As we often see with Nvidia’s GPUs, the dies for the GeForce RTX 2080 and GeForce RTX 2070 were binned into two discrete GPU types. A GeForce RTX 2070 is either equipped with a TU106-400-A1 or a TU106-400A-A1 GPU, for example, with the best silicon branded with the ‘A’ type GPU and therefore permitted for factory overclocks. Conversely, the non-A chips can usually have a few cores disabled and are also capable of lower clock speeds overall, resulting in a performance disparity between chips in the same GPU family.
It has now been confirmed that Nvidia will stop producing the non-A chips for the GeForce RTX 2070 and RTX 2080, meaning just the higher quality ‘A’ dies will be manufactured.
All GeForce RTX 2070 and RTX 2080 GPUs from AIB partners going forward should be using the faster chips now, meaning more reliable overclocking capabilities. It could take a little while for the effects of this change to ripple through the supply lines, but it does mean AIB partners should be able to offer higher, more reliable overlocks on their custom GeForce RTX graphics cards.
To the end user, this shouldn’t mean more expensive graphics cards either. This is much the same silicon as is already used for the RTX 2070 and RTX 2080; Nvidia has just matured the fabrication and can now more reliably deliver the higher-specced components. Essentially, this can somewhat absolve the silicon lottery component which can factor into any graphics card purchase.
One final point to consider is that Nvidia’s yield has evidently improved on the TU106 and TU104 GPUs. In theory, this means the price of Nvidia’s high-end graphics cards could actually drop due to the more reliable supply line. We’ll have to take a wait and see approach on that one though, as savings may not necessarily be passed on to the consumer.