AMD has introduced Ryzen 9000 processors for desktop PCs

As part of the annual Computex 2024 exhibition, AMD announced the release of long-awaited desktop processors based on the Zen 5 architecture — flagship and mid-budget solutions will be the first to see the light.

The first processors based on Zen 5 will be Ryzen 9 9950X, 9900X, Ryzen 7 9700X, as well as Ryzen 5 9600X. Novelties from AMD still support Socket AM5 and are highly differentiated in terms of performance, heat generation, and the amount of cache memory. The internal layout of the chips has not changed much and is represented by one I/O and two computational crystals (CCD) with eight cores each.

It is noteworthy that AMD persistently does not equip its solutions with small energy-efficient cores (big .LITTLE), which have been featured in Intel solutions for many years. Nevertheless, it is difficult to call this a disadvantage, since the classic layout, which consists only of "large" cores, facilitates the work of the task scheduler and guarantees maximum performance in any calculations.

The older version of the Ryzen 9 9950X is equipped with 16 physical cores running at 4.3GHz (up to 5.7GHz per core), 16MB of L2 cache, and 64MB of L3 cache. The declared heat release of the novelty — up to 170 W. The advantage of the 9950X over the flagship Intel Core i9-14900K reaches 56% in content creation tasks and up to 23% in games.

The most budget option for Zen 5 — Ryzen 5 9600X — boasts only 6 cores clocked at 3.9GHz (up to 5.4GHz per core) with class-typical heat dissipation of 65W. L2 cache size is 6 MB, and L3 — 32 MB.

Microarchitecture — the key to the advantage

It is interesting that the new processors got a noticeable advantage in speed thanks to the improved microarchitecture of Zen 5, and not to an increase in clock frequencies.

AMD itself claims a 16% increase in performance per clock , increased bandwidth between different levels of cache memory and an extended window of out-of-order execution of commands. In addition, the new architecture uses an improved branch prediction module, which increases performance by loading and executing instructions before they are needed.

The new architecture is expected to receive enhanced AI support and increased AVX-512 instruction processing speed. . Thus, on average, a 16% advantage of Zen 5 over Zen 4 is achieved.

It is interesting to see how fast the new chips will be in real usage scenarios — the first real tests will appear on the network soon.