Nvidia stops distinguishing Turing GPUs for overclocked video cards

Nvidia would have stopped supplying specific Turing GPUs for video cards that are overclocked by manufacturers. Presumably the 12nm production process of TSMC has been improved so that all dies are now suitable for overclocking.

According to Germany’s Tom’s Hardware website, Nvidia has started production of the TU104-410 and TU106-410, as successors to the TU104-400 and TU106-400. In all cases it concerns the GPUs of the GeForce RTX 2070 and 2080 and the new versions would be supplied for video cards with and without overclock. At the end of this month, video cards GPUs of the new revision would be released.

When the GeForce RTX cards appeared last year, Nvidia supplied specific GPUs to manufacturers who wanted to release a video card with overclock. That was about the 400A variants. The cheapest versions of video cards, without overclocking from the factory, had the regular TU10x-400 GPU. Customers could overclock the GPU themselves, but Nvidia used the best chips for the 400A variants.

Nvidia has its GPUs for RTX video cards made on a 12nm process from TSMC. Presumably, the yields of the process have now been improved, so that all chips are of the same quality and no distinction is made anymore. According to the sources of Tom’s Hardware, it only concerns the GPUs for the RTX 2070 and 2080. Whether the same will happen with the Turing GPUs of other RTX video cards is not yet known.