The performance of "Ryzen 75800U" exceeds that of Apple M1 and achieves the same long-time driver as Intel.

Zen3 has become a Ryzen5000 with greatly improved CPU performance.

　Ryzen5000 is the successor to Ryzen4000 which was put into use last year (2020).

　Ryzen4000 uses the GPU of the CPU,Vega (Berga) generation of microarchitecture called Zen2, and the memory controller is SoC, which is based on a single chip of the PCI Express O controller (PCI Express and USB etc.).

　At this time, as the processing capacity of CPU as Zen2,CPU is greatly improved, and the manufacturing process used for manufacturing is refined to 7 nm, compared with the old model, the power consumption is greatly reduced, and as a processor for thin notebook PC, its attraction is greatly improved.

　In its follow-up Ryzen5000, the manufacturing technology is the same as before, staying at the structure of 7 nm,SoC, and there is no big strengthening point in this sense. In addition, like previous generations, GPU has become a Vega,Compute Unit of up to 8, which has not been greatly strengthened at this point.

　But the reinforcement is that CPU is evolving towards a third-generation Zen microarchitecture called Zen3 (in addition, some SKU still keep Zen2 unchanged).

　AMD is the first generation of Ryzen, which introduces a CPU micro-architecture designed from scratch, completely eliminating the reputation of Intel for losing performance. Then, in Zen2, it surpassed Intel with multithreaded performance by further re-examining the internal structure.

　For AMD, the last thing that compares poorly with Intel in CPU performance is single-threaded performance. This time Zen3 has made a variety of improvements in order to achieve the strengthening there.

　The biggest theme is the improvement of the cache hierarchy. In the traditional Zen 2 generation CPU, the eight CPU cores in a die are divided into two regions, and the quad core + 4MB L3 cache is divided into two blocks. In Zen3, eight CPU cores share the structure of 16MB's L3 cache, and one CPU core can directly access the larger L3 cache.

　Cache improvements result in reduced time for CPU to read data from memory, known as storage latency (latency). As the wait time decreases, CPU can process instructions faster, thus improving the performance of CPU.

　In addition, the internal improvement of CPU is added, as the front-end scheduler and branch prediction engine, as well as the parallelism of the execution unit are also improved. IPC is more efficient than Zen2.