And more than likely a crippled 256-bit memory interface.
Honestly, I expected nothing and I am still disappointed with how NVIDIA’s rumored RTX 4000 series “Super” refresh is turning out by the minute, with the supposed RTX 4080 Super now speculated to contain the full AD103 chip of which the regular RTX 4080 utilizes, with 10,240 CUDA cores in total as opposed to the 9,728 feature on the RTX 4080.
Kopite is negligent on the apparent memory configuration of the aforementioned RTX 4080 Super, but it’s unlikely to change much from the 16GB of GDDR6X memory found on the standard 4080, it was first thought that the RTX 4080 Super was to contain a severely cut down AD102 core boasting 20GB of GDDR6X memory with a slightly cucked memory bus as opposed to the 4090.
This is SUPER disappointing, with a mere 5.2% bump in actual core count, and more than likely with some slightly faster memory on the same 256-bit memory interface found on the RTX 4080, there’s going to be various examples of games where the $999 Radeon RX 7900 XTX still manages to pull ahead of this alleged SUPER in terms of rasterized performance.
NVIDIA has rather shallow bridges to fill in terms of performance and pricing between the likes of the RTX 4070, RTX 4070 Ti and the RTX 4080 which are priced at $499, $799 and $1199 respectively but most of which can be found for lower than their MSRPs quite frequently due to diminishing demand and inventory left on store shelves.
Which is why NVIDIA intends to mediate between these three graphics cards with “SUPER” refresh alternatives, with Kopite, a notorious insider regarding GeForce hardware revealing alleged specifications for the RTX 4070 Super and RTX 4070 Ti Super.
There’s a bit of confusion surrounding the latter due to its name, blending the “Ti” and “SUPER,” monikers which certainly opens the floodgates of NVIDIA quite literally spamming the market with a new product segment seemingly between every 10% in performance variance.
The predicted configuration for the RTX 4070 Ti Super is much more exciting than the “4080 Super”, it’s suggested that it could feature 8,448 CUDA cores with a mixture of two die variants which would either be the AD103-275 or a notably reduced AD102-175, it could very well contain a mixture of both as a means of discarding AD102 chips that failed to pass QA.
The largest performance gaps inside NVIDIA’s current lineup are between the GeForce RTX 4090 and the 4080 followed by the GeForce RTX 4080 and the 4070 Ti, with the rough performance gap between both SKUs coming in at around ~22% at 1440p resolution, or in the case of the RTX 4080 and the RTX 4090 the gap widens to roughly 30% at 4K resolution.
A potential RTX 4070 Ti Super with 10% more CUDA cores along with increased memory bandwidth would most certainly bridge the gap between the RTX 4070 Ti and RTX 4080 by a substantial margin, while more than likely a supposed RTX 4070 Ti wouldn’t leverage an equal 16GB of VRAM found on the 4080.
If priced correctly by effectively taking the $799 price tag of the current 4070 Ti, this Super refresh would undoubtedly provide better value than the RTX 4080 which would otherwise cost hundreds of dollars more, though with a lackluster 12GB of VRAM most likely on the cards I’d still prefer purchasing the Radeon RX 7900 XT.
The RTX 4070 SUPER exhibits a conservative shift in specifications from the RTX 4070 Ti, featuring a lowered CUDA core count of 7168. This version, akin to the previously mentioned SKU, might also offer two GPU variations, presenting either the AD104-350 or the AD103-175.
Versus the traditional RTX 4070, this supposed SUPER revision would come featuring a near 22% increase in core count, but most importantly it’ll come packing 48 MB of L2 cache as opposed to the 38 MB found on the standard RTX 4070, out of all the rumored specifications, the RTX 4070 Super is by far the most substantial increase.
Again, there’s no real reason why NVIDIA would be pushing to incorporate a greater quantity of VRAM on these SUPER renditions versus the likes of the 4080, 4070 Ti and 4070 which come in 16GB and 12GB flavors respectively, though undeniably it would be great if NVIDIA were to simply give consumers more memory capacity their real issues versus AMD’s RDNA 3 offerings come in the form of outright gaming performance and value.
NVIDIA desperately needs to bridge the performance gap between the RTX 4070 and 4070 Ti, as AMD currently offers a substantially superior product (for rasterization) in the form of the Radeon RX 7800 XT which undercuts NVIDIA’s offering by a whole hundred dollars.
It’s a shame that the aforementioned RTX 4080 Super will turn out to be a miniscule performance increase, because the more exciting of this SUPER revision is undoubtedly the RTX 4070 Super and the RTX 4070 Ti Super which has a very stupid name. But of course, NVIDIA isn’t a charity and more than likely they’re not going to release these graphics cards and simply slash the prices of the RTX 4070, 4070 Ti and the 4080.
NVIDIA cements themselves as being a “premium” brand, which is something that does not exist in the tech sector. With the RTX 4070 and 4070 Ti priced at $499 and $799 respectively, NVIDIA can just as easily get away with pricing the RTX 4070 Super at $600, it also wouldn’t be surprising to see a $949 RTX 4070 Ti Super either.
I certainly hope I’m wrong regarding such prices, but NVIDIA isn’t a company that really likes to give consumers the “edge”, and with the RTX 4080 Super coming with a full AD103 core, it has no purpose even existing at all. With such a marginal lackluster performance increase there is absolutely no redeeming qualities to an alleged AD103 fitted RTX 4080 Super, unless NVIDIA simply dissolve the RTX 4080 outright and have this “Super” take its place at $1199 there is absolutely nothing worthwhile going its way.
It’s pure narcissism and greed, because NVIDIA would absolutely love to charge consumers $1299 for that additional 5-7% performance increase over the standard RTX 4080, whatever it takes to market a product that’s “better” than the competitors flagship across a handful of cherry picked scenarios while doing the absolute least they can.
Clearly we were expecting too much from NVIDIA with the thought of a further cut-down AD102 SKU with 20GB of VRAM and a nerfed memory bus, because this Super lineup is shaping up to be a massive disappointment in the making.
