We reported previously on the possibility that Nvidia might wheel out a revised version of its popular RTX 4070 graphics chipset with cheaper and slower GDDR6 memory. Now it's been officially announced by Nvidia.
As Nick explained, the reason why this is happening at all is thought to be a short supply of the GDDR6X graphics memory used by the existing RTX 4070. Indeed, Nvidia's announcement implies just that.
«To improve supply and availability to meet strong demand, we’re introducing the GeForce RTX 4070 with extra fast GDDR6 memory,» Nvidia said in a conspicuously short entry towards the bottom of a blog post (via Tom's Hardware) that was mostly about the release of a new Game Ready driver with optimizations for Black Myth: Wukong and Star Wars Outlaws.
Immediately, you're probably wondering what impact the shift to ostensibly cheaper and lower-spec GDDR6 will have. After all, if plain old GDDR6 is just fine for the RTX 4070, wouldn't Nvidia have specified that memory from the get-go?
Unfortunately, aside from styling the GDDR6 memory in question as «extra fast», Nvidia isn't providing much by way of details. For now, it is only saying that when it comes to the revised RTX 4070 with GDDR6, «all of the other specs remain the same. It offers similar performance in games and applications.»
The OG Founders Edition RTX 4070 and all third-party models so far have come with 12 GB of Micron GDDR6X clocked at 21 Gbps. By way of example, Samsung sells 20 Gbps GDDR6. So, if Nvidia has gone for that particular brand and spec of GDDR6, the total memory bandwidth will drop from 504 to 480 GB/s.
That's a fall of just 5%. Would that hit frame rates? Probably a little bit but typically frame rates aren't directly proportional to memory bandwidth. So, the likely impact would be in the very low single-digit percentiles. Not enough that you'd ever feel it, in other words.
Keep up to date with the most important stories and the best deals, as picked by the PC Gamer team.
All that
Read more on pcgamer.com