Nvidia’s new version of the RTX 4070 has been officially announced, and the graphics card is pretty much just as previously rumored, carrying slightly slower video memory – but the good news is that there’s seemingly no meaningful impact on performance.
Wccftech was quick to spot the release of the new variant of the 4070, which was revealed by Team Green in a lowkey manner as part of the release notes for its latest graphics driver.
That Game Ready Driver brings in support for the GeForce RTX 4070 with GDDR6 video RAM (as opposed to the GDDR6X which is the memory employed on the graphics card currently). As previously rumored, these are 20Gbps GDDR6 modules and this lowers the memory bandwidth to 480GB/s (from 504GB/s).
That sounds bad, but apparently Nvidia has tuned the memory timings and latencies in order to pep up the GDDR6 modules a touch, and the end result is that the new and old versions of the RTX 4070 effectively offer the same level of performance.
Nvidia informs us: “To improve supply and availability to meet strong demand, we’re introducing the GeForce RTX 4070 with extra fast GDDR6 memory. All of the other specs remain the same. It offers similar performance in games and applications.”
Wccftech has some benchmarks which back up this assertion and show that the performance of the new RTX 4070 equates to the existing GDDR6X model across a clutch of games and apps. The relative performance of both graphics cards is within 1% in all cases except for one app (Chaos V-Ray, where the GDDR6 model is actually 3% faster – but that can be classed as an outlier).
As expected, the new RTX 4070 is pitched at the same price, which is why Nvidia had to ensure performance was pretty much identical. Team Green notes that its card making partners will be selling the new graphics card from September, so the week after next (there won’t be a Founders Edition from Nvidia).
Analysis: Is it worth buying?
Is this new version of the RTX 4070 worth buying, then? The short answer is – yes. Or at least it’s every bit as worthwhile as the older GDDR6X-toting RTX 4070, which will still probably represent the majority of models out there on shelves for quite some time (it takes a good while for existing stock to sell through, of course).
The good news here is that PC gamers feared that the GDDR6 variant would be a bit slower, but as we can see, from both Nvidia’s claims and the purported benchmarks aired above, there’s no meaningful difference at all between the two spins on the RTX 4070.
The not-so-great news is that we were also hoping that maybe the use of GDDR6 might make the new RTX 4070 better in the power-efficiency stakes, but the TGP (power usage) of the graphics card hasn’t changed – it remains at 200W as before. Incidentally, there’s no difference in the chip used in either version (AD104-250 is in both – whereas previous speculation pointed to a slightly different AD104-251 spin).
In short, none of this will help the RTX 4070 reclaim its spot on our roundup of the best GPUs, from which it was ousted by the RTX 4070 Super. However, the vanilla RTX 4070 with GDDR6 still remains a solid buy, and one other consideration you might want to take into account here is the rumor that it’s possible we’ll see price tags for the RTX 4070 creeping up.
That price hike is purportedly bound up in the reason for the switch to GDDR6, because as a previous report claimed, the faster GDDR6X VRAM is thinner on the ground stock-wise – and that’s something Nvidia hints at when it mentions the need to “improve supply and availability” for the RTX 4070. Given this possible price increase, if you’re mulling an RTX 4070, you might want to move sooner rather than later – and the new variant is every bit as good as the current 4070.
