~8.5% if it remains on AD103 and the information supplied in the article is correct.

4090 prices have been going up and are theorized to continue rising in cost. They start at $1700 and go up to $3200 now.

$1200 for an AD103 die is absurd, in my opinion (though I know I’m not alone). Honestly, I don’t think it would be unfair for the 4080 to drop to $950 and the theorized 4080 Super to drop in at $1100.

Though, I personally believe that $950 for an AD103 die is still too much. The 3080 was on GA102 die and had a MSRP of only $800. Die foe die comparison, the 4080 would fall between the 3070 ti ($600) and 3080 ($800)… $700. Adjust for margins and inflation and costs, etc., sure $800 would be fair.

The 4090 can cost whatever Nvidia wants it to cost; it’s unmatched in performance and is the best out there. $2000? The consumers who want this would still pay. But Nvidia made it clear last gen; 3080 is top of the line for gaming and the 3090 was more in line with the Titan class cards of generations prior. So, if they wanna add a little more margin on top of that $800 I mentioned earlier; sure, go for it. $850 still seems fair enough. But $950, $1100, and $1200 are all way too much.

But thats just my two cents.

If the unit offers a slight performance upgrade versus its predecessor, it will be good enough to further solidify its position on the market of desktop discrete graphics boards.

I mean, sure. But 4080 cards are currently selling for $1100-1600 (in the US). Unless this performance lift comes in at the same price or at a price reduction, idk how this would “further solidify it’s position”. According to the Steam Hardware Survey, only 0.53% of are using the RTX 4080.

Notable cards the 4080 is behind (in no particular order):

• RTX 1080

• RTX 2080 Super

• RTX 3080

• RTX 3090

• RTX 4060

• RTX 4060 Ti

• RTX 4090

• GTX 960

• RX 570

Also, if this ends up being a full-fledged AD103, how big of a performance difference can we really expect? 8%? It’s like the 3070 and 3070 Ti all over again. That was a 4.34% increase in cores, which results in ~7% increased performance. This would be a 5.2% increase in cores for (if we scale linearly off of the previous set of numbers) a ~8.5% increase in performance.

I understand it cannot scale linearly, as the architecture is entirely different, but it gives us a good ballpark for expectations… and those expectations will need to be low.

Edit: To further expand, I just wanted to input that this difference would be just enough to allow the 4080 Super or 4080 Ti (whatever this ends up being) to just surpass the 7900XTX in rasterized performance. If I had to guess, that’s all Nvidia cares about for marketing this. However, this all comes down value and the cost of the product. If they increase the MSRP by $100, that’s less than a 1% scaling in performance to cost ratio, from the 4080.

And what exactly does it do? It “optimizes” how?