AMD is without a doubt the king of integrated graphics, with their Ryzen APUs providing the highest performance in a power-saving design. The AMD Ryzen APUs have continued to outperform their predecessors, and the recent switch to the RDNA 2 architecture has resulted in even higher performance. However, it appears that the next generations will be a big win for the budget gaming industry, as well as a fierce competitor for the blue team.
AMD has continued to develop despite being the market leader in integrated graphics. The business continued to improve the performance of their Radeon Vega graphics architecture even with their Pre-Rembrandt APUs. Shifting the Vega design to the brand new 7nm production node added power optimizations and enhanced clock speeds by a factor of ten to achieve greater performance.
AMD completed a complete architectural move to RDNA 2 with their most recent lineup, the Ryzen 6000 ‘Rembrandt.’ The APUs were the first to use a 6nm Radeon graphics architecture, which resulted in significant frequency and efficiency improvements.
The little Radeon 680M iGPU has so far shown impressive performance, approaching the desktop-grade GeForce GTX 1060 and even landing in better efficiency than the desktop RX 6400 graphics card, with up to 12 RDNA 2 Compute Units for a total of 768 SPs and the same architectural features as the desktop lineup. TechEphipany, a friend of ours, has a YouTube channel dedicated to gaming benchmarks and the performance of AMD’s APU, which you can find here.
However, the Ryzen APU’s iGPU has yet to reach its full potential. The AMD Rembrandt APU for the desktop platform is still on the way, and a higher TDP rating would allow users to overclock and extract more performance from the monolithic die’s RDNA 2 chips. Given how well the Radeon 680M performs in the laptop segment, budget PC gamers will be able to enjoy 1080p gaming without having to invest in a $150 US graphics card.
The next-generation Phoenix APUs from AMD could be a game-changer in the affordable PC gaming market
These APUs will purportedly use 5nm Zen 4 cores and RDNA iGPUs, despite the fact that they are still a year away from release. And, according to previous sources, AMD would use RDNA 2, but given that the APUs will be released a few months after the new Radeon lineup, it’s possible that AMD will use RDNA 3. RDNA 3 IP is known to exist on both 5nm and 6nm nodes.
There’s no news regarding the number of compute units in the Phoenix APUs yet, although new rumours suggest ranging from 16 to 24. That’s up to double the CUs of the existing iGPU, the Radeon 680M, and it’s the same CU count as the Radeon RX 6500 XT, a desktop graphics card, even at the lowest end.
Within a year of its availability, AMD might fine-tune the RDNA 2 for its Phoenix APU, and we could see performance quicker than a desktop Radeon RX 6400 or possibly the RX 6500 XT. This would be a game-changer for budget PC gamers, as they could buy a CPU instead of a graphics card for $150-$199 US and receive equivalent or better performance.
Furthermore, AMD has been working on technologies such as Phoenix that could be included in next-generation Ryzen APUs. The first and most obvious benefit is 3D V-Cache, which a gaming APU can greatly benefit from because Monolithic designs do not have access to a higher cache than their MCM brothers. On the 5800X3D, 3D V-Cache has already proven its usefulness, and it could pave the way for even better performance on Ryzen APUs that are both cache and bandwidth constrained. Infinity Cache is the other technology.
Infinity Cache is a big aid to the GPU, since it provides a near-GPU cache solution with plenty of high-speed bandwidth. AMD is working on adding it to its upcoming RDNA 3 MCM GPUs in the form of MCD through vertical stacking, but only time will tell if this technology makes it to Phoenix. Of course, these technologies would be nice, but they would require a lot of work and R&D to make them work, so if Phoenix doesn’t use them, perhaps the next generation will. However, even without them, Ryzen APUs can deliver current low-end graphics card performance by 2023 with a CU count hike, faster speeds, and power improvements.
AMD isn’t the only company investing in more powerful iGPUs. Intel has also revealed in its roadmap how they want to use their Arc GPU IPs in future CPUs. AMD may face stiff competition from Intel, which plans to integrate its latest graphics IP into Meteor Lake CPUs as early as 2023. Intel will use a multi-tiled technique for their iGPU, dubbed a GPU, in contrast to AMD’s monolithic solution (Tiled GPU).
All of the IP’s functionality, including as DeepLink, XeSS, raytracing, and more, will be retained on the GPUs. So it’ll all come down to who provides the best user experience, and AMD’s more complete Radeon Software suite may give it an advantage here. Intel still has a long way to go after the problems they had with their first Arc launch, so we’ll see how they do in 2023.
While this will signal the death of low-end discrete graphics cards, it will be for the better, since we have seen that entry-level graphics cards aren’t nearly as good as iGPUs. This is especially true in the laptop market, where the Navi 24 has performed significantly worse than the Radeon 680M.
Also Read:
NVIDIA is rumoured to have started testing its AD102 GPU for GeForce RTX 4090 Graphics Card
