Report – Nvidia bans factory overclocking on select Turing/RTX silicon

Report - Nvidia bans factory overclocking on select Truring/RTX silicon

Report – Nvidia bans factory overclocking on select Turing/RTX silicon

As with all new graphics architectures, there remain several mysteries surrounding Nvidia’s Turing architecture that are unsolved, but now it appears as if one of these secrets has finally been solved. 

One of these mysteries is the “-A” problem. Why is there a “-A” variant of all of Nvidia’s RTX gaming GPU models? For example, there is a TU102-400, an RTX 2080 Ti, and there is also a TU102-400-A, which is also an RTX 2080 Ti. What does the -A mean, and should consumers be concerned about it? 

While creating their next version of GPU-Z, Techpowerup took note of this phenomenon and have asked “industry sources” if they can provide an answer to their “what is -A?” question. Usually, Nvidia only creates a single ASIC code for a product, whether it be Reference, Founders Edition, custom or factory overclocked. These codes are typically used to differentiate Workstation products from their gaming-grade counterparts. 

The answer to this question is simple, -A products are intended to custom models that AIB partners can release with Factory overclocks and non-A models are designed for reference graphics cards with MSRP pricing. They are the same chips, but they are separated by binning and are sold to Nvidia’s AIB partners with different price points.    

When it comes to -400/non-A models, Nvidia’s AIB partners must ship products that are limited, without any factory overclocks over Nvidia’s reference specifications. Thankfully this limitation is not forwarded onto end-users, who are allowed to overclock their graphics cards as they see fit.  

-A graphics card models are allowed to be sold with factory overclocks, allowing AIBs to create graphics cards that offer equal or greater performance than Nvidia’s Founders Edition models. These graphics cards are likely to be using higher binned silicon and offer increased overclocking potential. These chips are likely more expensive than Nvidia’s -400/non-A silicon, making it difficult or impossible for these SKUs to match Nvidia’s MSRP. 

  

Report - Nvidia bans factory overclocking on select Turing/RTX silicon

(Image from Techpowerup)

  
Assuming that these findings are correct, the mystery of the -A has been solved, showcasing the strict limitations that Nvidia has placed on their AIB partners, and what they have done to ensure to their AIB partners create low-cost MSRP graphics card models. 

Should end-users be concerned about this? No, this is no reason for alarm, these restrictions do not limit overclocking from the perspective on users, though it does mean that reference RTX 20 series graphics cards are likely to use lower-binned silicon. 

You can join the discussion on Nvidia banning factory overclocks on select Turing/RTX silicon on the OC3D Forums.Â