APUs or Accelerated Processing Units are simply AMD processors that pack in a graphical chip inside the main CPU package. AMD has finally announced the hotly anticipated Ryzen 5000 APUs based on their latest and wildly successful Zen 3 architecture integrating Vega graphics. As of the present moment, these chips are only available to OEMs like Dell, HP, Lenovo, etc. and possibly a few SIs or System Integrators like Origin PC, CyberPower, etc. However, AMD does promise that the chips would be available for purchase to the general public later this year.
Ryzen 5000G APU Specifications
|Processor||Cores/Threads||Base Frequency||Turbo Frequency||No. of CUs||GPU Frequency||TDP|
|Ryzen 7 5700G||8/16||3.8 GHz||4.6 GHz||8||2.0 GHz||65W|
|Ryzen 7 5700GE||8/16||3.2 GHz||4.6 GHz||8||2.0 GHz||35W|
|Ryzen 5 5600G||6/12||3.9 GHz||4.4 GHz||7||1.9 GHz||65W|
|Ryzen 5 5600GE||6/12||3.4 GHz||4.4 GHz||7||1.9 GHz||35W|
|Ryzen 3 5300G||4/8||4.0 GHz||4.2 GHz||6||1.7 GHz||65W|
|Ryzen 3 5300GE||4/8||3.6 GHz||4.2 GHz||6||1.7 GHz||35W|
Every APU in the Ryzen 5000G lineup also ships with 24 PCIe lanes.
AMD’s 5000 series APUs are very easily distinguishable from the main lineup of Ryzen 5000 CPUs launched late last year. Every AMD APU has its name suffixed with a ‘G’. However, AMD also announces processors suffixed with ‘GE’ instead of ‘G’, for example, Ryzen 7 5700G as opposed to Ryzen 7 5700GE. What purpose does the ‘GE’ APUs serve? As is evident from the table above, the ‘GE’ APUs are essentially their ‘G’ counterpart with a significantly cut-down base clock and a 30W lower TDP.
What do the cut-down specs mean for you in terms of performance? We expect the more aggressive power limits and lower base frequencies in the ‘GE’ APUs would result in them performing 10-3% worse in multi-threaded workloads when compared to their ‘G’ counterpart. However, in single-threaded benchmarks, we do not expect any significant relative difference since such workloads do not push the entire CPU package, only a single core at a time, resulting in lower power consumption. Therefore, we assume the ‘GE’ APUs are targeted towards OEM or customer systems with more aggressive thermal limitations, possibly due to a constrained case in terms of volume, insufficient airflow, or both.
The Ryzen 5000G APUs use AMD’s latest Zen 3 architecture which is currently dominating the silicon industry in terms of raw performance. It is the same architecture on which the Ryzen 5000 and the EPYC 7003 CPUs are based. Naturally, these processors would feature the same 8-core compute complex (CCX) design sharing a unified L3 cache across all cores, as found in the Ryzen 5000 CPUs. However, while the Epyc and Ryzen 5000 CPUs have 32 megabytes of L3 cache, the APUs would feature only 16.
The top-end Ryzen 7 5700G comes with 8 cores, 16 threads and a Vega 8 graphics chip. It might come as disheartening news that the APUs would feature only a maximum of 8 CUs or Compute Units; with the number of CUs dropping by one, the more affordable the APU gets. Thus, the top-of-the-line Ryzen 7 APUs would feature 8 CUs, the Ryzen 5 APUs would pack in 7, while the Ryzen 3 APUs would come with only 6. AMD has shifted to TSMC’s 7 nm manufacturing process from the 12 nm process used in the previous-gen APUs with the current lineup.
AMD has backed up their decision to reduce the number of CUs from 11 to 8. The reasoning is that due to their switch to the more efficient 7 nm process from TSMC, they have been able to significantly improve the APU’s graphical capabilities despite reducing the number of CUs. However, how much performance AMD has been able to squeeze out of the new Vega chips compared to last-gen is yet to be objectively confirmed.
Capable memory is crucial for getting the most performance out of an APU, as we are aware. Therefore, the claimed increase in the APU’s graphical performance can be attributed to the very impressive 2.0 GHz frequency (up from 1.4 GHz on 12 nm); the 8 Vega CUs operate out of the box along with the 16 MB unified L3 cache, which is essential for further driving down latency and optimizing the APUs operation.