In a rather controversial (and rather hollow) claim, NVIDIA’s CEO, Jensen Huang has claimed that the company’s GPUs are “easily twice as energy efficiency compared to the second-best architecture (AMD) on the market, regardless of the number in the transistor side”. Speaking to analysts at the Q2 earnings report after the company’s Data Center revenue outpaced gaming, Huang was quick to downplay the impact of the process node on the final product.
There’s a vast difference between our architecture and the second best architecture and the rest of the architectures. The difference is incredible. We are easily twice the energy efficiency, all the time, irrespective of the number in the transistor side. And so it must be more complicated than that.NVIDIA CEO, Jensen Huang
This claim may have been correct with the older GCN based Vega graphics cards, but the newer Navi GPUs based on the RDNA architecture are as efficient as NVIDIA’s Turing products. The Radeon RX 5700 and 5700 XT level with the competing GeForce parts, namely the RTX 2060 Super and 2070 Super in terms of the power consumption, all the while offering similar performance at markedly lower prices.
- NVIDIA RTX 2060 Super vs AMD RX 5700: 1080p|1440p Performance Benchmarks
- NVIDIA GTX 1650 Super vs AMD RX 5500 XT: Gaming Performance
Perhaps the reason why the NVIDIA CEO said this is because the upcoming consumer-class Ampere GPUs will be based on Samsung’s 8nm node (enhanced 10nm) which is way less efficient compared to TSMC’s latest 7nm process. We don’t expect AMD’s Navi 2x GPUs to be much different from the RTX 30 series parts in terms of power consumption though.
That in itself speaks volumes about the efficiency of NVIDIA’s GPU architecture. However, the claim that the process technology is unimportant (or less important) when it comes to determining the power efficiency of a GPU/CPU is completely false. If you want proof, you have to look no further than AMD’s 1st Gen Navi graphics cards.