 driver: vc4_hdmi

Device-2: bcm2712-hdmi1 driver: vc4_hdmi

API: ALSA v: k6.6.51+rpt-rpi-2712 status: kernel-api

Network:

Device-1: Raspberry Pi RP1 PCIe 2.0 South Bridge driver: rp1

IF: wlan0 state: up mac: 2c:cf:67:b2:8d:ea

IF-ID-1: eth0 state: down mac: 2c:cf:67:b2:8d:e8

Bluetooth:

Device-1: bcm7271-uart driver: bcm7271_uart

Report: hciconfig ID: hci0 state: up address: 2C:CF:67:B2:8D:EB bt-v: 3.0

Device-2: bcm7271-uart driver: N/A

Drives:

Local Storage: total: 267.6 GiB used: 5.17 GiB (1.9%)

ID-1: /dev/mmcblk0 type: Removable vendor: Samsung model: BJTD4R

size: 29.12 GiB

ID-2: /dev/nvme0n1 vendor: Samsung model: MZ9LQ256HBJD-00BVL

size: 238.47 GiB

Partition:

ID-1: / size: 234.2 GiB used: 5.1 GiB (2.2%) fs: ext4 dev: /dev/nvme0n1p2

Swap:

ID-1: swap-1 type: file size: 200 MiB used: 0 KiB (0.0%) file: /var/swap

Sensors:

System Temperatures: cpu: 68.8 C mobo: N/A

Fan Speeds (RPM): cpu: 3516

Info:

Processes: 214 Uptime: 4h 54m Memory: 3.95 GiB used: 1.33 GiB (33.5%)

gpu: 8 MiB Init: systemd target: graphical (5) Shell: Sudo inxi: 3.3.26

The camera is not detected, I might have to double-check the connection, and I might have set some jumpers. I’ll check that in the second part of the review:

pi@raspberrypi:~ $ rpicam-hello --list
No cameras available!

1 2	pi@raspberrypi:~ $ rpicam-hello --list No cameras available!

The idle temperature is a little high at 68.8°C because there’s no heatsink and we only rely on the fan. Let’s run sbc-bench.sh script to see if thermal throttling does occur:

pi@raspberrypi:~ $ sudo ./sbc-bench.sh -r
Starting to examine hardware/software for review purposes...

sbc-bench v0.9.68

Installing needed tools: distro packages already installed, 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 (12 minutes elapsed).

Results validation:

  * Measured clockspeed not lower than advertised max CPU clockspeed
  * No swapping
  * Background activity (%system) OK
  * Too much other background activity: 0% avg, 6% max -> https://tinyurl.com/mr2wy5uv
  * Throttling / frequency capping (under-voltage) occured -> https://tinyurl.com/4ky59sys / https://tinyurl.com/3j2c66kd


Full results uploaded to https://0x0.st/XRKl.txt

# Raspberry Pi Compute Module 5 Rev 1.0

Tested with sbc-bench v0.9.68 on Wed, 27 Nov 2024 15:17:58 +0700. Full info: [https://0x0.st/XRKl.txt](http://0x0.st/XRKl.txt)

### General information:

    Information courtesy of cpufetch:
    
    SoC:                 Broadcom BCM2712
    Technology:          16nm
    Microarchitecture:   Cortex-A76
    Max Frequency:       2.400 GHz
    Cores:               4 cores
    Features:            NEON,SHA1,SHA2,AES,CRC32
    
    BCM2712, Kernel: aarch64, Userland: arm64
    
    CPU sysfs topology (clusters, cpufreq members, clockspeeds)
                     cpufreq   min    max
     CPU    cluster  policy   speed  speed   core type
      0        0        0     1500    2400   Cortex-A76 / r4p1
      1        0        0     1500    2400   Cortex-A76 / r4p1
      2        0        0     1500    2400   Cortex-A76 / r4p1
      3        0        0     1500    2400   Cortex-A76 / r4p1

4041 KB available RAM

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

Original governor settings:

    cpufreq-policy0: performance / 2400 MHz (conservative ondemand userspace powersave performance schedutil / 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400)

Tuned governor settings:

    cpufreq-policy0: performance / 2400 MHz

Status of performance related policies found below /sys:

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

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

Before at 73.2°C:

    cpu0 (Cortex-A76): OPP: 2400, ThreadX: 2400, Measured: 2399 

After at 84.8°C (throttled):

    cpu0 (Cortex-A76): OPP: 2400, ThreadX: 2400, Measured: 2403 

### Performance baseline

  * memcpy: 5304.3 MB/s, memchr: 13456.6 MB/s, memset: 11936.3 MB/s
  * 16M latency: 119.0 118.8 117.2 115.3 122.0 119.1 139.2 134.3 
  * 128M latency: 135.0 134.2 136.5 141.3 135.2 134.0 135.7 136.4 
  * 7-zip MIPS (3 consecutive runs): 9831, 9085, 8992 (9300 avg), single-threaded: 3191
  * `aes-256-cbc     540403.95k  1003620.69k  1256036.52k  1332896.09k  1365554.52k  1367932.93k`
  * `aes-256-cbc     540634.79k  1003600.96k  1256047.70k  1332894.72k  1365325.14k  1368255.15k`

### PCIe and storage devices:

  * Raspberry RP1 PCIe 2.0 South Bridge: Speed 5GT/s, Width x4, driver in use: rp1, ASPM Disabled
  * 238.5GB "SAMSUNG MZ9LQ256HBJD-00BVL" SSD as /dev/nvme0: Speed 8GT/s, Width x1 (downgraded), 0% worn out, drive temp: 40°C, ASPM Disabled
  * 29.1GB "Samsung BJTD4R" HS400 Enhanced strobe eMMC 5.1 card as /dev/mmcblk0: date 07/2024, manfid/oemid: 0x000015/0x0100, hw/fw rev: 0x0/0x0300000000000000

### Swap configuration:

  * /var/swap on /dev/nvme0n1p2: 200.0M (0K used)

### Software versions:

  * Debian GNU/Linux 12 (bookworm)
  * Build scripts: http://archive.raspberrypi.com/debian/ bookworm main
  * Compiler: /usr/bin/gcc (Debian 12.2.0-14) 12.2.0 / aarch64-linux-gnu
  * OpenSSL 3.0.15, built on 3 Sep 2024 (Library: OpenSSL 3.0.15 3 Sep 2024)    
  * ThreadX: 26826259 / 2024/09/23 14:02:56 

### Kernel info:

  * `/proc/cmdline: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe  smsc95xx.macaddr=2C:CF:67:B2:8D:E8 vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000  console=ttyAMA10,115200 console=tty1 root=PARTUUID=c15065e6-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles`
  * Vulnerability Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl
  * Vulnerability Spectre v1:             Mitigation; __user pointer sanitization
  * Vulnerability Spectre v2:             Mitigation; CSV2, BHB
  * Kernel 6.6.51+rpt-rpi-2712 / CONFIG_HZ=250

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        fake/real   load %cpu %sys %usr %nice %io %irq   Temp    VCore    PMIC   DC(V)
15:18:04: 2400/2400MHz  3.33   2%   0%   2%   0%   0%   0%  78.2°C  0.8948V   3.0W   5.12V 
15:19:04: 2400/2400MHz  1.22   0%   0%   0%   0%   0%   0%  70.5°C  0.8948V   2.9W   5.12V 
Cleaning up. Done.
Checking cpufreq OPP again. Done.

Clockspeeds now at 73.8°C:

    cpu0 (Cortex-A76): OPP: 2400, ThreadX: 2400, Measured: 2399 

Results validation:

  * Measured clockspeed not lower than advertised max CPU clockspeed
  * No swapping
  * Background activity (%system) OK
  * Throttling / frequency capping (under-voltage) occured -> https://tinyurl.com/4ky59sys / https://tinyurl.com/3j2c66kd


Querying ThreadX on RPi for thermal or undervoltage issues:

11100000000000000000
|||             |||_ under-voltage
|||             ||_ currently throttled
|||             |_ arm frequency capped
|||_ under-voltage has occurred since last reboot
||_ throttling has occurred since last reboot
|_ arm frequency capped has occurred since last reboot

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

pi@raspberrypi:~ $ sudo ./sbc-bench.sh -r

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

sbc-bench v0.9.68

Installing needed tools: distro packages already installed, 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 (12 minutes elapsed).

Results validation:

* Measured clockspeed not lower than advertised max CPU clockspeed

* No swapping

* Background activity (%system) OK

* Too much other background activity: 0% avg, 6% max -> https://tinyurl.com/mr2wy5uv

* Throttling / frequency capping (under-voltage) occured -> https://tinyurl.com/4ky59sys / https://tinyurl.com/3j2c66kd

Full results uploaded to https://0x0.st/XRKl.txt

# Raspberry Pi Compute Module 5 Rev 1.0

Tested with sbc-bench v0.9.68 on Wed, 27 Nov 2024 15:17:58 +0700. Full info: [https://0x0.st/XRKl.txt](http://0x0.st/XRKl.txt)

### General information:

Information courtesy of cpufetch:

SoC: Broadcom BCM2712

Technology: 16nm

Microarchitecture: Cortex-A76

Max Frequency: 2.400 GHz

Cores: 4 cores

Features: NEON,SHA1,SHA2,AES,CRC32

BCM2712, Kernel: aarch64, Userland: arm64

CPU sysfs topology (clusters, cpufreq members, clockspeeds)

cpufreq min max

CPU cluster policy speed speed core type

0 0 0 1500 2400 Cortex-A76 / r4p1

1 0 0 1500 2400 Cortex-A76 / r4p1

2 0 0 1500 2400 Cortex-A76 / r4p1

3 0 0 1500 2400 Cortex-A76 / r4p1

4041 KB available RAM

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

Original governor settings:

cpufreq-policy0: performance / 2400 MHz (conservative ondemand userspace powersave performance schedutil / 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400)

Tuned governor settings:

cpufreq-policy0: performance / 2400 MHz

Status of performance related policies found below /sys:

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

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

Before at 73.2°C:

cpu0 (Cortex-A76): OPP: 2400, ThreadX: 2400, Measured: 2399

After at 84.8°C (throttled):

cpu0 (Cortex-A76): OPP: 2400, ThreadX: 2400, Measured: 2403

### Performance baseline

* memcpy: 5304.3 MB/s, memchr: 13456.6 MB/s, memset: 11936.3 MB/s

* 16M latency: 119.0 118.8 117.2 115.3 122.0 119.1 139.2 134.3

* 128M latency: 135.0 134.2 136.5 141.3 135.2 134.0 135.7 136.4

* 7-zip MIPS (3 consecutive runs): 9831, 9085, 8992 (9300 avg), single-threaded: 3191

* `aes-256-cbc 540403.95k 1003620.69k 1256036.52k 1332896.09k 1365554.52k 1367932.93k`

* `aes-256-cbc 540634.79k 1003600.96k 1256047.70k 1332894.72k 1365325.14k 1368255.15k`

### PCIe and storage devices:

* Raspberry RP1 PCIe 2.0 South Bridge: Speed 5GT/s, Width x4, driver in use: rp1, ASPM Disabled

* 238.5GB "SAMSUNG MZ9LQ256HBJD-00BVL" SSD as /dev/nvme0: Speed 8GT/s, Width x1 (downgraded), 0% worn out, drive temp: 40°C, ASPM Disabled

* 29.1GB "Samsung BJTD4R" HS400 Enhanced strobe eMMC 5.1 card as /dev/mmcblk0: date 07/2024, manfid/oemid: 0x000015/0x0100, hw/fw rev: 0x0/0x0300000000000000

### Swap configuration:

* /var/swap on /dev/nvme0n1p2: 200.0M (0K used)

### Software versions:

* Debian GNU/Linux 12 (bookworm)

* Build scripts: http://archive.raspberrypi.com/debian/ bookworm main

* Compiler: /usr/bin/gcc (Debian 12.2.0-14) 12.2.0 / aarch64-linux-gnu

* OpenSSL 3.0.15, built on 3 Sep 2024 (Library: OpenSSL 3.0.15 3 Sep 2024)

* ThreadX: 26826259 / 2024/09/23 14:02:56

### Kernel info:

* `/proc/cmdline: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe smsc95xx.macaddr=2C:CF:67:B2:8D:E8 vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000 console=ttyAMA10,115200 console=tty1 root=PARTUUID=c15065e6-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles`

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

* Vulnerability Spectre v1: Mitigation; __user pointer sanitization

* Vulnerability Spectre v2: Mitigation; CSV2, BHB

* Kernel 6.6.51+rpt-rpi-2712 / CONFIG_HZ=250

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 fake/real load %cpu %sys %usr %nice %io %irq Temp VCore PMIC DC(V)

15:18:04: 2400/2400MHz 3.33 2% 0% 2% 0% 0% 0% 78.2°C 0.8948V 3.0W 5.12V

15:19:04: 2400/2400MHz 1.22 0% 0% 0% 0% 0% 0% 70.5°C 0.8948V 2.9W 5.12V

Cleaning up. Done.

Checking cpufreq OPP again. Done.

Clockspeeds now at 73.8°C:

cpu0 (Cortex-A76): OPP: 2400, ThreadX: 2400, Measured: 2399

Results validation:

* Measured clockspeed not lower than advertised max CPU clockspeed

* No swapping

* Background activity (%system) OK

* Throttling / frequency capping (under-voltage) occured -> https://tinyurl.com/4ky59sys / https://tinyurl.com/3j2c66kd

Querying ThreadX on RPi for thermal or undervoltage issues:

11100000000000000000

||| |||_ under-voltage

||| ||_ currently throttled

||| |_ arm frequency capped

|||_ under-voltage has occurred since last reboot

||_ throttling has occurred since last reboot

|_ arm frequency capped has occurred since last reboot

Note the devkit is placed on my terrace outdoors with a 32°C ambient temperature. Here’s the CPU temperature from rpimonitor during the benchmark.

Looking at the log, The real CPU frequency dropped as low as 1.5 GHz during 7-zip:

System health while running 7-zip multi core benchmark:

Time        fake/real   load %cpu %sys %usr %nice %io %irq   Temp    VCore    PMIC   DC(V)
15:10:35: 2400/2400MHz  1.07   0%   0%   0%   0%   0%   0%  76.5°C  0.8945V   3.4W   5.12V 
15:10:45: 2400/2256MHz  1.67  92%   0%  92%   0%   0%   0%  83.7°C  0.8948V   5.3W   5.11V 
15:10:56: 2400/1500MHz  2.03  94%   0%  93%   0%   0%   0%  84.2°C  0.8958V   6.0W   5.11V 
15:11:06: 2400/1500MHz  2.33  90%   0%  89%   0%   0%   0%  84.8°C  0.8953V   5.5W   5.11V 
15:11:17: 2400/2201MHz  2.74  97%   0%  96%   0%   0%   0%  84.8°C  0.7489V   3.2W   5.13V 
15:11:28: 2400/2201MHz  3.15  94%   0%  94%   0%   0%   0%  85.3°C  0.8950V   5.2W   5.13V 
15:11:38: 2400/1500MHz  3.44  94%   0%  93%   0%   0%   0%  83.2°C  0.8961V   5.7W   5.12V 
15:11:48: 2400/2146MHz  3.53  93%   0%  92%   0%   0%   0%  83.2°C  0.8932V   5.1W   5.10V 
15:12:01: 2400/1500MHz  3.68  93%   1%  91%   0%   0%   0%  85.3°C  0.7497V   4.1W   5.13V 
15:12:12: 2400/1500MHz  3.73  96%   0%  96%   0%   0%   0%  85.9°C  0.7497V   3.7W   5.12V 
15:12:23: 2400/2146MHz  3.86  94%   0%  93%   0%   0%   0%  84.2°C  0.7505V   4.0W   5.15V 
15:12:33: 2400/2256MHz  3.88  96%   0%  95%   0%   0%   0%  83.2°C  0.7497V   3.0W   5.15V 
15:12:46: 2400/1500MHz  3.90  88%   1%  86%   0%   0%   0%  84.8°C  0.8948V   5.6W   5.12V

System health while running 7-zip multi core benchmark:

Time fake/real load %cpu %sys %usr %nice %io %irq Temp VCore PMIC DC(V)

15:10:35: 2400/2400MHz 1.07 0% 0% 0% 0% 0% 0% 76.5°C 0.8945V 3.4W 5.12V

15:10:45: 2400/2256MHz 1.67 92% 0% 92% 0% 0% 0% 83.7°C 0.8948V 5.3W 5.11V

15:10:56: 2400/1500MHz 2.03 94% 0% 93% 0% 0% 0% 84.2°C 0.8958V 6.0W 5.11V

15:11:06: 2400/1500MHz 2.33 90% 0% 89% 0% 0% 0% 84.8°C 0.8953V 5.5W 5.11V

15:11:17: 2400/2201MHz 2.74 97% 0% 96% 0% 0% 0% 84.8°C 0.7489V 3.2W 5.13V

15:11:28: 2400/2201MHz 3.15 94% 0% 94% 0% 0% 0% 85.3°C 0.8950V 5.2W 5.13V

15:11:38: 2400/1500MHz 3.44 94% 0% 93% 0% 0% 0% 83.2°C 0.8961V 5.7W 5.12V

15:11:48: 2400/2146MHz 3.53 93% 0% 92% 0% 0% 0% 83.2°C 0.8932V 5.1W 5.10V

15:12:01: 2400/1500MHz 3.68 93% 1% 91% 0% 0% 0% 85.3°C 0.7497V 4.1W 5.13V

15:12:12: 2400/1500MHz 3.73 96% 0% 96% 0% 0% 0% 85.9°C 0.7497V 3.7W 5.12V

15:12:23: 2400/2146MHz 3.86 94% 0% 93% 0% 0% 0% 84.2°C 0.7505V 4.0W 5.15V

15:12:33: 2400/2256MHz 3.88 96% 0% 95% 0% 0% 0% 83.2°C 0.7497V 3.0W 5.15V

15:12:46: 2400/1500MHz 3.90 88% 1% 86% 0% 0% 0% 84.8°C 0.8948V 5.6W 5.12V

The fan turned at full speed, but it was still not enough. I’ll repeat the test indoors, and then with the board outside the case using the official Cooler. I’ll also try to get a CM5 active cooler from EDATEC as shown below.

The company did their own test at 25°C using sysbench, and the CPU temperature topped at around 60°C with the active cooler.

Rapsberry Pi CM5 active cooler CPU temperature frequency sysbench — A: Raspberry Pi IO board with CM5 without any heatsink at all, B: Raspberry IO board with CM5 and EDATEC CM5 active cooler.

They also tested the same setup at 60°C, and the Raspberry Pi CM5 (without a heatsink) only ran for about 16 minutes before dying. With the active cooler, the temperature stabilized at 74°C under those conditions.

I’ll do more testing in the second part of the review. I’d like to thank Raspberry Pi for sending the Development Kit for CM5 for review. It can be purchased for $130 plus taxes and shipping on authorized distributors.

Jean-Luc Aufranc (CNXSoft)

Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.