AMD’s upcoming Radeon RX 8000 series, built on the RDNA 4 architecture, is anticipated to feature enhanced GDDR6 memory speeds, reaching up to 20 Gbps. This new information, from a source known for reliable hardware leaks, suggests improvements over previously expected 18 Gbps memory modules.
The Radeon RX 8000 lineup will include two main chips: the Navi 48 “GFX1201” and the Navi 44 “GFX1200”. Navi 48, the higher-end chip, is expected to support 256-bit and 192-bit memory bus interfaces, while the Navi 44 will use a 128-bit bus interface.
For the Navi 48 GPUs:
– 256-bit bus models will use either 18 or 20 Gbps GDDR6 memory, providing bandwidths of 576 GB/s and 640 GB/s, respectively.
– A 192-bit bus model will use 19 Gbps GDDR6 memory, offering 456 GB/s of bandwidth.
Navi 44 GPUs, configured with a 128-bit bus, are expected to utilize 18 Gbps GDDR6, yielding a bandwidth of 288 GB/s.
In terms of memory capacity, the higher-end models will top out at 16 GB on a 256-bit bus, while configurations with a 192-bit bus will include 12 GB of memory.
These GPUs are not intended to compete with NVIDIA’s top-tier RTX 50 “Blackwell” series but will instead target the mid-range and entry-level market segments, where AMD will face competition from both NVIDIA and Intel. The choice to use GDDR6 instead of GDDR7 is seen as a strategic decision to balance performance and cost.
The release of the AMD Radeon RX 8000 lineup is expected around CES 2025, with broader availability slated for the first quarter of 2025.
Here’s a look at AMD’s generational progress in GPU architecture:
