Last month we saw the presentation and testing of the most powerful model of the new generation desktop graphics card from Nvidia – the GeForce RTX 4090. While the jump in performance was huge between generations, the price was quite interesting, at least compared to the GeForce RTX 3000 generation in terms of price and performance.
Nvidia is now releasing another and this year’s last model of the new generation – the GeForce RTX 4080. Originally there were to be two variants of this model, but due to the completely different performance and great criticism, Nvidia will most likely launch this version as the GeForce RTX 4070 Ti.
GeForce RTX 4080: a heavily trimmed 4090
Compared to the highest GeForce RTX 4090, the lower model in the form of the GeForce RTX 4080 has a significantly reduced number of cores and thus also performance. The 4nm AD103-300 chip used has only 9,728 cuda cores, while the GeForce RTX 4090 has 16,384. This also results in a much lower number of cores specializing in ray tracing acceleration – 76 and only 304 tensor cores. The chip works at a similar boost frequency of up to 2.51 GHz, but everything will still differ depending on the overclocking of individual models from the factory.
|Model||GeForce RTX 4090||GeForce RTX 4080 16 GB|
|Production||4N (TSMC)||4N (TSMC)|
|Boost frequency||2520 MHz||2505 MHz|
|Memory capacity||24 GB (GDDR6X)||16 GB (GDDR6X)|
|E. memory frequency||21 GHz||22.4 GHz|
|Memory interface||384 bit||256 bits|
|Memory throughput||1008 GB/s||717 GB/s|
|TGP Consumption (Max)||450W (660W)||320W (516W)|
|Price (recommended)||47,990 CZK||CZK 35,990|
The card has the standard 16 GB of fast GDDR6X memory with a 256-bit interface, but despite the slightly higher frequency, the throughput is only 717 GB, while the 4090 has 1,008 GB/s. Nvidia used the problematic 16-pin 12VHPWR connector again, but this time only three eight-pins need to be connected to the adapter.
A lower configuration also means a lower TGP, which is a relatively acceptable 320 W, while it is 450 W for the GeForce RTX 4090, with the fact that these are not peak consumptions.
Tests from around the world
If you have your own favorite game or component configuration, you are sure to find exactly the comparison you are interested in in tests from all over the world. A large number of editors received the card, so there are tests on websites and as video reviews. And how is the card doing?
The GeForce RTX 4080 is on average 25% to 30% slower in games than the GeForce RTX 4090. That’s quite a lot, and the significantly reduced price corresponds to it. However, this is still a performance that makes a mockery of even the best of the last generation – the GeForce RTX 3090 Ti.
The GeForce RTX 4080 card is on average 15% to 20% more powerful in games than the GeForce RTX 3090 Ti, which at the same time had almost double the starting price. But if we were to focus on individual games, the differences can be even greater on both sides. It goes without saying that the card is ideal for playing in 4K, in lower resolutions it is unnecessarily limited by the processor. However, if you need an increased number of frames per second, for example in QHD or Full HD resolution, the card can push out admirable performances here as well. When you add support for technologies such as DLSS 2.0, DLSS 3.0 and latency optimization, it is definitely a solution even for very demanding gamers.
As is common today, with overclocking using a standard cooler, you cannot get to very high frequencies or increase the performance significantly. Expect to squeeze an additional 5% or so from the card. The price will, of course, be higher needs and temperatures.
The GeForce RTX 4080 is aimed at the target group of gamers for whom the GeForce 4090 is unnecessary or too expensive. Yes, it has brutal power, but you have to be able to use it or it’s pointless.
Nvidia has set the recommended price at 36,000 crowns, which, from the perspective of the distant past, sounds like a very high price, but when compared to the last generation and performance, it is a very interesting value. It is questionable if the prices of the cards will ever reach a slightly more normal level again, because this price is still among the hi-end, which most players cannot afford. Entire gaming PCs are built for this price, and as you can see from the Steam statistics, the average gamer’s PC configuration is completely out of this level.
The most popular graphics cards
There is much more hope for other lower models, which can be expected only next year and which will bring prices much lower, but the performance can still be expected as with the hi-end of the last generation.