AMD is poised to transform the landscape of integrated graphics with their upcoming Strix Point APUs, which are rumored to deliver performance on par with some entry-level discrete graphics cards. These next-generation APUs, featuring the RDNA 3+ integrated GPU (iGPU), are expected to bring significant advancements across both CPU and GPU capabilities, including the innovation of Zen 4 cores and the integration of the latest Ryzen AI architecture.
The Strix Point APU lineup will consist of two main configurations: Strix Point 1, a standard monolithic design, and Strix Point 2, a premium chiplet design. Both designs are set to harness the same cutting-edge architectures with differing configurations catering to a range of needs, from mainstream users to enthusiasts seeking high performance in their PCs.
Expected to arrive in the second half of 2024, the AMD Ryzen 8050 ‘Strix Point Mono’ will sport a plethora of features including a Zen 5 (4nm) monolithic design with up to 12 cores, a hybrid configuration of Zen 5 and Zen 5C, shared L3 cache of 32 MB, an integrated XDNA 2 engine, and approximately 25 TOPS AI Engine. Notably, these APUs are predicted to enhance CPU performance by up to 35% over their predecessor, the Phoenix, when operating at 50W.
For graphics, the 16 RDNA 3+ Compute Units on the high-end monolithic Strix Points are anticipated to challenge the GeForce RTX 3050 GPU, which will likely be a game-changer for mainstream gaming. The 12 Compute Unit variant of the RDNA 3+ iGPU may produce around 3150 points in the 3DMark Time Spy benchmark at a conservative TDP of 22-24W, further reaching 4000 points at full capacity.
Currently, AMD’s Radeon RX 6400 and NVIDIA’s GeForce RTX 2050 discrete notebook GPUs average about 3500 points in the same benchmark, generally at higher TDPs. This indicates a promising uplift of performance by roughly 35% for the new RDNA 3+ iGPU compared to its RNDA 3 predecessor, even more so when clock speeds and architectural tunings are considered.
In the era of AI and machine learning, it’s worth noting that these new chips were once rumored to feature larger caches that would have bolstered CPU and GPU performance, but the focus on AI might have led to different developmental choices. Despite the absence of innovations like on-die cache similar to Infinity Cache found in discrete GPUs, AMD is leaning toward implementing faster LPDDR5x memory solutions to address bandwidth demands. There are also ongoing discussions in the industry about using LPDDR6 to further reduce bandwidth limitations.
Moreover, these upcoming Strix Point chips are expected to significantly benefit the burgeoning market of Mini PCs and handheld gaming devices, offering powerful graphics performance in smaller, more portable form factors.
As AMD’s RDNA 3+ graphics architecture is projected to remain in Ryzen APUs until at least 2027, users can look forward to continuous advancements and optimizations similar to the improvements seen in the Vega iGPU line.
In summary, AMD’s Strix Point APUs represent a highly anticipated development in integrated graphics, promising substantial performance gains that could transform the accessibility and capabilities of APU-based systems, particularly for mainstream gamers and those fond of compact gaming solutions.
