Home » Technology » “New Leaks about AMD Ryzen 8000 Mobile CPUs: Ryzen HS Phoenix, Ryzen HX Dragon Range, Hawk Point, Fire Range, Strix Point Mono and Strix Point Halo”

“New Leaks about AMD Ryzen 8000 Mobile CPUs: Ryzen HS Phoenix, Ryzen HX Dragon Range, Hawk Point, Fire Range, Strix Point Mono and Strix Point Halo”

On the YouTube channel Moore’s Law is Dead, new information has appeared about the upcoming mobile CPUs from AMD. Today we have the Ryzen 7000s, which are divided into two series, the Ryzen HS Phoenix, which use a powerful iGPU on the RDNA3 architecture (essentially an APU), and the Ryzen HX Dragon Range, which are usually expected to be paired with a dedicated GPU, so their iGPU is weaker and only has RDNA2 architecture. Mobile phones should also appear next year Ryzen 8000 and the new leaks show that there will be a lot of surprising specifications and the situation may seem very confusing.

In the case of the APU, there should be a new one Hawk Point, which will replace Phoenix (but it is only being introduced in new laptops these weeks). This should happen in the first half of 2024. There will be Zen 4 cores, production using the 4nm process, and the main change should be the use of a more modern RDNA3+ (RDNA3.5) GPU architecture, 12 CUs will remain. Another significant change should be the XDNA core (Xilinx’s AI processing chip). It counts with a maximum of 8 CPU cores.

While the previous one is a more pronounced “refresh”, we will get the new generation in the form of processors Fire Range, which will replace Dragon Range. Here we are talking about Ryzen mobile processors with the Zen 5 architecture, which is supposed to bring a further increase in performance. It may surprise someone that it still remains on 5nm production technology. The maximum is supposed to be 16 cores, the graphics chip is also supposed to be with the RDNA3+ architecture (some sources still say RDNA2), however these are processors for the HX series that will typically be paired with a dedicated GPU, so the integrated one is still only talked about 2 CU. Fire Range should appear in the second half of next year.

We get confused a lot with the next row, Strix Point. It is supposed to replace today’s Phoenix and the future Hawk Point, and there are two different variants being discussed. Strix Point “Mono” should continue to have a monolithic design, so it will continue to be an APU. According to leaks, a hybrid CPU architecture should appear here, so we can expect up to 12-core models (up to 8 for this product line so far), with 4 standard Zen 5 cores and 8 more economical Zen 5C cores. Compared to Intel, it does not lose the ability to process 2 threads, so this 12-core processor can handle 24 threads at once. Production is to take place using a 4nm process. There is talk of 24 MB of L3 cache memory.

We can also expect a faster GPU, it should be an RDNA3+ architecture and the number of CUs should be increased from 12 to 16. The performance should thus match the 35W variant of the GeForce RTX 3050, while it will still only be a GPU in the processor. Compared to Phoenix, it should be 35% more powerful at 50 W (according to the first benchmarks in Cinebench R23). There is talk of support for 128-bit LPDDR5X memories for the GPU, there is supposed to be an AI accelerator XDNA Engine with a performance of 20 TOPS, it is expected to be launched at the beginning of summer 2024. As for consumption, the range of 15 to 45 W is talked about here.

But he was also mentioned Strix Point “Halo”. Here it will be an extreme solution of CPU and GPU in one component. It will not be a monolithic, but a chiplet design, and instead of having a separate CPU and GPU in your computer, which even in gaming laptops take over 100 W each, here it will be in one component (and perhaps even lower consumption). We can find up to 16 cores with the Zen 5 architecture, and there should be a total of 64 MB of shared L3 cache. Compared to the 16-core Dragon Range at 90W consumption, it should deliver 25% more performance.

But the GPU chiplet can be RDNA3+ graphics with up to 40 CU (that’s how much the desktop Radeon RX 6700 XT has!) and the performance of this integrated solution should be equivalent to a 90W GeForce RTX 4070 Max-Q. In a laptop review Gigabyte Aero 16 BSF we had the 115W version of the RTX 4070 using Dynamic Boost and it was a really powerful GPU. There should be support for 256-bit LPDDR5X memories (this time also an additional 32 MB cache for the GPU), the XDNA Engine will come with a performance of even 40 TOPS. While the most economical processors of this architecture could only have a TDP of 25W, the most powerful configurations are expected to have a TDP of 120W. That sounds like an insane value for a mobile CPU, but the truth is that today’s CPU and GPU combinations can get well above that value in gaming laptops. Here, the launch is expected at the end of next year. Only then, however, will we see what will be filled from these leaks and what will not.

2023-04-25 11:25:28
#Mobile #AMD #Ryzen #iGPU #level #90W #RTX

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.