With Ada Lovelace, Nvidia takes a huge step forward. The tested Zotac GeForce RTX 4090 Trinity consistently achieved new maximum values in the speed tests and did not allow itself any more energy than the predecessor RTX 3090.
However, Nvidia is also setting new standards in terms of price, which are forcing graphics card manufacturers to make immensely high calculations: Zotac estimated it At the time of testing, the GeForce RTX 4090 Trinity cost a whopping 2,500 euros – that’s too much of a good thing for a graphics card.
nvidia’s RTX 3000 series graphics chips were codenamed “Ampere” – named after the French physicist André-Marie Ampère, who also gave the name to the unit of measurement for electrical current.
Ampere is followed by Nvidia’s RTX 4000 series – named after the British mathematician Ada Lovelace. The top graphics chip RTX 4090 proved in the test: Ada has it all! For the tests, COMPUTER BILD had one of the first finished RTX 4000 graphics cards in the test laboratory – the Zotac GeForce RTX 4090 Trinity.
RTX 4090 in the test: everything is fine
Nvidia’s RTX 4000 series graphics chips are completely new developments. It starts with the manufacturing process. In the middle of the development phase, Nvidia decided to have the 4000 chips manufactured by the contract manufacturer TSMC with 4 nanometer fine structures instead of Samsung with 8 nanometers.
Why? The same structure width as the predecessor would have increased the power consumption in the direction of 600 watts and higher with the planned increase in performance.
With 4 nanometer fine structures, Nvidia can now accommodate many more circuits (transistors) on a chip area of around 600 mm2, which is about the same size – up to 76 billion instead of “only” 28 billion. The transistors are therefore much closer together, and the switching paths are shorter, which brings a further speed advantage.
24 gigabytes of graphics memory GDDR6X
Nvidia distributes the 76 billion transistors to 16,384 CUDA computing units in the top model RTX 4090. The old top model, the RTX 3090, had to make do with 10,496 cores.
The Cuda cores work with a base clock of 2.23 gigahertz, with boost technology it goes up to 2.5 gigahertz for a short time.
In comparison to the RTX 3090, everything remains the same with the graphics memory: The RTX 4090 can also temporarily swap out calculated data to a graphics memory that is very large at 24 gigabytes – of the fast GDDR6X type. And the data exchange between the graphics chip and memory continues to take place via 384 lines (384-bit interface).
Improved ray tracing cores
Nvidia also gives the RTX 4090 revised ray tracing cores for accelerated calculation of effects. In adapted games, for example, rapidly rotating spotlights project fine light effects with different intensities into the game environment.
At the same time, the chips adjust the strength of the shadows cast by people and objects in real-time. So that the performance does not collapse, the new RTX 4090 has 128 separate RT cores of version 3.0, and the RTX 3090 had 84 RT cores of version 2.0
DLSS 3.0: More images
In addition, the RTX 4090 gets 512 Tensor Core processing units in version 4.0. Among other things, they are responsible for Deep Learning Super Sampling (DLSS) of variant 3.0.
While DLSS in version 2.0 still rendered a game scene pixel by pixel and then extrapolated to the resolution used, DLSS 3.0 can now generate entire individual images (frames) in adapted games itself.
An example: When playing in 4K, the Tensor cores with DLSS 3.0 calculate an additional image between two frames of the respective scene. To do this, the RTX 4090 analyzes the entire content of the existing frames – including movements, particles, reflections, shadows, and lighting.
An AI (artificial intelligence) creates the additional image from the data determined.
Full of speed!
What does that bring? Significantly more speed. The additional frames also ensure better playback and fewer details are lost on top of that.
The speed advantage over Nvidia’s RTX 3090 was particularly noticeable in 4K games – the RTX 4090 worked 75 percent faster (see comparison above). But the DLSS 3.0 technology also shows how efficiently the RTX 4090 works.
In the test, the Zotac worked twice as fast in this test point, in some cases even three times as fast as the predecessor RTX 3090 with DLSS 2.0. Disadvantage: The game manufacturers have to adapt their titles with DLSS 3.0. And there aren’t that many of them yet.
After all, well-known titles such as “Cyberpunk 2077”, “Microsoft Flight Simulator” and “F1 2022” should receive the technology via an update for the beginning, the game
“A Plague Tale: Requiem” will be available for sale on October 18, 2022, with DLSS 3.0. Good: Under full load, it consumed a lot of energy with around 450 watts, but no more than its predecessor – despite the immensely higher performance.
RTX 4090: Big chunk
Important: If you actually have the necessary change, you should first take stock of the PC. In addition to an 850-watt power supply, sufficient space is required, after all, the tested Zotac GeForce RTX 4090 Trinity is 36 centimeters long, 13 centimeters high, and 7 centimeters wide.
The case must therefore be sufficiently deep, the mainboard must not be fully equipped – the RTX 4090 occupies three adjacent PCI Express slots with its thickness. With a weight of almost 2 kilograms, graphics card support is absolutely necessary for horizontal installation.
RTX 4090 in the test: conclusion
With Ada Lovelace, Nvidia takes a huge step forward. The tested Zotac GeForce RTX 4090 Trinity consistently achieved new maximum values in the speed tests and did not allow itself any more energy than the predecessor RTX 3090.
However, Nvidia is also setting new standards in terms of price, which are forcing graphics card manufacturers to make immensely high calculations: Zotac estimated it At the time of testing, the GeForce RTX 4090 Trinity cost a whopping 2,500 euros – that’s too much of a good thing for a graphics card.
