Each NVIDIA generation has its performance flagship that can easily blow the wallet of most users. In the previous generation, this was the Titan X, which for over 2,000 dollars was more interesting for professionals than for gamers.
The classification of the GeForce RTX 3090 in the number sequence of the new Ampere cards actually suggests that the card is also interesting for gamers. But a price of well over 1,500 dollars and a rather small increase in performance compared to the RTX 3080 clearly speak against it, as we can demonstrate using the example of a ZOTAC GeForce RTX 3090 Trinity.
With a current price of well over 1,500 dollars, the RTX 3090 is currently the flagship of the new amp generation from NVIDIA. This makes it about twice as expensive as the RTX 3080, but the plus in gaming performance is significantly lower than the price associated. We’ll explain why later in our review of the ZOTAC GeForce RTX 3090 Trinity. Similar to the Titan X of the previous generation, the RTX 3090 is less intended for gamers than for professional users.
But let’s start with the external features of the present model. By and large, the RTX 3090 is pretty much identical to the recently tested RTX 3080 Trinity from ZOTAC. With dimensions of 317.8mm x 120.7mm x 58mm, you definitely need a spacious case and enough space for the wide 2.5-slot design. A powerful power supply of at least 750W should also be in the computer, because with a TDP of 350W the card slurps a lot of electricity, our measurement even went up to 369W under full load. The power supply runs via two 8-pin connections, adapters from 6- to 8-pin are included.
Like the smaller version, the RTX 3090 also uses three fans with ZOTAC’s own IceStorm 2.0 cooling system, which keeps the card very happy when it is fully loaded. We recorded a maximum temperature of 81 degrees, which is a good value. There are no differences in the connections either. An HDMI 2.1 port with HDCP 2.3 is also available, as well as three DisplayPort 1.4a connections, which enables resolutions up to 8K. Spectra 2.0 RGB Lighting is also on board, as is the ability to control the card using the FireStorm utility.
The differences between the two models are under the hood and if you give a statement, it quickly becomes clear that no miracles can be expected from the RTX 3090 compared to the cheaper RTX 3080 in terms of gaming performance. The same GA102 CPU works in both cards, with a boost clock of 1,695 MHz, which is even slightly lower than the 1,710 MHz of the RTX 3080. The maximum clock rate climbed to 2,025 MHz and remained minimally below the RTX 3080 there, too there are others, however, and in part clearly.
The CUDA cores are merciless 10,496 (RTX 3080: 8,704). If the RTX 3080 works with just 10GB GDDR6X toil a full 24GB in the RTX 3080, and that with a wider 384-bit interface, which enables the bandwidth to be increased to 936 Gbps compared to the RTX 3080. There are 82 instead of 68 ray tracing cores, 328 instead of 272 for the tensor cores. If there are little differences in the actual CPU, the peripherals are much more generously equipped, but this has only a minor effect on the gaming performance depending on the game and benchmark.
This makes it immediately clear which sticking point players can expect from the RTX 3090. The card is a lot more powerful in terms of features, but games hardly benefit from it due to the identical chip. The card is therefore particularly interesting for professionals who need a lot of cores and a lot of memory, i.e. pretty much everyone who is involved in media production or development. For the average gamer, the additional financial expense compared to an RTX 3080 just doesn’t pay off. 100 percent more price for 10 percent more gaming performance makes little sense.
This is quickly evident in the benchmarks that focus on games. The refresh rates of most in-game benchmarks only spit out slightly higher frame rates, if at all, because the increased memory, bandwidth and cores cannot be used at all. Bottlenecks on the part of the CPU, which can no longer fail to feed the graphics card, sometimes even result in a few frames less than with the RTX 3080. Even with more neutral benchmarks, the increase in performance remains manageable and mostly in a range of around ten percent. Only in the RTX benchmark Port Royale is a slightly larger jump due to the higher number of RT cores.
Finally, it should be noted that every player, even with high demands on resolution and frame rate, can currently build the currently most powerful card in their computer with the RTX 3080 in terms of price and performance, while the RTX 3090 is not a sensible alternative for pure gamers. Like the Titan X, the RTX 3090 is currently seen as a flagship for professional users who also use a high-end CPU. Our eight-core from the R7-3700X series from AMD was sometimes a bit overwhelmed when it came to feeding the card.