AMD processors are now considered to be far ahead in performance, allowing the company to take a relaxed approach in its roadmap planning. AMD can move at its own pace and, in some cases, it has started to “milk” the market, especially when it comes to its mainstream laptop APUs.
Currently, AMD’s mainstream laptop APU is the Ryzen AI 300 series, code-named Strix Point, which represents a significant upgrade. It features the new Zen 5 CPU architecture, RDNA 3.5 GPU architecture, and XDNA 2 NPU architecture.
According to a roadmap disclosed by LG, the upgrade next year will be codenamed “Gorgon Point”, and it will likely be branded as the Ryzen AI 400 series.
Unfortunately, the Gorgon Point will see no changes in its architecture. There will be no Zen 6, no RDNA 4, and no PCIe 5.0 support. The only changes are slight frequency boosts, improved AI performance, better yield, and reduced costs.
In other words, this is similar to the transition from the Ryzen 7040 series to the Ryzen 8040 series, and now to the entry-level Ryzen 200 series—no significant architectural change.
The Gorgon Point lineup consists of seven models, with the top-end Ryzen AI 9 HX being an upgrade from the current 375/370 series, featuring 12 CPU cores, 16 GPU cores, a base clock of 5.2+ GHz, and an NPU computing power of 55+ TOPS.
The Ryzen AI 9 model corresponds to the current 365, with 10 CPU cores, 12 GPU cores, but no noticeable changes in frequency or computing power.
The two Ryzen AI 7 models feature 8 CPU cores and 8 GPU cores, with a base clock of 5.2+ GHz, and NPU computing power of up to 55+ TOPS.
There are two Ryzen AI 5 models with 6 CPU cores and 4 GPU cores, with one of them having a 100 MHz lower CPU frequency.
The new Ryzen AI 3 features 4 CPU cores and 2 GPU cores, making it an entry-level option, but it still offers an NPU computing power of 50 TOPS.
LG also released a set of performance comparisons, but without specific project details or scores, only percentages. Whether single-core or multi-core performance, the improvements are only around 5%, which is marginal.
Based on information from the leaked shipment list, AMD’s upcoming Zen 6 architecture Medusa Point APU will feature the FP10 package interface.
It has been revealed that Medusa Point will no longer use the FP8 package interface like the previous Strix Point, but will adopt a slightly larger FP10 package interface, measuring 25mm x 42.5mm, which is about 6% larger than the FP8.
This change not only indicates a shift in the processor’s physical size but could also signal improvements in design and performance for the Medusa Point.
Medusa Point is expected to be manufactured using TSMC’s 3nm process, compared to the 4nm process used for Strix Point. Medusa Point will also adopt a Chiplet design, with one CCD (Core Complex Die) housing 12 Zen 6 cores and a separate I/O chip, which differs from Strix Point’s monolithic design.
In terms of integrated graphics, Medusa Point will feature an RDNA 3.5-based GPU, not the more advanced RDNA 4 architecture, as the latter will be reserved for discrete GPUs.
However, the RDNA 3.5 architecture should still provide solid graphical performance for most users, meeting daily usage and gaming needs.
According to AMD’s GPUOpen driver code, the next-gen Zen 6 architecture APU (codenamed Medusa Point) will not adopt the RDNA 4 architecture, but will continue using the RDNA 3.5 architecture.
The driver code clearly marks the device ID for RDNA 4 architecture (GfxIp12) as being exclusive to discrete GPUs, meaning RDNA 4 will only appear in AMD’s RX 9000 series discrete GPUs, not in APUs.
This decision may be based on AMD’s current success in mobile platforms. The Zen 5-based Strix Halo series already delivers leading integrated GPU performance with RDNA 3.5.
For APUs, the primary goal is balancing power consumption, heat dissipation, and performance, rather than targeting high-end gaming performance. Therefore, AMD likely believes continuing with RDNA 3.5 is sufficient for its next-gen APUs.
Skipping RDNA 4 could limit the use of FSR 4 (FidelityFX Super Resolution 4), as FSR 4 currently only supports RDNA 4 architecture.
Earlier, the blogger “Gold Pig Upgrade Pack” also revealed that AMD’s next-gen Zen 6 mobile APUs will not feature RDNA 4 or RDNA 5, but will continue to use an improved version of RDNA 3.X.
It’s clear that AMD now has the ability to “milk” the market, as its products are already performing well and in high demand.
According to the latest data, in February, AMD’s CPU shipment share on Amazon USA was an impressive 84.18%, while Intel’s share was only 15.82%.
In terms of specific products, AMD’s Ryzen 7 9800X3D continues to dominate the sales charts, with over 8,000 units sold in a single month.
Trailing behind are the Ryzen 5 5600X, Ryzen 9 9700X, Ryzen 7 7800X3D, and Ryzen 7 7700X, each of which saw sales of over 3,000 units.
Additionally, AMD’s entry-level products, like the Ryzen 5500, performed well, further solidifying its position in the mid-to-low-end market.
In contrast, Intel’s performance was underwhelming, with no products in the top 10. The only slightly better-performing product from Intel’s Arrow Lake-S series was the Core Ultra 7 265KF, which sold just 200 units.
Overall, Intel’s February sales champion was the Core i9 14900K, with slightly over 1,000 units sold, but it still lagged behind AMD’s strong performance.
In terms of revenue, AMD’s average selling price was $290.71, higher than Intel’s $251.18. AMD’s revenue share in February was 86.03%, while Intel’s was just 13.97%.
