AMD Navi die area data will shatter NVIDIA's self-confidence to the ground

At AMD's morning presentation, CEO Lisa Su demonstrated from the stage the 7nm Navi-architecture (RDNA) GPU that will power the Radeon RX 5700 family of graphics cards to be unveiled in July. Taking clear photos at this distance was problematic, but selected members of the press were allowed to hold this GPU in their hands. Alas, not all of them are so concerned about the size that they constantly carry accurate measurement tools with them, and censors from AMD could not approve such manipulations with samples of products that have not yet been presented.

AMD Navi die area data will shatter NVIDIA's self-confidence to the ground

And yet, the representatives of the site AnandTech managed to get a rough idea of ​​the Navi GPU die area. According to them, it does not exceed 275 mm2. Even if we consider that this is a very rough calculation, the benefit of using the 7nm process technology in the execution of TSMC is obvious here. As previously stated in the presentation, the first generation of RDNA GPUs improves the performance-to-power ratio by 50% compared to the GCN architecture. In addition, the 7-nm manufacturing process makes it possible to manufacture a fairly compact crystal.

AMD Navi die area data will shatter NVIDIA's self-confidence to the ground

AMD compared a conventional Radeon RX 5000 series graphics card with an NVIDIA GeForce RTX 2070 graphics card at the morning presentation, and in Strange Brigade, the Navi-architecture product was at least 10% faster. While there is no exact data on the cost of new AMD video cards, they have much more “margin for price maneuvering”, because the TU106 graphics processor underlying the NVIDIA product is produced using 12nm technology, and its die area is approximately equal to 445 mm2. Roughly speaking, AMD has a 62% area advantage.

AMD Navi die area data will shatter NVIDIA's self-confidence to the ground

Of course, without knowing the nuances of the contractual relations between AMD and NVIDIA with TSMC, it is difficult to draw categorical conclusions about the cost of 7nm graphics processors of the first and 12nm graphics processors of the second. However, NVIDIA founder Jen-Hsun Huang's recent quarterly conference was remembered for his arrogant remarks that there was no need to switch to 7nm manufacturing technology. He said that NVIDIA's existing offerings are unmatched in terms of performance and power consumption, although they are produced on 12nm technology, and there is no point in comparing them with competitor's 7nm products. Let's wait until July and see how the rhetoric of the head of NVIDIA will change after the release of independent reviews of the latest AMD video cards ...



Source: 3dnews.ru

Add a comment