NVIDIA and AMD’s next-generation GeForce RTX and Radeon RX graphics cards, respectively, will be offering double-digit performance gains, according to a rumour by the reliable leakster, Kopite7kimi.
Kopite7kimi has had a flawless record as a leakster, we will still advise treating this info with a grain of salt. In his tweets, he reports that he has received tips from his sources but since these products are still more than a year away from launch and performance estimates could change dramatically once we get close to that point of time. He mentions performance estimates and some specifications for both next-generation AMD and NVIDIA GPUs.
Starting with the details, it is reported that either TSMC’s or Samsung’s design will be used for the next-gen GPUs that are likely to be based on a 5nm process node. The key architectures that have been discussed include NVIDIA’s Ada Lovelace (AD102), Hopper (GH202), and RDNA 3 (Navi 3X). Based on the details, Ada Lovelace seems like a small transition architecturally from Ampere, but it will bring a massive boost in performance. And Hopper will be the revolution that NVIDIA has been waiting for, both in terms of performance (raster/ray-tracing) and design (power/package).
Kopite7kimit believes that we are likely to get up to 100 TFLOPs of FP32 compute power in a single graphics card, which sounds a bit too optimistic. However, given that GPUs today rely on efficient and faster compute models, 100 TFLOPs doesn’t sound that extreme, and we could see a sparsity model for consumer-specific chips to increase their computing performance by a huge margin.
When it comes to performance numbers, according to WCCFTECH, the NVIDIA Ampere GA102 architecture is used as a baseline and when compared to that, “the NVIDIA Ada Lovelace (AD102) chip is said to offer a 2.2x increase in perf while AMD’s RDNA 3 (Navi 3X) is said to offer over 2.5x perf increase.”
There have been several rumours stating that the upcoming RDNA 3 GPUs will outperform whatever the Santa Clara, California-based GPU-maker, NVIDIA, has to offer in terms of rasterization performance. It looks like Dr Lisa Su’s AMD, under its Radeon RX graphics card lineup, will take the lead by offering the first MCM-powered GPUs. But at the same time, NVIDIA is expected to quickly transition to its Hopper MCM GPU lineup, which will offer over 3x performance boost over Ampere GPUs. Following is the expected performance increase:
Memory bandwidth is stated as one of the major bottlenecks for both AMD and NVIDIA next-gen GPUs. GDDR6X won’t be able to cut it so we might see a return to HBM or a newer GDDR memory standard that’s available around 2022-2023. It is likely that, to bypass bandwidth bottlenecks and conserving cost, GPU makers will be investing more in close-to-GPU caches such as the Infinity Cache featured on RDNA 2 graphics card. Power consumption is also expected to go dramatically up.
AMD & NVIDIA GPU Roadmaps:
NVIDIA Lovelace AD102 Rumored GPU Specifications
As for AMD’s Navi 31 GPU, even KittyYYuko has leaked information about it offering more than 2.5x performance. As for Hopper, we were informed that it was to be delayed in favor of Ada Lovelace a while ago but once again, “it is possible we can see the AD10* and GH20* GPUs to be introduced close to one another however NVIDIA’s own roadmap cites that their ‘Ampere Next Next’ GPU won’t be here till 2024.”
