AMD Zen 4 vs. Zen 4C Performance, Zen 4C Core Scaling With Ryzen 5 8500G

Besides the integrated RDNA3 graphics making the Ryzen 8000G series desktop APUs interesting, making the AMD Ryzen 5 8500G a fun benchmarking target besides its sub-$200 price tag is having a mix of Zen 4 and Zen 4C cores. Here are some benchmarks looking at the Zen 4 vs. Zen 4C performance and power efficiency when offlining various core combinations on the Ryzen 5 8500G desktop processor.

After looking at the Ryzen 8000G series performance at 35 Watt and 45 Watt cTDPs and 500+ benchmarks of the Ryzen 8500G/8600G against the Intel Core i3 14100 / i5 14500, one of the other follow-up benchmarking specials I wanted to explore was more testing on the Zen 4 vs. Zen 4C performance for the Ryzen 5 8500G.

The Ryzen 5 8500G six-core / 12-thread processor is comprised of two Zen 4 cores and then the other four cores are the smaller Zen 4C cores. I've already shown how this sub-$200 processor is quite capable especially with its integrated RDNA3 graphics, but I wanted to dive deeper on the Zen 4 vs. Zen 4C aspect. The Zen 4 cores on the 8500G have a maximum frequency of 5.07GHz while the Zen 4C cores on this processor have a maximum frequency of 3.73GHz.

Unfortunately from the BIOS there isn't a way exposed for disabling all Zen 4 or Zen 4C cores. But under Linux there is at least the ability to offline CPU cores... Though CPU0 can't be offlined, which is a Zen 4 core, so it wasn't possible to test the 8500G simply as a Zen 4C APU.

Looking at the contents of the /sys/devices/system/cpu/cpu*/cpufreq/scaling_max_freq files makes it possible to determine what CPU cores are Zen 4 and which are Zen 4C... In the case of the Ryzen 5 8500G, the Zen 4 cores all reported a 5.076GHz frequency while the Zen 4C cores were at 3.731GHz. Once figuring out the Zen 4(C) cores, looking at /sys/devices/system/cpu/cpu*/topology/thread_siblings_list exposes the relation of the CPU cores and the SMT sibling when wanting to offline a particular core and its SMT sibling thread.

So while being limited by not being able to offline the Zen 4 CPU0 core, I was able to benchmark various other core combinations of the Ryzen 5 8500G including:

- 1 x Zen 4
- 1 x Zen 4, 1 x Zen 4C
- 2 x Zen 4
- 2 x Zen 4, 1 x Zen 4C
- 2 x Zen 4, 2 x Zen 4C
- 2 x Zen 4, 3 x Zen 4C
- 2 x Zen 4, 4 x Zen 4C

SMT was left enabled across all the core configurations.

There's also the avenue of testing by ensuring task placement / isolation to particular CPU cores for workloads, but then the other CPU0 or similar is still online and can be spinning kernel work... Thus for this testing it was about benchmarking combinations while offlining the other Zen 4(C) cores and also then being able to accurately measure the CPU power consumption too for the difference. CPU core temperature monitoring was also recorded across the tested core combinations.

Let's take a look at how the benchmarks and power efficiency compare for these different core combinations in various Linux workloads on Ubuntu 23.10.