Nvidia’s RTX 40-series graphics card generation seems to be one of U-turns. First we had the unlaunching of its RTX 4080 12GB card and now we’re seeing rumours of a 16GB version of its upcoming RTX 4060 Ti. Which is all kinds of odd.
The rumours have all sprung from a single set of tweets from Zed Wang (via Videocardz), a regular GPU leakerer, who has mentioned release dates of RTX 40-series cards in the past which have more or less lined up.
Their latest lists the RTX 4060 Ti 8GB, RTX 4060 Ti 16GB, and RTX 4060 8GB cards citing the first launching at the end of May, with the other two cards following in July. The 8GB versions of the RTX 4060/Ti themselves aren’t a surprise; we had been expecting that VRAM capacity from the get-go. But that 16GB level certainly is.
We are pretty certain about many of the RTX 4060 Ti cards’ specifications by now, with an ADA106 GPU sporting 4,352 CUDA cores, a bunch of extra cache, and a 128-bit memory bus. That last one is the key one here if we’re talking about video memory capacities, because it limits the amount available to be attached to the card.
With only 16Gb (2GB) modules available for GDDR6 as a maximum, you’re going to be severely restricted by how much memory you can physically fit on a board. And with a 128-bit aggregated memory bus you can only work in multiples of four in order to maintain balance across each of the 32-bit discrete memory bus built into the GPU.
That means you can have 4GB, 8GB, and 16GB capacities. Nvidia was never going to stick 4GB on there, and we had expected 16GB to be too rich for the green team to attach to an effectively mid-range GPU. But apparently not. At least not now.
Were it not for the current furore over the level of VRAM demands from modern games, particularly the slew of poorly optimised PC ports we’ve had to endure recently, I doubt Nvidia would have thought of releasing a 16GB RTX 4060 Ti. At least not just a couple of months after an initial 8GB version had launched.
But the noise around the rumoured GPU specs and the perilous state of PC game releases, has pushed VRAM capacities to the forefront of gamers’ minds. AMD has certainly been keen to point out that it has been slapping 16GB of GDDR6 onto cards costing as low as $499, and that all looks to have had an impact on Nvidia’s plans for these new GPUs.
It’s rare to see a company such as Nvidia making such changes seemingly so close to launch, but it’s gratifying to see it listening. Or at least looking like it is. The ‘unlaunching’ was unprecedented, and this itself is a surprise. We don’t know what either the 8GB or 16GB cards are going to cost, though the latest rumours are for a $399 price point for the 8GB RTX 4060 Ti coming at the end of May.
That’s a cut from the $499 we were expecting, but does give Nvidia that window in which to drop a more expensive 16GB version of the card and still make bank. Theoretically then it gives the green team the chance to have its cake and eat it—it can still ship a cheaper 8GB card at a price point that sounds good in marketing taglines, and a more expensive one that both matches the expected original $499 price and satiates demand for a 16GB card.
The issue, however, is that such a 16GB card will undoubtedly cannibalise the sales of the initial RTX 4060 Ti 8GB launching at the end of May, especially given that the effectively superior card will be coming a scant two months later at most. Even if the actual GPU performance or memory bandwidth won’t differ between the two options, if a specific game is hoovering up as much VRAM as it can get then you’ll see stuttering on the lower spec card you won’t on the 16GB version.
And that possibility alone will likely be enough for most people to want the security blanket of more VRAM. Which could then result in a glut of the 8GB cards sitting in retail as there won’t be much wiggle room for price cuts to encourage demand with a non-Ti RTX 4060 coming in just below it.
It really is interesting times in the mid-range of this year’s GPUs, and how it all shakes out is going to be fascinating to watch.