Nvidia’s RTX 4070 GPU may not be as powerful as we’ve heard in the rumor mill recently, with Team Green apparently still not finishing any major specs – and we’re talking CUDA Core count here, as well as video memory count.
The latest update from prolific Twitter-based leaker Kopite7kimi gives us two different possible specs for the RTX 4070 graphics card (apply your own taste and don’t be stingy on that front either).
Check out the two possible specifications of RTX 4070: PG141-SKU340/341, 7680FP32, 12G 21Gbps GDDR6X, total card power 285W, TSE <11000. PG141-SKU336/337, 7168FP32, 10G 21Gbps GDDR6X, total card power 250W, TSE <10000. I found that I made a typo earlier, accidentally wrote 341 as 331.29 August 2022
As you can see from the tweet, one of them is a more powerful configuration of the RTX 4070 that offers 7,680 CUDA cores and 12 GB GDDR6X VRAM (at 21 Gbps), with a power consumption of 285 W.
The other option Nvidia could work with would reduce the number of CUDA cores to 7,168 and the VRAM to 10GB, with power consumption set at a less demanding 250W.
The difference between those two possible RTX 4070 GPUs could be that the formerly beefier card would be about 10% faster, at least according to Kopite7kimi’s predictions for a 3DMark (Time Spy Extreme benchmark) result.
Analysis: Nvidia engaged in more than one careful balancing act?
Rumor lovers will no doubt remember that the lesser of the specs mentioned above is an older configuration that aired on the grapevine some time ago. Kopite7kimi has since changed the tone of the RTX 4070, claiming it would be like the more powerful configuration mentioned above, with 12GB of VRAM and 7,680 cores; the only difference between that previous leak and this new one is that the power consumption has apparently been adjusted to 285W (instead of 300W).
So that’s good news if Nvidia is indeed running with the stronger capability in the above tweet, but a disappointing step back to previous speculation if the lesser spec turns out to be the route taken. There have long been concerns about the RTX 4070 with just 10GB of VRAM, and many feel it does the card way too short – so there was a lot of relief when the rumor mill chatter turned into a 12GB kit. But now we have cast a cloud of doubt on that.
Is Nvidia still in doubt whether to save 12 GB for the RTX 4070 Ti? Indeed, looking at the lesser spec option Kopite7kimi offers is about considering the overall alignment of the Lovelace GPU range, and making sure the RTX 4070 slots are in and logical compared to the RTX 4080. We after all, we’ve been wondering if that rumored more powerful RTX 4070 is going to take sales away from the 4080 (and its bigger profit margin), and presumably Nvidia will want to be careful there.
All this assumes, of course, that Kopite7kimi is right, and perhaps the leaker is wrong. But assuming this isn’t the case, and that Nvidia is still effectively putting the Lovelace puzzle pieces in place as to where the various RTX 4000 models sit relative to each other in terms of specs, this seems to indicate that the release of the RTX 4070 is not within reach.
Another possibility that has come out of the joke is that Nvidia could launch the RTX 4090 in October, and instead of the RTX 4080 and 4070 that follow in a month (or two), they could be delayed until 2023. shift of excess RTX 3000 GPU stock, which will otherwise not be sold when next-generation cards are out or imminent).
If Team Green has still not decided on the exact specs of the RTX 4070 as suggested, it seems to indicate that the release time may be further away. Board partners, of course, need a lot of time and knowledge of the official configuration for these products to start designing their third-party graphics cards and consider cooling solutions and so on.
So there are two ways in which this latest rumor is a little disappointing – it suggests that the less powerful theoretical spec could still be used for the RTX 4070, and that we may not see this Lovelace graphics card for a while.