At an event in Los Angeles last week, AMD went through in more detail all the changes it's introducing with the Zen 5 CPU architecture. For a brief while, the chip giant also explained what's new in RDNA 3.5—a «fractional improvement» that's been «bolted onto» the current graphics processor design. In short, it's all about optimising rendering performance in mobile applications.
The updated design was introduced by Mark Papermaster, AMD's chief technology officer, and he began by pointing out that the changes culminated from its collaboration with Samsung, which licenses AMD's graphics tech for the Exynos range of smartphone and tablet processors.
«A lot of the techniques are ideal for notebooks,» he said. «They're ideal to give you that same great Radeon graphics experience, but at a much-reduced power and a much higher efficiency.»
There are no sweeping changes but that's to be expected from the architecture's codename. RDNA 3.5 exists to improve some of the performance bottlenecks that AMD's GPUs come across when used in low-power, low shader count configurations—namely the integrated Radeon GPUs in its mobile APUs, used in laptops and most handheld gaming PCs.
In the case of the latter, these typically run with power budgets of 15 W or so and while they can be given more, it's still significantly less power than the lowest of discrete GPUs get to enjoy. For example, a Radeon RX 6400 can use up to 54 W, which is 80% more power than the GPU in the Asus ROG Ally can demand.
Coupled with having a small number of Compute Units (CUs), it means that certain rendering operations, that would normally be of no concern to a desktop GPU, become more of a limiting factor in the overall performance. The first that Papermaster identified was the texture sampling rate.
In RDNA 3, each CU houses four texture units, each of which can sample and return one bilinearly-filtered texel per clock cycle. Papermaster said that AMD has doubled this figure to eight, in RDNA 3.5, though you
Read more on pcgamer.com