Monolithic vs MCM
Although the AMD Ryzen 5000, part of the 5000U and 5000H are based on Zen 3. It should be taken into account that in the case of 5000U and 5000H we are not talking about a processor made up of several chips but of a monolithic processor. This means that although the architecture and therefore the organization and communication of the elements is the same, in the case of SoCs and CPUs that we discuss in this article, we are referring to processors built on a single chip.
Núcleo Zen 3 and AMD Ryzen 5000 Mobile
The first point is the Zen 3 kernel, its architecture is exactly the same as in the desktop version based on chips, but there is a difference in input and it is that for reasons of space on the sound chip cache is only 16MB for the 8 cores, instead of 32MB, however compared to its direct predecessor, Ryzen 4000 also known as “Renoir”, the amount of L3 cache has increased from 1MB per core to 2MB per core, and a total of 8MB to 16MB.
In the case of all Zen architectures, the L3 cache serves as what we call a “Victim Cach2”, this is important because it adopts the cache lines that are deleted from the L2 cache, preventing them from falling into memory and having what saving from there, this is important because recovering data from memory has an order of magnitude higher energy cost than internal processor memory.
The slide above these lines is illustrative of “Cache Victim” in Zen architectures and does not belong to the Ryzen 5000 Mobile, we are clarifying this in case you see the 8MB L3 cache confused instead of the 16MB.
At the kernel level, the changes made by AMD in Zen 3 are mainly based on increasing the IPC, this was achieved by redoing the kernel in such a way that there is less conflict between instructions so that the average number of instructions resolved by increasing the clock cycle.
Thanks to these changes, the processor can not only achieve a higher clock speed with the same consumption as the Ryzen 4000, but also
Legacy memory controller, Northbridge and Southbridge
Apart from the Zen 3 cores, the rest of the Ryzen 5000U and Ryzen 5000H, at least in the communication section between the different processors and accelerators with memory is exactly the same as in the Ryzen 4000U and Ryzen 4000H, which allowed AMD to deploy its AMD Ryzen 5000 for notebook, just a few months after the launch of its predecessors.
In the Ryzen 4000, however, AMD used two memory controllers for the different versions of its chips, one for DDR4 and one for LPDDR4, for the Ryzen 5000 for laptops, AMD’s decision. was to use LPDDR4 memory., which can achieve the same bandwidths as DDR4 but with much lower power consumption than DDR4.
The best counterpart? There are no DIMMs or SO-DIMMs in LPDDR4 memory, so in most configurations the memory will be soldered to the board. The Ryzen 5000 Mobile, like its predecessor, inheriting the same communication structure, supports up to 4 LPDDR4 memory chips.
On the other hand, AMD also announced the support of LPDDR4X memory, this variant of LPDDR4 memory has the ability to achieve the same transfer capacity with much lower voltage and therefore with less power consumption. Its density is much higher than that of DDR4 and LPDDR4, although like the latter it is not sold as modules.
Ryzen 5000 Mobile power consumption optimizations
One of the things AMD has created for the Zen architecture is to separate the different elements of the SoC into small “islands” or differentiated “geographic” areas. In this way, AMD can turn off the parts of the SoC or CPU that are not in use at all times and ensure that they are not wasting power. Something in the Ryzen 5000 Mobile AMD has done with the audio codec and screen driver, two accelerators that are not always or partially active.
In the case of the second monitor’s screen controller, unless we connect an external monitor to the laptop, it doesn’t make sense to use power, the same with audio and video codecs. if, for example, we don’t use them.
But the most important change is the so-called CCPC, which consists of the Ryzen 5000 Mobile, each of the Zen 3 processor cores has its own domain, this means that at times these cores can fluctuate in their clock speed. and their tension. independently of each other.
Before, they all went up and down in unison, but with the new system we can see that depending on the application and the workload, not all cores go up in the same way unless it is is necessary. So it’s a way to save power and increase battery life and prevent the SoC from generating more heat than needed.
Changes to the Ryzen 5000 Mobile graphics system
For the umpteenth time, AMD has again used the 7nm AMD Vega architecture, so that RDNA architectures remain unheard of as an iGPU outside of game consoles. The main novelty? The ability to reach 2.12 GHz.
Keep in mind that AMD has spent the last few years fine-tuning its GPU architectures so that higher clock speeds can be achieved from the same ISA, be it RDNA, RDNA 2, GCN, or CDNA. Increasing the clock speed without changing the manufacturing node means redesigning all the stages of each of the instructions, increasing the quantity but decreasing the time per stage and thus increasing the clock frequency.
AMD has performed this process on all of its GPUs so that they can achieve higher clock speeds under TSMC’s 7nm node, whether dedicated or integrated.
Which models of Ryzen 5000 Mobile are Zen 3 and which are not?
So that there is no confusion during the purchase:
- All Ryzen 5000U models with an odd second number use Zen 2 cores, ie: 5100U, 5300U, 5500U, 5700U.
- All Ryzen 5000U models with the same second number use Zen 2 cores, ie: 5200U, 5400U, 5600U, 5800U.
- All 5000H models are Zen 3 core.
In addition, it should be added that like their predecessors, the H models have a TDP of 45 W and do not have or do not have the integrated GPU active, while in the U models their TDP is 15 W and have a Integrated GPU.