The rumor mill is abuzz as soon as once more as Nvidia’s next-gen Ada Lovelace GPUs draw nearer. Notable tipster @kopite7kimi is providing new particulars within the Twittersphere on what Nvidia has in retailer for players within the fall. This time round they’re shedding some new gentle on high three GPUs within the product stack, which can possible be the primary to launch.
In a collection of recent tweets, @kopite7kimi details the specs for these upcoming GPUs. As has been reported beforehand, the flagship die will likely be AD102. Curiously, the top-shelf “Lovelace” GPU will likely be outfitted with 24GB of GDDR6X operating at 21Gb/s. That is the very same configuration as the present RTX 3090 Ti. It must be famous that the earlier technology Turing-based Titan RTX additionally had 24GB of GDDR6 reminiscence. Plainly for now, that is so far as Nvidia desires to take reminiscence allocation on its high-end SKUs. There have been earlier rumors that AD102 may come geared up with GDDR7 reminiscence operating at 24Gb/s, however these could be incorrect. What’s seemingly firmer at this stage is the GPU’s Whole Board Energy (TBP). It’s as soon as once more listed as 600W, which looks like a given since Nvidia desires to push this card to absolutely the most stage of efficiency. It may need to so as to fend off AMD for this spherical.
Transferring down the stack we come to the RTX 4080. Though the Ampere-based RTX 3080 used a cut-down model of the flagship die, GA102, that received’t be the case for the 4080. As an alternative of utilizing the massive AD102 die, or the smaller 104 model like with Turing and former generations, it’s getting its personal slice of silicon. It’ll be an AD103 chip, slotting in between the massive chip on the 4090 and the smaller chip on the 4070.
This card is rumored to supply 16GB of GDDR6, which will likely be a welcome improve for many who thought the current-gen’s 10/12GB was inadequate. Nonetheless, it seems Nvidia will likely be downgrading the reminiscence from GDDR6X to slower GDDR6. For comparability, the GDDR6X within the present RTX 3080 runs at 19Gb/s, whereas GDDR6 can hit about 14Gb/s. It’s potential Nvidia will compensate for this with a wider reminiscence bus, and even additional L2 cache. This GPU will supposedly have the identical TBP because the top-shelf GA102 die, which is 450W within the RTX 3090 Ti.
Lastly, we come to the RTX 4070. This “tween” GPU could possibly be the candy spot by way of efficiency and energy consumption for the Lovelace structure. It’s notable that for this technology it’ll be getting the AD104 die, which is bigger than what’s been reserved for this class beforehand. With Turing and former generations, the 104 die was for the x80 GPU, and the smaller 106 die was for the x70 playing cards. Along with a much bigger die it’s additionally getting an enormous improve within the reminiscence division. It’ll be outfitted (reportedly) with 12GB of GDDR6 operating at 18Gb/s. That’s roughly the identical because the boosted RTX 3070 Ti, which options 19Gb/s GDDRX reminiscence. Each 3070 playing cards solely have 8GB of reminiscence nevertheless. Whole board energy is a quaint 300W. That used to appear like loads, way back. Now it’s virtually lovely. The leaker notes that neither the RTX 4080 nor the 4070 have begun the testing section. This does indicate the 4090 is already at that stage.
This period of GPUs guarantees to be fairly attention-grabbing, as each Nvidia and AMD are rumored to be unleashing every thing they’ve of their arsenals. We all know that doesn’t sound any completely different from earlier eras since these two firms have been battling for ages. Nonetheless, Nvidia going as much as 600W is unquestionably a brand new factor. It signifies it anticipates some heavy resistance from the AMD and its chiplet-based RDNA3 GPUs.