AMD is reportedly preparing a significant leap in its GPU technology with the upcoming RDNA 5 or UDNA architecture. Leaks suggest the flagship models could boast 96 Compute Units (CUs) and a 384-bit memory bus, representing a substantial 50% increase in core count over the current RDNA 4 generation. This move signals AMD’s intent to re-enter the high-end GPU market with a competitive offering.
A New Era for AMD GPUs
Rumors from reliable leakers, particularly “Kepler_L2,” indicate that AMD’s next-generation GPU lineup, potentially dubbed UDNA or RDNA 5, will feature a significant increase in processing power. The top-tier Navi 5X SKU is expected to house 96 CUs, a substantial upgrade from the 64 CUs found in the flagship RDNA 4 GPUs. This architectural shift aims to deliver enhanced performance across various workloads, including gaming, ray tracing, and AI-powered upscaling technologies like FSR 5.
- Flagship GPU: 96 CUs with a 384-bit memory bus
- Mid-range GPU: 64 CUs with a 256-bit memory bus
- Entry-level GPU: 32 CUs with a 128-bit memory bus
Memory and Performance Expectations
The rumored 384-bit memory bus for the flagship model is a notable improvement over the 256-bit bus in current RDNA 4 cards. While the specific memory type is yet to be confirmed, GDDR7 is a strong possibility, which would further boost bandwidth and overall performance. This increased memory bandwidth is crucial for handling the demands of modern games and applications, especially at higher resolutions and with advanced graphical settings.
AMD’s focus on a larger GPU with more CUs and a wider memory bus suggests a direct challenge to NVIDIA’s high-end offerings. The company is also expected to leverage its collaboration with Sony on the PlayStation 6 console, with insights from this partnership potentially influencing the design and features of the RDNA 5/UDNA architecture. This could lead to significant advancements in ray tracing and AI capabilities, making path tracing more viable for gamers.
Market Positioning and Release Timeline
The introduction of a 96-CU flagship GPU would allow AMD to compete more directly with NVIDIA’s top-tier cards, a segment where AMD has been less present in the RDNA 4 generation. While the exact release date is not set, mass production of these new GPUs is anticipated to begin in Q2 2026, with availability expected in the latter half of 2026. This timeline positions AMD to compete with NVIDIA’s next-generation GPUs, potentially including the GeForce RTX 6000 series or the RTX 50 SUPER series, depending on their respective launch schedules.
This strategic move by AMD, if the rumors hold true, could mark a significant turning point for the company in the competitive GPU market, offering enthusiasts a powerful new option for high-performance computing and gaming.
Via Kepler_L2
