That reported a hack to Nvidia allegedly leaked the codenames of several next-generation GPUs that have now been leaked to the press. The most suitable for us would be Lovelace GeForce GPUs, of which there are six, but Hopper and Blackwell server GPUs are also listed here. This appears to have been due to the initial leak of some stolen documents supplied by Videocardz.
The green team is alleged to be demanding a ransom for the Ethereum hash rate limiter attached to its latest graphics card after the Lapsus$ hacker group stole about 1TB of sensitive data.
The group is demanding that Nvidia disable the limiter, with the dubious claim that it has “decided to help the mining and gaming community” or release the contents of a “large folder” it apparently stole. For its part, all Nvidia has told us is that it is “investigating the incident” but can’t share anymore at the moment.
I’m guessing it’s supposed to check first what was taken and how damaging a total leak can be. This initial disclosure only revealed some GPU codenames and nomenclature that we either already knew or could have guessed. This also means that there is no guarantee that images posted online are actually straight from Jen-Sun’s repository.
But the leak suggests there will be six GPUs based on the AD100 architecture, named after Ada Lovelace. They pair well with Ampere generation graphics chips, and the promised AD102, AD103, AD104, AD106, AD107, and AD10B chips are designed to form the next generation of GeForce GPUs.
None of this really tells us anything of note about new graphics cards, where the chips will be made, how big they are, or even when we can reasonably expect to see them “out of stock” at retail.
Tips & Tricks
How to buy a video card: tips for buying a video card in the desert silicon landscape of 2022
The latest rumors are that the RTX 40 series will arrive sometime in September, which coincides exactly with the launch of the RTX 30 series in 2020. Lovelace GPUs should not be architecturally very different from Ampere in terms of basic design but are expected to be manufactured using the TSMC 5N process, which promises improved efficiency and higher performance per watt.
Although there are rumors that the total graphics power (TGP) is between 450 and 850 watts for top-end AD102 chips. Which will negate any gain inefficiency. For reference, the RTX 3090 has a 350W TGP listed if you’re interested. Yes, relatively eep, right?
We might end up getting more information if the group does have and decides to release more details about the new chips, but that’s not the only reason we’d prefer Nvidia not to give in to the group’s demands. Besides not condoning stealing, I also think that the excuse of helping the gaming community sounds pretty ridiculous when the last thing we want is for GeForce GPUs to be more effective for miners. However, I want to hear what the new “falcon” is that Lapsus$ claims everyone knows about. The original was a RISC-V system processor for the GPU, but it was due to be recently discontinued. So maybe it comes back as something new.