NVIDIA GeForce RTX 3090 Founders Edition Review: Between Value and Decadence – When Price is Not Everything

Game performance

In the following result overviews I have on the one hand the average FPS of all cards and in all games cumulated anyway this then still percentage calculated and compared. I have tested 10 games in total and Ultra-HD is still going strong, but we must never forget that this is not a cherry picking, there are games in the selection in every situation that can easily break through the fair weather front.

If you look at the percentages now, we end up with an average of 10.5% for gaming. In professional applications such as Arnold or Blender, it was a good 20% and thus also the theoretically possible maximum increase. More would not be possible anyway, unless the RAM would not come into play.

In Summary: this card is a real giant, especially at higher resolutions, because even if the lead over the GeForce RTX 3080 isn’t always as high as dreamed, it’s always enough to reach the top position in playability. Right stop of many quality controllers included. Especially when the games of the GeForce RTX 3090 and the new architecture are on the line, the mail really goes off, which one must admit without envy, whereby the actual gain is not visible in pure FPS numbers.

If you have looked at the page with the variances, you will quickly understand that the image is much better because it is softer.  The FPS or percentiles are still much too coarse intervals to be able to reproduce this very subjective impression well. A blind test with 3 perons has completely confirmed my impression, because there is nothing better than a lot of memory, at most even more memory. Seen in this light, the RTX 3080 with 10 GB is more like Cinderella, who later has to make herself look more like Cinderella with 10 GB if she wants to get on the prince’s roller.

But the customer always has something to complain about anyway (which is good by the way and keeps the suppliers on their toes) and NVIDIA keeps all options open in return to be able to top a possible Navi2x card with 16 GB memory expansion with 20 GB later. And does anyone still remember the mysterious SKU20 between the GeForce RTX 3080 and RTX 3090? If AMD doesn’t screw it up again this time, this SKU20 is sure to become a tie-break in pixel tennis. We’ll see.

For a long time I have been wrestling with myself, which is probably the most important thing in this test. I have also tested 8K resolutions, but due to the lack of current practical relevance, I put this part on the back burner. If anyone can find someone who has a spare 8K TV, I’ll be happy to do so, if only because I’m also very interested in 8K-DLSS. But that’s like sucking on an ice cream that you’ve only printed out on a laser printer before.

Application Performance and Studio

A total of 17 benchmarks and many hours later, you can almost raise more enthusiasm in this area than it was already the case with the GeForce RTX 3080. Whenever FP32 calculations are involved and you think you are getting a little closer to the retirement point every time you render, the RTX 3090 is a spoilsport per excellence. Partly, even a Quadro-RTX 6000 with 24 GB memory is declassed to such an extent that one has to ask oneself which movie one is watching at the moment. Speaking of movies, the NvEnc has been making life easier since Turing, but if all the postprocessing filters get a good run, the forced pause at the editing table is not even enough for a nice coffee break and you are even faster than with the GeForce RTX 3080.

And because we were just talking about moving images, the new amp cards support AV1 natively in the decoder (Intel’s Xe will probably do the same). But why AV1? The codec is up to 50% more efficient than H.264, which means you only need half the Internet bandwidth to transmit the same quality – including HDR and even 10-bit encoding. This could be up to 8K – theoretically. Unfortunately, even among cable providers, the practice of Germany looks rather meagre. But the beginnings have been made and somehow the will counts, even if I miss the 8K-TV.

In summary: Yes, you can work really well with the GeForce RTX 3090 and you can find it even more beautiful than the RTX 3080. The 3D-community will probably strike with relish and hyperventilate as long as the yellow bag plays along, because no matter if Octane, Arnold or Blender – here really the post office goes off. With OptiX the GeForce RTX 3090 mutates to a cannibal, which acts partly more than twice as fast as the Quadro RTX 6000 or the Titan RTX. So Quadro-Punishment is already included. Which of course also applies to video editing and all those workstation or CAD applications that do not require certified and optimized drivers for sprinting.

GeForce RTX 3090 Founders Edition

The in-house Founders Edition does everything right this time and is clearly better than its predecessors, even than the GeForce RTX 3080. Optically and haptically the part is first class, but unfortunately also a bit heavy. Well, baroque curves have their price. The one with the 12-pin connector is now a running gag, but tolerable. The cooler lives up to its name and it is comforting to know that it can do both: cool AND quiet. Of course it’s not.

This is exactly why this card will find its friends and buyers despite the price, I am sure. The implementation of the FE is smart, as the RTX 3090 has been equipped with just about everything that the paver racks can handle. At the latest the circuit board will certainly make it a sought-after collector’s item. The size of the map is in its size, the rest is to get over. Buying? It’s up to each person to decide for themselves. Conversion to water cooling? It could be done, but it’s almost too bad.

Conclusion and final remark

This time, there is no such thing as the all-inclusive purchase tip, because the target group is almost too special for such an emotionally loaded watering can recommendation. Just buy it? Doesn’t fit. Just think about it? Yeah, that’s more like it. However, I think the chip of the GeForce RTX 3090 and the implementation as a Founders Edition is great, even if the crucial question of efficiency still arises. Whereby it is quickly put into perspective with the absolute wattage figures, if you only look at the page with the efficiency evaluation. Whom a Quadro RTX was too expensive up to now or who has always considered the Titan RTX to be the goal of dreams, gets a much faster and even much cheaper alternative here. This way around you can (and should) also see it.

The increase in value of the RTX 3090 in relation to the RTX 3080 for the only gamer is, up to the memory extension, to be rather neglected and one understands also, why many critics will never pay the double price for 10 to 15% more gaming performance. Because I wouldn’t either. Only this is then exactly the target group for the circulated RTX 3080 (Ti) with double memory expansion. Their price should increase visibly in comparison to the 10 GB variant, but still be significantly below that of a GeForce RTX 3090. This is not defamatory or fraudulent, but simply follows the laws of the market. A top dog always costs a little more than pure scaling, logic and reason would allow.

And the non-gamer or the not-only-gamer? The added value can be seen above all in the productive area, whether workstation or creation. Studio is the new GeForce RTX wonderland away from the Triple A games, and the Quadros can slowly return to the professional corner of certified specialty programs. What AMD started back then with the Vega Frontier Edition and unfortunately didn’t continue (why not?), NVIDIA has long since taken up and consistently perfected. The market has changed and studio is no longer an exotic phrase. Then even those from about 1500 Euro can survive without a headache tablet again.