AMD only revealed the bare-minimum details of their new Radeon RX Navi GPUs at their Computex keynote. It was mentioned that the graphics cards will be based on a new macro-architecture, the RDNA (Radeon DNA?) and like the Radeon VII and Zen 2 CPUs, will leverage TSMC’s 7nm process. Other than that the first party benchmarks were revealed showing that the top-end Navi was up to 10% faster than the NVIDIA GeForce RTX 2070.
The RDNA architecture features a redesigned compute engine design along with multi-level cache hierarchy to deliver faster clock speeds at low TDPs, resulting in a 1.25x higher IPC and 1.5x more performance per watt, when compared to the previous GCN architecture. Yes, there’s no hardware level ray-tracing support, otherwise, AMD would have mentioned it for sure.
Now according to WCCFTech, the Radeon RX 5700 series will feature two graphics cards, a 150W model with a total board power (TBP) of 180W and a 180W TDP model with a TBP of 225W. The later was demoed at AMD’s press conference and is expected to go up against the 175W RTX 2070 which retails for a bit less than $500. Considering that the Radeon VII is priced at $699, the faster Navi should go for $499 and the 150W part set to compete with the RTX 2060 will probably start from $349 to $399. Unless these cards are faster than their GeForce opponents in the majority of the games, it’ll be hard to recommend them, especially considering that they neither come with H/W ray-tracing nor something extra such as NVIDIA’s DLSS.
- AMD Desktop GPU Shipments Increased by 21% in Q1 2019, Intel’s Drop by 22.5%
- Intel Working on Dual Screen Honeycomb Glacier Gaming Laptop
ASRock demoed some early concept designs of the AMD Radeon Navi heatsinks, and all of them are triple fan designs. Both the Phantom and Taichi variants were present and it’s not clear which GPUs will get which heatsink: