AMD is committed to providing the same great performance enhancement for its next generation of graphics cards, at least according to Rick Bergman, AMD’s vice-president, who interviewed him on The Street clarifies his company’s vision for the coming year. .
AMD will compete with 5nm of TSMC and aspire to win the crown
The interview has no space other than useful information, in which Bergman makes it very clear that RDNA 3 will be an extension of the current RDNA 2 and other performance improvements:
AMD is targeting its RDNA 3 GPUs, which will use the advanced production process to deliver watt performance improvements similar to the 50% overtime upgrades brought by its RDNA 2 GPUs, as well as its future Infinity Cache Technology programs used by RDNA 2 GPUs.
Bergman: “Let’s go back and talk about the benefits of both of them. So why did we point to working per watt cruelly on our RDNA 2 GPUs? «.
“This is very important in many ways, because when electricity is very high, as we have seen from our competitors, suddenly our potential users should buy large electrical appliances and very high cooling solutions. And in addition, it greatly enhances the BOM of the card. This is a desktop view. And always, that means the sale price goes up or the GPU costs have to go down. “
RDNA 3 will maintain the same performance enhancement per watt as RDNA 2
“So there is actually a lot of performance… if you can significantly improve your performance per watt. On the laptop side, that is very obvious, because you are still very limited space, you can bring more functionality to that platform and without some cooling solutions… Focus on that in RDNA 2. It’s good Focus on RDNA 3 so we have the same commitment.
“Infinity Cache is also somewhat related to that, to a degree. If you have been in the graphics world for a long time, you will find that there is a good connection between memory bandwidth and performance. And often the way it is done is to increase the memory speed and increase the bus [de memoria] to increase efficiency. Unfortunately, both factors increase energy consumption. ‘
AMD goes so far as to claim that they will increase the efficiency of their performance with NVIDIA today, so we are facing a new generation of GPUs that promise to remove more performance without increasing usage. Can NVIDIA challenge this or will it fall behind by having a more efficient node than its competitor even though it has better internal structure?
