RTX 30 vs RTX 40: As I’ve had enough of saying for the past few weeks, ignore the price drops in the world of PC hardware! Forget the price drop on the side of Ryzen processors, and above all, do exactly the same thing, even more vigorously, with the current generation of graphics cards.
Yes, I understand, it’s been almost 2 years without being able to buy a graphics card at decent prices. But don’t be fooled! Prices have not fallen because of the kindness of manufacturers, or because production has improved (the crisis is still very active), prices have fallen, because demand has fallen extremely important (especially in the mining sector), and of course, because we are about to undergo a giant renovation of the product lines.
But do you know what the difference is? Are we going to take it back with scary prices?
RTX 30 vs RTX 40: What’s the performance difference? and in the price?
So you like “leaks”? It’s just that if you happen to like a good leak in the world of mobile phones, in the world of hardware, we already know almost everything almost 1 year before the launch. As is now the case with the NVIDIA GeForce RTX 40! Where we already have a list of specs that shouldn’t be too far from the truth, as well as an expected level of performance. But let’s go in parts.
How many RTX 40s will we see at launch?
At least 4 different models should be announced. However, they will not all arrive at the same time.
The first graphics card to be released, probably already in July, should be the RTX 4090 (24 GB), based on the AD102-350 GPU. Subsequently, we will welcome the most powerful RTX 4090 Ti (48 GB), itself based on the AD102-300. A little later we will have the best friends of the RTX 4080 (16 GB) wallet based on the AD103-300, and of course, finally, the RTX 4070 (12 GB) based on the AD104-400.
What are the “numbers”? (Technical specifications)
The most powerful RTX 4090 Ti will feature the fully unlocked AD102 GPU, which in turn means 18432 CUDA cores. The slightly weaker RTX 4090 sits at 16128 cores. Next we have the RTX 4080 with 10752 cores and the RTX 4070 with 7680 CUDA cores.
The number of processing cores does not tell the whole story, because with the new architecture, the improvement in performance will obviously be evident across the range. After all, the RTX 4070 should match or even surpass the RTX 3090 Ti in both raster and ray tracing performance.
To get an idea of the performance boost, you need to be aware of what the current generation is capable of delivering.
Well, the RTX 4090 Ti should be able to exceed 100 TFLOPs in performance, with the RTX 4090 remaining at around 90 TFLOPs. In direct comparison to the current RTX 30, we’re talking about a near tripling of performance, as the RTX 3090 Ti offers something like 40 TFLOPs, while the original RTX 3090 is around 36 TFLOPs.
Price and availability
It is almost undeniable at this stage of the championship. New graphics cards will be more expensive compared to their launch prices in 2020. I don’t think we’re going to relive the craziness of 2020 and 2021 as the market has stabilized and mining is on the way out.
In addition to all that, we also have the Wildcard that is Intel, which is also about to put its spoon in the gigantic soup that is the graphics card market, and of course, AMD, which apparently will be able to impress with its RDNA 3 architecture, this time with a good stock of products.
However, the production crisis remains a reality, as does the rise in inflation in Europe, and even throughout the world. So you can put a price 10-20% above the RTX 30’s recommended launch price of the RTX 30, and pray a lot that the thing stays there. (This is the estimate from multiple leaks for AMD and NVIDIA.)
As for the launch, all the graphics cards described here should be available by the end of the year. The RTX 4090 will be the first, already in July. Later we will have the RTX 4090 Ti, which will obviously serve as TITAN, most likely in August, and finally we will have the RTX 4080 followed by the RTX 4070, in September~October.