Unless AMD surprises us with other applications of the architecture, Zen 3+ is closely linked to the APU of which it is a part, Ryzen 6000 or Rembrandt, and, therefore, many of the improvements it has received are not only exclusive to the new integrated CPU, but also to the rest of the elements that define the SoC in its entirety and that are shared.
So AMD has had to create a new Northbridge or integrated memory controller in order to support DDR5, this has allowed them to optimize the energy consumption of the central intercommunication in Ryzen 6000, so if we discard the new processor, good part of the chip is totally new, since to the new IMC we have to add an IO Hub or controller for peripherals improved with USB4 support and the adoption of the RDNA 2 architecture, with greater efficiency than the RX Vega that had been used until now.
As for the range of processors, this is divided into three families, Ryzen 6000U for ultrathin computers with a 15W TDPRyzen 6000HS for ultrathin gaming computers with a 28W TDP and to finish Ryzen 6000H for mainstream gaming laptops and a 45W TDP.
Zen 3+
This is the processor architecture used in Intel’s Ryzen 6000 APUs, where unlike other improvements made by AMD on the Zen architecture in the past, this time they have not focused on improving overall performance. So the average number of instructions per clock cycle that it can solve has not grown, but it has been optimized at the consumption level. Put another way, at equal clock speed Zen 3+ is no better than Zen 3.
However, its performance per watt is higher and this allows it to achieve higher clock speeds under the same consumption or be more suitable for ultra-thin high-performance laptops, especially those designed for gaming.
TSMC N6 Node
TSMC’s 6nm node is one of the keys to greater efficiency in Zen 3+, its peculiarity is that said manufacturing process was designed to translate designs from 7nm and AMD could have settled with it. However, the node in theory offers an increase in the density or number of transistors, but in theory it does not offer improvements in terms of consumption compared to the 7nm or N7 node.
The key is in the EUV, which is the novelty of this manufacturing process. Which has been used in the development of Zen 3+ to make a more efficient design. So from AMD they have not dedicated themselves to transferring the design of Zen 3 without any change, but have taken advantage of its strengths to create a more efficient version of said architecture.
The Power Delivery Network in Zen 3+ and Ryzen 6000
One of the first things that is done when creating a processor is to also design the internal circuitry that distributes power between the different components. Here we have to differentiate between the architecture of the different pieces that make up a processor and their relationship and, therefore, what we call architecture, from what is the organization of these components with respect to the PDN of the processor. It is quite common that when designing a processor for laptops and desktops, despite the fact that the architecture is common, the Power Delivery Network can be totally different due to the disparity in consumption.
What AMD has done in Zen 3+ is to redesign everything in PDN again with the goal of increasing efficiency per watt. Among the changes that have been added is greater control over the consumption states of each of the elements of the Ryzen 6000 APU of which Zen 3+ is a part. In total, they have made 50 changes in this aspect, which, due to the fact that they are the same chip, also affect the design of the integrated GPU and elements such as the memory controller.
From AMD they claim that the performance per watt compared to the Intel Core i9-1290HS is up to 2.62 times higher. It can be said without fear of being wrong that the path that has been taken with Zen 3+ has not been to improve performance, but to create a much more efficient CPU that meets the needs of a laptop.
iGPU RDNA 2 on Ryzen 6000
The latest news in Ryzen 6000 is an iteration of the RDNA 2 architecture optimized for the 6nm node and that adds several of the optimizations in terms of consumption and temperature that we will see in the monolithic versions of the future RDNA 3 for PC, as well as the recently launched RX 6500 XT. The performance per watt of AMD’s new graphics architecture has already been shown in gaming graphics cards to be much better than the old RX Vega, but we’ve had to wait almost three years to see them integrated into a PC APU.
For the same number of GPU cores, the architecture is 50% more powerful, so even the Ryzen 6000U models have superior performance in that aspect compared to any 5000 series APU and if we talk about the 12-core configuration of the H models, 45 W, and HS, 28 W, then it is very clear that the advances in terms of consumption or performance per watt are not reduced only to the CPU,
