AMD is seemingly closing the efficiency gap between its Radeon GPUs and Nvidia’s GeForce rivals. Unlike RTX 50 Series GPUs which remained on the same TSMC 4N process as their predecessors, RX 9000 Series is expected to move away from RX 7000’s TSMC 5N manufacturing. This should put both brands on a similar footing, leaving architectural differences to define efficiency.
According to a rumour shared by leaker Hoang Anh Phu on X before being removed, AMD may target a 220W TBP (total board power) for RX 9070 non-XT and 304W for RX 9070 XT. While the tweet didn’t mention the GPUs name, the frequency figure alongside this power claim is identical to the one we saw on previous RX 9070 XT rumours. And besides, AMD isn’t planning many releases in the near future aside from these RDNA 4 GPUs.
If correct, this would mark a slight but welcome improvement compared to RX 7900 XTX‘s 355W and RX 7900 XT‘s 300W targets. 304W would put the RX 9070 XT head-to-head with Nvidia’s RTX 5070 Ti, which is also expected to offer similar graphics performance. This is understandable since the underlying RDNA 4 architecture is seemingly also using TSMC’s 4N node.
VideoCardz also confirmed that the leaked PCIe 5.0 interface is correct, making RX 9000 as the third GPU series supporting it after Moore Threads’ MTT S80 and Nvidia’s RTX 50 Series. Likewise, the RX 9000 lineup is set to include the latest DisplayPort 2.1a and HDMI 2.1b video outputs.
With this, the specs of RX 9070 Series are complete, though not official. As things stand, RX 9070 XT is set to feature a 4,096-core Navi 48 chip clocked up to 2.97GHz, coupled with 16GB of GDDR6 memory running on a 256-bit bus. Radeon RX 9070 non-XT, on the other hand, leverages a cut-down Navi 48 chip packing 3,584 cores clocked up to 2.52GHz, also linked to 16GB of GDDR6 memory via a 256-bit bus.
We should have confirmation of these specs plus pricing on February 28 during AMD’s announcement event.
