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.

 

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:

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:

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

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:

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

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:

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:

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:

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

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:

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.

Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress