The announcement of three new cards today, the NVIDIA GeForce RTX 2080 Super, RTX 2070 Super, and RTX 2060 Super is about as unexpected as ice melting in Florida. NVIDIA, along with the rest of the world, watch AMD announce Navi, its competitor’s next-gen 7nm GPU. Here is the problem: AMD announced against then-current NVIDIA GPUs, but left a gap to when it would ship. Instead, NVIDIA waited. Today, it announced its response, updated cards. This is exactly what we would expect.
NVIDIA has had around ten months since the NVIDIA GeForce RTX 2080 Ti 2080 and 2070 Launched. In that time, yields improved, and new memory generations came out. NVIDIA simply took advantage of higher chip yields and updated a portion of its lineup to bring performance one price tier downmarket. There was so little competition in the market that, other than the RTX 2080 Ti, NVIDIA has been able to hold pricing. With AMD finally getting a competitive part, NVIDIA can use volume advantages to lower pricing. This is about as strategically easy to understand as watching young children play checkers.
NVIDIA GeForce RTX 2080 Super, RTX 2070 Super, and RTX 2060 Super
Instead of going through these one-by-one, here is the table that shows the NVIDIA GeForce RTX 2080 Super, RTX 2070 Super, and RTX 2060 Super along with the rest of the GeForce RTX family:
Perhaps the big news is that the NVIDIA GeForce RTX 2060 moves up to an 8GB part. That is something that STH noted at the original NVIDIA GeForce RTX 2060 6GB launch and our ASUS Turbo-RTX2060-6G Blower-Style Performance Review that we wanted to see.
We still think that the GeForce RTX 2080 Ti will be safe with 11GB of memory for those who are doing compute. For those who were buying at the RTX 2080 level, the price of entry is going down considerably. We think GPU compute users will enjoy more memory bandwidth.
We also wanted to point out here that if you are power constrained, the NVIDIA GeForce RTX 2080 Ti offers significantly more compute, memory, and bandwidth for only 10W more than the NVIDIA GeForce RTX 2080 Super.
Final Words
What is next? We fully expect NVIDIA will take its newer faster memory and do a refresh of the NVIDIA GeForce RTX 2080 Ti and NVIDIA Titan RTX lines. Until AMD launches a proper halo product, this is not necessary. At the same time, it will start to feel some pressure to update since the lower-priced cards have bumped specs up. Simply adding faster GDDR6 memory and slightly raising clock speeds could yield a NVIDIA GeForce RTX 2080 Ti Super and Super Titan RTX.
An interesting aspect of these new “Super” products is that they all follow the same price/performance ratio of the existing RTX products. But going by several reviews so far like at German Computerbase the increase in performance is achieved with a even stronger increase in power consumption, making these new cards less power efficient that the already existing cards.
Radeon VII is caught flat-footed and in an odd place now, relative performance charting in between the 2070S and 2080 FE/OE, with the now 2080S a step above at the same price. Vega 64 owners might find themselves with little incentive to upgrade to either VII or the Navi 5700XT as it only offers 1.15~X performance over the previous part. Pretty poor execution on AMD’s part, IMO, especially in light of their successes with Ryzen/TR/Epyc as of late. I think they need to get their house in order, and quickly.
Patrick, “what is next” you ask: HotChips 2019, August 19: https://www.hotchips.org/program/
1:45 PM – 2:45 PM: Keynote 1: “Delivering the Future of High-Performance Computing with System, Software and Silicon Co-Optimization” by Dr. Lisa Su, CEO, AMD
Most of the rest of the programs seem to involve Machine Learning, maybe AMD will say something about a future Instinct GPU.