Comparison of Raspberry Pi 5 with 2GB and 8GB RAM - Hardware, benchmarks, and power consumption

The Raspberry Pi 5 with 2GB RAM was launched last week, and since I got a sample for review, I decided to compare it to the Raspberry Pi 5 with 8GB RAM to see if I could find any noticeable differences between the two boards.

I’ll start with a visual inspection to show differences on the PCBA, then check system information, run some benchmarks, check power consumption, and finally try to open as many tabs in Firefox until the 2GB RAM is filled and the system becomes unusable.

Raspberry Pi 5 2GB vs Raspberry Pi 8GB – visual inspection

We should first have a quick look at the boards and packages there’s no obvious difference apart from seeing 2GB RAM and 8GB RAM on the respective packages.

But if we look closer, we can see the resistors for memory capacity detection are in different locations for “2G” and “8G”, and markings on the BCM2712 SoC and memory chip don’t quite match between the two boards.

2GB RAM SBC resistor BCM2712 D0 — Close up on the 2GB RAM version with BCM2712D0

Raspberry Pi 5 8GB RAM resistor BCM2712C1 — Close up on the 8GB RAM model with Broadcom BCM2712C1

We can look up the memory parts on the Micron website:

Raspberry Pi 5 2GB – D8BQM = Micron MT53E512M32D1NP-046 WT:B 16Gbit LPDDR4 @ 2133 MHz (2GB RAM)
Raspberry Pi 5 8GB – D8CJN = Micron MT53E2G32D4DE-046 WT:C 64Gbit LPDDR4 @ 2133 MHz (8GB RAM)

The two chips are from the same family and just have different capacities, so we should not expect any differences in benchmarks.

Test setup and system information

I’ll be using the Raspberry Pi 5 2GB I received on August 19, and a new Raspberry Pi 8GB we were sent earlier this month to review the CrowView Note laptop shell. So both boards should be new and will be using the same storage device (a 128GB NVMe SSD) and active cooler as shown below. I’ll test the 8GB RAM board first, then move both the HAT+ with SSD and active cooler to the 2GB RAM version to test the boards in the exact same configuration.

The boards will also be connected to two RF dongles for a mouse and a keyboard, an Ethernet cable (GbE), an HDMI display, and the official 5V/5A USB-C power adapter. Raspberry Pi OS updated to the latest version with:

sudo apt update
sudo apt dist-upgrade

1 2	sudo apt update sudo apt dist-upgrade

I also disabled WiFi and the room temperature was around 27-28°C during the tests.

This is the system information I got on the Raspberry Pi 5 8GB with inxi:

pi@raspberrypi:~ $ sudo inxi -Fc0
System:
  Host: raspberrypi Kernel: 6.6.31+rpt-rpi-2712 arch: aarch64 bits: 64
    Console: pty pts/1 Distro: Debian GNU/Linux 12 (bookworm)
Machine:
  Type: ARM System: Raspberry Pi 5 Model B Rev 1.0 details: N/A rev: d04170
    serial: c31f47f82441d0f8
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: wayfire v: 0.7.5 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: k6.6.31+rpt-rpi-2712 status: kernel-api
Network:
  Device-1: Raspberry Pi RP1 PCIe 2.0 South Bridge driver: rp1
  IF: wlan0 state: down mac: 2c:cf:67:28:fd:28
  IF-ID-1: eth0 state: up speed: 1000 Mbps duplex: full
    mac: 2c:cf:67:28:fd:27
Bluetooth:
  Device-1: bcm7271-uart driver: bcm7271_uart
  Report: hciconfig ID: hci0 state: up address: 2C:CF:67:28:FD:29 bt-v: 3.0
  Device-2: bcm7271-uart driver: N/A
Drives:
  Local Storage: total: 119.24 GiB used: 13.07 GiB (11.0%)
  ID-1: /dev/nvme0n1 model: PCIe SSD size: 119.24 GiB
Partition:
  ID-1: / size: 116.77 GiB used: 12.99 GiB (11.1%) 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: 59.0 C mobo: N/A
  Fan Speeds (RPM): cpu: 3373
Info:
  Processes: 204 Uptime: 1m Memory: 7.87 GiB used: 704.5 MiB (8.7%) gpu: 8 MiB
  Init: systemd target: graphical (5) Shell: Sudo inxi: 3.3.26

pi@raspberrypi:~ $ sudo inxi -Fc0

System:

Host: raspberrypi Kernel: 6.6.31+rpt-rpi-2712 arch: aarch64 bits: 64

Console: pty pts/1 Distro: Debian GNU/Linux 12 (bookworm)

Machine:

Type: ARM System: Raspberry Pi 5 Model B Rev 1.0 details: N/A rev: d04170

serial: c31f47f82441d0f8

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: wayfire v: 0.7.5 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: k6.6.31+rpt-rpi-2712 status: kernel-api

Network:

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

IF: wlan0 state: down mac: 2c:cf:67:28:fd:28

IF-ID-1: eth0 state: up speed: 1000 Mbps duplex: full

mac: 2c:cf:67:28:fd:27

Bluetooth:

Device-1: bcm7271-uart driver: bcm7271_uart

Report: hciconfig ID: hci0 state: up address: 2C:CF:67:28:FD:29 bt-v: 3.0

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

Drives:

Local Storage: total: 119.24 GiB used: 13.07 GiB (11.0%)

ID-1: /dev/nvme0n1 model: PCIe SSD size: 119.24 GiB

Partition:

ID-1: / size: 116.77 GiB used: 12.99 GiB (11.1%) 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: 59.0 C mobo: N/A

Fan Speeds (RPM): cpu: 3373

Info:

Processes: 204 Uptime: 1m Memory: 7.87 GiB used: 704.5 MiB (8.7%) gpu: 8 MiB

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

and I repeated the same with the Raspberry Pi 5 2GB later:

pi@raspberrypi:~ $ sudo inxi -Fc0
System:
  Host: raspberrypi Kernel: 6.6.31+rpt-rpi-2712 arch: aarch64 bits: 64
    Console: pty pts/1 Distro: Debian GNU/Linux 12 (bookworm)
Machine:
  Type: ARM System: Raspberry Pi 5 Model B Rev 1.0 details: N/A rev: b04170
    serial: 696387f5901070da
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: wayfire v: 0.7.5 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: k6.6.31+rpt-rpi-2712 status: kernel-api
Network:
  Device-1: Raspberry Pi RP1 PCIe 2.0 South Bridge driver: rp1
  IF: wlan0 state: down mac: 2c:cf:67:83:d7:8f
  IF-ID-1: eth0 state: up speed: 1000 Mbps duplex: full
    mac: 2c:cf:67:83:d7:8e
Bluetooth:
  Device-1: bcm7271-uart driver: bcm7271_uart
  Report: hciconfig ID: hci0 state: up address: 2C:CF:67:83:D7:90 bt-v: 3.0
  Device-2: bcm7271-uart driver: N/A
Drives:
  Local Storage: total: 119.24 GiB used: 13.2 GiB (11.1%)
  ID-1: /dev/nvme0n1 model: PCIe SSD size: 119.24 GiB
Partition:
  ID-1: / size: 116.77 GiB used: 13.13 GiB (11.2%) 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: 49.1 C mobo: N/A
  Fan Speeds (RPM): cpu: 0
Info:
  Processes: 207 Uptime: 1m Memory: 1.97 GiB used: 664 MiB (33.0%) gpu: 8 MiB
  Init: systemd target: graphical (5) Shell: Sudo inxi: 3.3.26

pi@raspberrypi:~ $ sudo inxi -Fc0

System:

Host: raspberrypi Kernel: 6.6.31+rpt-rpi-2712 arch: aarch64 bits: 64

Console: pty pts/1 Distro: Debian GNU/Linux 12 (bookworm)

Machine:

Type: ARM System: Raspberry Pi 5 Model B Rev 1.0 details: N/A rev: b04170

serial: 696387f5901070da

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: wayfire v: 0.7.5 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: k6.6.31+rpt-rpi-2712 status: kernel-api

Network:

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

IF: wlan0 state: down mac: 2c:cf:67:83:d7:8f

IF-ID-1: eth0 state: up speed: 1000 Mbps duplex: full

mac: 2c:cf:67:83:d7:8e

Bluetooth:

Device-1: bcm7271-uart driver: bcm7271_uart

Report: hciconfig ID: hci0 state: up address: 2C:CF:67:83:D7:90 bt-v: 3.0

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

Drives:

Local Storage: total: 119.24 GiB used: 13.2 GiB (11.1%)

ID-1: /dev/nvme0n1 model: PCIe SSD size: 119.24 GiB

Partition:

ID-1: / size: 116.77 GiB used: 13.13 GiB (11.2%) 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: 49.1 C mobo: N/A

Fan Speeds (RPM): cpu: 0

Info:

Processes: 207 Uptime: 1m Memory: 1.97 GiB used: 664 MiB (33.0%) gpu: 8 MiB

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

I compared the two outputs in the Meld program and the main differences were the serial number, MAC addresses for WiFi, Ethernet, and Bluetooth, and the memory capacity (7.87 GB vs 1.97 GB). The revision of the board was also different: “d04170” for the 8GB model and “b04170” for the 2GB model. Some readers may also notice the differences in CPU temperature (59.0°C vs 49.1°C), but this can be explained easily since I ran inxi after updating Raspberry Pi OS on the 8GB model and ran the command right after booting the 2GB RAM board. I waited a bit more before running benchmarks, and the temperature of the new 2GB model crept up to over 56°C after a while just being idle.

Benchmarks

I’ll just use sbc-bench.sh script to run a few benchmarks starting on the 8GB model:

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

sbc-bench v0.9.67

Installing needed tools: apt-get -f -qq -y install sysstat lshw links mmc-utils smartmontools stress-ng. Something went wrong:

Trying to continue, tinymembench, ramlat, mhz, cpufetch, 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 (11 minutes elapsed).

Results validation:

  * Measured clockspeed not lower than advertised max CPU clockspeed
  * No swapping
  * Background activity (%system) OK
  * No throttling


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


# Raspberry Pi 5 Model B Rev 1.0

Tested with sbc-bench v0.9.67 on Mon, 26 Aug 2024 16:14:10 +0700. Full info: [https://0x0.st/XtHK.txt](http://0x0.st/XtHK.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

8048 KB available RAM

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

Original governor settings:

    cpufreq-policy0: ondemand / 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 57.3°C:

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

After at 71.0°C:

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

### Performance baseline

  * memcpy: 5370.2 MB/s, memchr: 13982.4 MB/s, memset: 12368.2 MB/s
  * 16M latency: 118.6 119.9 119.6 119.8 118.1 135.7 131.2 150.9 
  * 128M latency: 136.0 137.3 143.7 135.1 136.2 134.8 135.6 137.2 
  * 7-zip MIPS (3 consecutive runs): 10951, 10850, 10635 (10810 avg), single-threaded: 3172
  * `aes-256-cbc     580900.20k  1032045.01k  1266215.51k  1339040.77k  1365650.09k  1367965.70k`
  * `aes-256-cbc     581415.91k  1032843.29k  1265462.70k  1338715.14k  1365174.95k  1368227.84k`

### PCIe and storage devices:

  * Raspberry RP1 PCIe 2.0 South Bridge: Speed 5GT/s, Width x4, driver in use: rp1, ASPM Disabled
  * 119.2GB "PCIe SSD" SSD as /dev/nvme0: Speed 8GT/s, Width x1 (downgraded), 0% worn out, drive temp: 35°C, ASPM Disabled

### 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 3.0.13, built on 30 Jan 2024 (Library: OpenSSL 3.0.13 30 Jan 2024)    
  * ThreadX: 6fe0b091 / 2024/06/05 16:41:49 

### Kernel info:

  * `/proc/cmdline: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe  smsc95xx.macaddr=2C:CF:67:28:FD:27 vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000  console=ttyAMA10,115200 console=tty1 root=PARTUUID=a0dd71de-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles cfg80211.ieee80211_regdom=TH`
  * 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 6.6.31+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)
16:14:10: 2400/2400MHz  3.52  53%   0%  52%   0%   0%   0%  64.5°C  0.8905V   3.3W   5.13V 
16:15:10: 2400/2400MHz  1.29   0%   0%   0%   0%   0%   0%  59.5°C  0.8902V   3.1W   5.14V 
^C

Cleaning up. Done.
Checking cpufreq OPP again. Done.

Clockspeeds now at 59.5°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
  * No throttling
  * ondemand cpufreq governor used by distro but io_is_busy not set to 1 on all cores -> http://tinyurl.com/44pbmw79

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

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

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

sbc-bench v0.9.67

Installing needed tools: apt-get -f -qq -y install sysstat lshw links mmc-utils smartmontools stress-ng. Something went wrong:

Trying to continue, tinymembench, ramlat, mhz, cpufetch, 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 (11 minutes elapsed).

Results validation:

* Measured clockspeed not lower than advertised max CPU clockspeed

* No swapping

* Background activity (%system) OK

* No throttling

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

# Raspberry Pi 5 Model B Rev 1.0

Tested with sbc-bench v0.9.67 on Mon, 26 Aug 2024 16:14:10 +0700. Full info: [https://0x0.st/XtHK.txt](http://0x0.st/XtHK.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

8048 KB available RAM

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

Original governor settings:

cpufreq-policy0: ondemand / 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 57.3°C:

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

After at 71.0°C:

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

### Performance baseline

* memcpy: 5370.2 MB/s, memchr: 13982.4 MB/s, memset: 12368.2 MB/s

* 16M latency: 118.6 119.9 119.6 119.8 118.1 135.7 131.2 150.9

* 128M latency: 136.0 137.3 143.7 135.1 136.2 134.8 135.6 137.2

* 7-zip MIPS (3 consecutive runs): 10951, 10850, 10635 (10810 avg), single-threaded: 3172

* `aes-256-cbc 580900.20k 1032045.01k 1266215.51k 1339040.77k 1365650.09k 1367965.70k`

* `aes-256-cbc 581415.91k 1032843.29k 1265462.70k 1338715.14k 1365174.95k 1368227.84k`

### PCIe and storage devices:

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

* 119.2GB "PCIe SSD" SSD as /dev/nvme0: Speed 8GT/s, Width x1 (downgraded), 0% worn out, drive temp: 35°C, ASPM Disabled

### 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 3.0.13, built on 30 Jan 2024 (Library: OpenSSL 3.0.13 30 Jan 2024)

* ThreadX: 6fe0b091 / 2024/06/05 16:41:49

### Kernel info:

* `/proc/cmdline: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe smsc95xx.macaddr=2C:CF:67:28:FD:27 vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000 console=ttyAMA10,115200 console=tty1 root=PARTUUID=a0dd71de-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles cfg80211.ieee80211_regdom=TH`

* 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 6.6.31+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)

16:14:10: 2400/2400MHz 3.52 53% 0% 52% 0% 0% 0% 64.5°C 0.8905V 3.3W 5.13V

16:15:10: 2400/2400MHz 1.29 0% 0% 0% 0% 0% 0% 59.5°C 0.8902V 3.1W 5.14V

Cleaning up. Done.

Checking cpufreq OPP again. Done.

Clockspeeds now at 59.5°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

* No throttling

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

and repeated it later on the Raspberry Pi 5 2GB:

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

sbc-bench v0.9.67

Installing needed tools: distro packages already installed. 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 (11 minutes elapsed).

Results validation:

  * Measured clockspeed not lower than advertised max CPU clockspeed
  * No swapping
  * Background activity (%system) OK
  * No throttling


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


# Raspberry Pi 5 Model B Rev 1.0

Tested with sbc-bench v0.9.67 on Mon, 26 Aug 2024 16:54:15 +0700. Full info: [https://0x0.st/XtHM.txt](http://0x0.st/XtHM.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

2005 KB available RAM

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

Original governor settings:

    cpufreq-policy0: ondemand / 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 47.4°C:

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

After at 65.0°C:

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

### Performance baseline

  * memcpy: 5530.3 MB/s, memchr: 14174.6 MB/s, memset: 12842.8 MB/s
  * 16M latency: 117.2 114.1 117.1 115.0 116.6 119.7 142.6 136.8 
  * 128M latency: 134.5 133.5 136.5 140.3 134.4 133.2 133.9 135.0 
  * 7-zip MIPS (3 consecutive runs): 11034, 11048, 11033 (11040 avg), single-threaded: 3193
  * `aes-256-cbc     580920.16k  1032536.21k  1266243.16k  1337069.91k  1365860.35k  1368156.84k`
  * `aes-256-cbc     581292.82k  1032156.12k  1266186.07k  1337005.40k  1365658.28k  1368309.76k`

### PCIe and storage devices:

  * Raspberry RP1 PCIe 2.0 South Bridge: Speed 5GT/s, Width x4, driver in use: rp1, ASPM Disabled
  * 119.2GB "PCIe SSD" SSD as /dev/nvme0: Speed 8GT/s, Width x1 (downgraded), 0% worn out, drive temp: 35°C, ASPM Disabled

### 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 3.0.13, built on 30 Jan 2024 (Library: OpenSSL 3.0.13 30 Jan 2024)    
  * ThreadX: 6fe0b091 / 2024/06/05 16:41:49 

### Kernel info:

  * `/proc/cmdline: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe  smsc95xx.macaddr=2C:CF:67:83:D7:8E vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000  console=ttyAMA10,115200 console=tty1 root=PARTUUID=a0dd71de-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles cfg80211.ieee80211_regdom=TH`
  * 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 6.6.31+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)
16:54:15: 2400/2400MHz  3.93  30%   0%  30%   0%   0%   0%  58.4°C  0.8521V   2.3W   5.13V 
16:55:15: 2400/2400MHz  1.44   0%   0%   0%   0%   0%   0%  55.6°C  0.8521V   2.4W   5.12V 
^C

Cleaning up. Done.
Checking cpufreq OPP again. Done.

Clockspeeds now at 55.1°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
  * No throttling
  * ondemand cpufreq governor used by distro but io_is_busy not set to 1 on all cores -> http://tinyurl.com/44pbmw79

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

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

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

sbc-bench v0.9.67

Installing needed tools: distro packages already installed. 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 (11 minutes elapsed).

Results validation:

* Measured clockspeed not lower than advertised max CPU clockspeed

* No swapping

* Background activity (%system) OK

* No throttling

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

# Raspberry Pi 5 Model B Rev 1.0

Tested with sbc-bench v0.9.67 on Mon, 26 Aug 2024 16:54:15 +0700. Full info: [https://0x0.st/XtHM.txt](http://0x0.st/XtHM.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

2005 KB available RAM

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

Original governor settings:

cpufreq-policy0: ondemand / 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 47.4°C:

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

After at 65.0°C:

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

### Performance baseline

* memcpy: 5530.3 MB/s, memchr: 14174.6 MB/s, memset: 12842.8 MB/s

* 16M latency: 117.2 114.1 117.1 115.0 116.6 119.7 142.6 136.8

* 128M latency: 134.5 133.5 136.5 140.3 134.4 133.2 133.9 135.0

* 7-zip MIPS (3 consecutive runs): 11034, 11048, 11033 (11040 avg), single-threaded: 3193

* `aes-256-cbc 580920.16k 1032536.21k 1266243.16k 1337069.91k 1365860.35k 1368156.84k`

* `aes-256-cbc 581292.82k 1032156.12k 1266186.07k 1337005.40k 1365658.28k 1368309.76k`

### PCIe and storage devices:

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

* 119.2GB "PCIe SSD" SSD as /dev/nvme0: Speed 8GT/s, Width x1 (downgraded), 0% worn out, drive temp: 35°C, ASPM Disabled

### 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 3.0.13, built on 30 Jan 2024 (Library: OpenSSL 3.0.13 30 Jan 2024)

* ThreadX: 6fe0b091 / 2024/06/05 16:41:49

### Kernel info:

* `/proc/cmdline: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe smsc95xx.macaddr=2C:CF:67:83:D7:8E vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000 console=ttyAMA10,115200 console=tty1 root=PARTUUID=a0dd71de-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles cfg80211.ieee80211_regdom=TH`

* 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 6.6.31+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)

16:54:15: 2400/2400MHz 3.93 30% 0% 30% 0% 0% 0% 58.4°C 0.8521V 2.3W 5.13V

16:55:15: 2400/2400MHz 1.44 0% 0% 0% 0% 0% 0% 55.6°C 0.8521V 2.4W 5.12V

Cleaning up. Done.

Checking cpufreq OPP again. Done.

Clockspeeds now at 55.1°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

* No throttling

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

Let’s check some of the results in a comparison table.

	Raspberry Pi 5 2GB	Raspberry Pi 5 8GB	Delta
memset	12842.8 MB/s	12368.2 MB/s	-3.8%
memcpy	5530.3 MB/s	5370.2 MB/s	-2.9%
7-zip	11,040 MIPS	10,810 MIPS	-2.1%
AES-256 16K	1,368,309.76k	1,368,227.84k	+0.07%
Max CPU temperature	66.7°C	71.6°C	+4.9°C

There are some differences but probably not significant. Still, most are in favor of the 2GB RAM variant, but no user will be able to “feel the difference” in terms of performance (as long as we stay within the 2GB RAM limit). The temperature delta is higher, but the 71.6°C was only one data point, and others were a bit lower, yet still higher than on the 2GB model:

System health while running cpuminer:

Time        fake/real   load %cpu %sys %usr %nice %io %irq   Temp    VCore    PMIC   DC(V)
16:09:02: 2400/2400MHz  3.92  31%   0%  30%   0%   0%   0%  68.3°C  0.8907V   6.3W   5.13V 
16:09:45: 2400/2400MHz  3.96 100%   0%  99%   0%   0%   0%  68.3°C  0.8905V   6.3W   5.13V 
16:10:27: 2400/2400MHz  4.02 100%   0%  99%   0%   0%   0%  68.8°C  0.8910V   6.4W   5.13V 
16:11:09: 2400/2400MHz  4.01 100%   0% 100%   0%   0%   0%  69.4°C  0.8905V   6.4W   5.13V 
16:11:51: 2400/2400MHz  4.04 100%   0% 100%   0%   0%   0%  71.6°C  0.8905V   6.4W   5.13V 
16:12:34: 2400/2400MHz  4.02 100%   0%  99%   0%   0%   0%  69.4°C  0.8905V   6.4W   5.13V 
16:13:16: 2400/2400MHz  4.01 100%   0% 100%   0%   0%   0%  69.4°C  0.8905V   6.4W   5.13V 
16:13:58: 2400/2400MHz  4.08 100%   0% 100%   0%   0%   0%  69.4°C  0.8905V   6.4W   5.12V

System health while running cpuminer:

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

16:09:02: 2400/2400MHz 3.92 31% 0% 30% 0% 0% 0% 68.3°C 0.8907V 6.3W 5.13V

16:09:45: 2400/2400MHz 3.96 100% 0% 99% 0% 0% 0% 68.3°C 0.8905V 6.3W 5.13V

16:10:27: 2400/2400MHz 4.02 100% 0% 99% 0% 0% 0% 68.8°C 0.8910V 6.4W 5.13V

16:11:09: 2400/2400MHz 4.01 100% 0% 100% 0% 0% 0% 69.4°C 0.8905V 6.4W 5.13V

16:11:51: 2400/2400MHz 4.04 100% 0% 100% 0% 0% 0% 71.6°C 0.8905V 6.4W 5.13V

16:12:34: 2400/2400MHz 4.02 100% 0% 99% 0% 0% 0% 69.4°C 0.8905V 6.4W 5.13V

16:13:16: 2400/2400MHz 4.01 100% 0% 100% 0% 0% 0% 69.4°C 0.8905V 6.4W 5.13V

16:13:58: 2400/2400MHz 4.08 100% 0% 100% 0% 0% 0% 69.4°C 0.8905V 6.4W 5.12V

I set my aircon to 27C during the tests so in theory, the room temperature should be stable, and there may be some variations… But it’s safe to say the Raspberry Pi 5 2GB was at least 3°C cooler than the 8GB model in those tests. Note that it may be board-dependent, and the tests would have to be repeated on several boards in order to come to a conclusion.

Power consumption

I measured the power consumption with a wall power meter using the test configuration mentioned above, so the only difference is the board used.

	Raspberry Pi 5 2GB	Raspberry Pi 5 8GB
Power off	1.8 Watts	2.7 Watts
Idle (ondemand)	2.7 Watts	3.5 Watts
Idle (performance)	4.4 Watts	5.5 Watts

By default, the governor is set to “ondemand”, but I also switched it to “performance” using the command:

sudo ./sbc-bench.sh -R

1	sudo ./sbc-bench.sh -R

which may also perform some other optimizations.

I was only expecting some small variation here, but we get up to around a Watt difference in terms of power consumption. This also happens when powered off, so it’s really strange.

So I check the EEPROM configuration on Raspberry Pi 5 2GB:

pi@raspberrypi:~ $ sudo rpi-eeprom-config
[all]
BOOT_UART=1
POWER_OFF_ON_HALT=0
BOOT_ORDER=0xf461

pi@raspberrypi:~ $ sudo rpi-eeprom-config

[all]

BOOT_UART=1

POWER_OFF_ON_HALT=0

BOOT_ORDER=0xf461

and confirmed it was the same on the 8GB RAM model:

pi@raspberrypi:~ $ sudo rpi-eeprom-config
[all]
BOOT_UART=1
POWER_OFF_ON_HALT=0
BOOT_ORDER=0xf461

pi@raspberrypi:~ $ sudo rpi-eeprom-config

[all]

BOOT_UART=1

POWER_OFF_ON_HALT=0

BOOT_ORDER=0xf461

This would have to be tested on multiple boards with 2GB and 8GB RAM to confirm the pattern. I’d like to try some other power measurement tools like the Otii Arc power supply and power meter which I used to test the Raspberry Pi Zero 2 W’s power consumption, but I don’t have it with me and I’m not sure it can deliver even power to boot the Pi 5 since it tops at 5V/2.5A.

I also used the default setting, and we can further lower the consumption when the board is powered off by changing the settings as follows:

pi@raspberrypi:~ $ sudo rpi-eeprom-config --edit
[all] 
BOOT_UART=1
POWER_OFF_ON_HALT=1
BOOT_ORDER=0xf461
WAKE_ON_GPIO=0

pi@raspberrypi:~ $ sudo rpi-eeprom-config --edit

[all]

BOOT_UART=1

POWER_OFF_ON_HALT=1

BOOT_ORDER=0xf461

WAKE_ON_GPIO=0

My Raspberry Pi 5 2GB board would (only) consume 0.8W to 0.9W when powered off in that configuration. The downside is the configuration may impact some specific HAT boards.

How many tabs can I open in Firefox with the Raspberry Pi 5 using 2GB RAM?

Let’s answer this question by opening several tabs in Firefox using CNX Software’s website. I’ll report the number of tabs and memory usage.

Idle – 603MB RAM used, 0MB swap
Firefox launched, no tabs – 1.4GB RAM used, 6MB swap
One tab – 1.1GB RAM used, 184MB swap
Five tabs – 1.4GB RAM used, 199MB swap (full)
Ten tabs – 1.6GB RAM used, 199MB swap
Fifteen tabs – 1.7GB RAM used, 199MB swap
Twenty tabs – The system hung while switching between tabs and I could not access the SSH window (actually I could but there was a lag of a few minutes between key presses)

For each step, I would go through all open tabs, and scroll them to make sure the system is still responsive. Firefox could load twenty tabs, and I managed to visit the first two new tabs, but my mouse froze when I went to the third new tab. The results will depend on the websites visited, but I’ve been able to load 15 tabs in Firefox when visiting CNX Software and the system was still responsive. This obviously assumes you don’t run other programs.

Conclusion

As one would have expected the Raspberry Pi 5 2GB delivers about the same performance as the Raspberry Pi 5 8GB in benchmarks, but my specific board at least appeared to run slightly cooler and consume less power than the Raspberry Pi 5 with 8GB RAM. This would have to be tested on a larger number of boards to definitely conclude these potential advantages of the Raspberry Pi 5 with 2GB RAM, or disprove the theory.

Most people won’t need to care about those, and the important part is to find out how much memory should need for your target application/project, and if 2GB RAM is enough, the cheaper $50 price tag makes it a no-brainer.

I’d like to thank Raspberry Pi Limited for sending the new Pi 5 with 2GB RAM for review. You should be able to purchase it from any local resellers that offer Raspberry Pi boards and accessories.

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.