AMD's upcoming Ryzen 7040-series Phoenix processors for laptops based on the Zen 4 microarchitecture will come in three different packages aimed at different types of laptops. Chinese technology review channel Golden Pig Upgrade has published a review of the Ryzen 7 7840HS processor that, among other things, compares AMD's FP8 and FP7/FP7r2 form factors (via VideoCardz).
AMD's Ryzen 7040-series CPUs will be available in all-new FP8 as well as in FP7r2 and proven FP7 packages. The newer FP8 is noticeably larger than FP7/FP7r2. In addition, the FP8 package is designed to support higher-performance interfaces, including, among other things, AMD's MIPI CSI, a high-speed interface protocol for transmitting video and images from camera to host. This package is more suitable for devices requiring higher data throughput and advanced camera capabilities.
By contrast, the FP7 package is smaller and lighter. It has the same advanced features or performance as the FP8 package but offers a more compact solution for manufacturers looking to build slimmer devices without compromising processing power. In general, FP7/FP7r2 is a better fit for lightweight and portable devices. Also, some FP7 CPUs will be compatible with Ryzen 6000-series PCB designs. Regarding PCBs, AMD will advise its partners to use different printed circuit boards with Zen 4-based products with FP8 and FP7/FP7r2 packages. For example, some next-generation AMD Ryzen 7040-series laptops will rely on Type3 10-layer PTH (plated through-hole) PCBs, whereas others will use Type 4 HDI (high-density interconnect) PCBs. PTH PCBs use through-hole mounting, have lower component density, and are more cost-effective, making them suitable for less complex designs.
By contrast, HDI PCBs have higher component density and provide better electrical performance due to advanced fabrication techniques. As a result, they are ideal for miniaturized, high-performance electronic devices but tend to be more expensive to manufacture. In the case of AMD's notebooks, Type 3 PCBs will support up to LPDDR5X-6400 memory, whereas Type 4 PCBs will enable LPDDR5X-7500.
Another interesting aspect of AMD's 7040-series CPUs is the Ryzen 7040HS and 7040H specs. On paper, they appear to be identical in terms of TDP. However, AMD's Ryzen 6000H processors are rated for up to 45W, whereas Ryzen 6000HS is rated for up to 35W. Unfortunately, neither the Golden Pig Upgrade review nor AMD's website clarifies the differences between the 7000HS and 7000H series, leaving us puzzled.
The review compares the Ryzen 7 7840H, featuring the Radeon 780M-badged RDNA3 integrated GPU, to the Ryzen 7735H (based on the Rembrandt silicon with Zen 3 and RDNA 2) and two Intel Raptor Lake CPUs (13700H and 13500H). Although the latest Ryzen CPU boasts superior GPU performance, the upgrade from the previous-gen RDNA2 iGPU is not substantial. Ultimately, the iGPU outperforms the GeForce MX550 discrete GPU and approaches the performance of the GeForce GTX 1650 Max-Q designs.
First up, for all reviews AMD laptops were configured with 55W while the Intel laptops were configured at 65W. The Intel laptops were using LPDDR5-5200 memory while the AMD laptops were using LPDDR5x-6400 memory.
He also posted some synthetic and gaming benchmarks. In Cinebench R23 and 7-zip Intel had a lead in single-core benchmark but AMD ends up faster in multi-threaded benchmark.
In H.265, VCN 4.0 leads both VCN 3.0 and NVENC by 30% and 42%, respectively. In H.265, VCN 4.0 leads both VCN 3.0 and NVENC by 30% and 42%, respectively.
For gaming the reviewer took the NVIDIA GeForce RTX 4060 at the same TGP across all 4 laptops. The AMD Ryzen 7 7840HS delivered performance on par with Intel Core i7-13700H with LPDDR5-5600 memory, and the use of LPDDR5x-7500 ram could increase the performance substantially as noted in the video.
LPDDR5x-7500 actually delivered a 3.5% boost over LPDDR5x-6400 and a 16.57% boost over LPDDR5x-5500 memory, although I think there is still some room for improvement here, since these might be early Laptop samples. I couldn't actually understand the review properly.
That's because the AMD's RDNA 3 core architecture used in Phoenix APUs is internally referred to as "RDNA 2.5" in AMD's documents, which means it is somewhat in between the RDNA 2 and RDNA 3 arch designs. Hence the performance uplift is less.