We’ve been hearing a lot of rumors about Nvidia’s upcoming RTX 4000 series, though most of them seem to revolve around the RTX 4070, its Ti variant, and the RTX 4090. The latest claim to come from a regular leaker involves the high-end-but-not-quite-flagship RTX 4080, which sounds like it will be an absolute beast of a card.
As usual, the new Ada Lovelace rumor comes from the Twitter account of prolific and accurate (when it came to Ampere) leaker Kopite7kimi (via VideoCardz). They posted what are claimed to be updated specs for the RTX 4080, with the significant changes relating to the memory speed and the card’s TDP.
It had been expected that the RTX 4070, 4080, and 4090 would use 21 Gbps GDDR6X modules—they were mentioned in this month’s update on Micron’s website—but the RTX 4080 is now thought to feature 23 Gbps speeds. Assuming the rumor is accurate, the RTX 4090 will likely feature the same 23 Gbps GDDR6X.
The next revised spec is the RTX 4080’s Total Board Power (TBP). This is one area that leakers seem to change every other week. It was initially said to consume a monstrous 450W, but that was recently changed to 320W. According to Kopite7kimi, Nvidia has increased the RTX 4080’s TBP again, albeit by just 20W, leaving the new number at 340W. For comparison, the RTX 3080 Ti Founders Edition has a TBP of 350 watts.
The rest of the RTX 4080 specs remain unchanged: an AD103 GPU, 9,728 CUDA cores, and 16GB of memory across a 256-bit bus.
Another part of the August update on Micron’s website was a listing for 24 Gbps GDDR6X. It could end up in what’s thought to be a new Titan card that was previously rumored to pack 48GB of VRAM, 18,176 CUDA cores, and an 800 TDP.
Nvidia experiments with the specs of its upcoming cards as the launch dates draw closer, which is why they seem to change so often; we might not see the final, locked-in specifications until the launch is on the doorstep. The good news is that day is now just a couple of months away.