Yesterday I claimed that Nvidia fans and customers should stop buying a new RTX GPU until Nvidia had actually demonstrated how much of a performance improvement these GPUs would deliver in today's shipping titles. The company, at that time, simply did not show that the new cards would provide enough additional performance to make advance reservation a wise decision.
Nvidia has just released a new set of 4K results claiming to show a gap of 50 percent between Nvidia GTX 1080 and Turing-based RTX 2080. This is the theoretical type of data we need to draw some conclusions about relative performance abandonment between RTX and GTX families. Unfortunately, the only thing that these data do is strengthen the argument that Nvidia does not bring much to the table for some but wealthy enthusiasts. Let's look at why.
The white line represents GTX 1080's nominal performance, the darker green is the RTX 2080 alone and the bright green is the RTX 2080 when you use Nvidia's new anti-aliasing method, DLSS. We'll ignore DLSS here, not because there's something wrong with it – I've been tough for antialiasing for decades and have picked my GPU before based on how good anti-aliasing The mechanisms worked – but because it represents a feature that only one card has and which requires software support from the developer to enable. Our focus is therefore strictly on the gray and dark green bars – and the prices of the two GPUs are compared. The price is where all this falls apart.
Price Equity (or as well as practically accomplished) is one of the most important ways to help end users determine if a component represents a good deal. GTX 1080 currently available for as little as $ 450, between 56 percent and 64 percent the price of a new RTX 2080 (depending on whether you go Founders Edition). By course the RTX 2080 is faster. The relevant question is, does it mean that increased performance represents a better deal?
The results depend on which starting number you choose since these cards have a number of prize points. The current best case for the GTX 1080 – RTX 2080 comparison is that the RTX 2080 is 1.55x more expensive and delivers ~ 1.5x more performance. If you need to buy FE or a similar OEM card to see these results, you will pay at least 1.77x more money for 1.5x performance. Yes, the final product is faster – but you also pay significantly more for it. And "More expensive hardware provides better performance at a steady poor price / performance ratio" is not the type of headline that makes people busy. There is also absolutely no help for people who may swing $ 450- $ 550 on a new card but can not afford to go up to $ 700- $ 800.
After Nvidia's price increases, the RTX 2080 is no longer appropriate comparison point for the GTX 1080. So what happens when we compare the RTX 2080 with its actual competitor, $ 700 GTX 1080 Ti? It only happens that [H] ardOCP recently published an article that compared the GTX 1080 to 1080 Ti. So let's look at the dataset and estimate how much 1080 Ti would cut into Nvidia's results.
[H] ardOCP test Crysis 3, Tomb Raider, GTA V, Witcher 3, Fallout 4, Rise of the Tomb Raider, Doom, Deus Ex Mankind Divided, Battlefield 1, Sniper Elite 4, Mass Effect Andromeda, Kingdom Come Deliverance and the Far Cry 5. The GTX 1080 Ti is an average of 1.27x faster than the GTX 1080. If we assume that these averages hold the ecosystem, we can expect the RTX 2080 to be about 1.23 times faster than GTX 1080 Ti at the same price point . There may be some shifts depending on resolution and detail level differences, but we expect them to favor 1080 Ti if anything. Any bottle neck that specifically reduced the GTX 1080 would give 1080 Ti's extra horsepower more space to shine.
There is nothing wrong with a 1.2x performance boost, but Nvidia knows it's not the kind of thing that makes players talk. It's certainly not the kind of improvement that makes anyone hurry and replace a 1080 Ti they've just bought within the last 18 months. So, instead of recognizing this point, they beat it by comparing the GTX 1080 with a much more expensive GPU that it would normally not compete with. When you set the competition correctly, up to half of Nvidia's claimed performance improvement may disappear.
Another point to remember is that Nvidia claims that these data represent GTX 1080, RTX 2080 and RTX 2080 with DLSS enabled. This means that the other GPUs both use a kind of performance-effective AA solution to begin with. It is the only way 2080's performance would improve by using AI for antialiasing, as opposed to the method previously used. There is nothing wrong with using AA when analyzing GPUs; Almost all of our GPU testing is performed with antialiasing enabled. But readers should be aware that using AA can change relative performance between two solutions, and since we have no idea what AA methods were used in the games Nvidia lists, we do not know how to enable it. affect the performance between Turing and Pascal. The size of the increase from the baseline 2080 to 2080 + DLSS suggests that the originally deployed solution was quite strict.
Final point: None of these tests appear to affect the beam tracking side of Nvidia's new capabilities. The only feature tested and displayed here is Deep Learning Super-Sampling. We do not yet know what kind of performance Turing takes into play with RTX-enabled versus disabled people.
So far, the debut for the RTX 2080 is mostly like Nvidia trying to justify the price increases as it relaxes on its video card. We have graded the launch price of Nvidia's flagship models, dating back to Fermi. It gives some interesting reading:
For four generations, (GTX 295 – GTX 680), Nvidia kept the same price of $ 500 for its flagship card. The GTX 780 increased to $ 649 for launch, but fell lower for six months thanks to AMD's Hawaii-based R9 290 and R9 290X. Maxwell thanked a modest $ 50 increase on the top price, but nothing wrong. However, beginning in 2016, Nvidia started aggressively charging more, especially if you bought a Founders card. If you want to purchase a RTX 2080 card in 2018, as you purchased a GTX 980 card in 2014, Nvidia wants an additional $ 150 – $ 250 for the privilege. It is far from what we used to see just a few years ago, when Nvidia brought up achievements at the same price point from year to year, even at the top of the market. AMD's difficulty in competing on the top of the GPU market is reflected in Nvidia's pricing. As costs increase, Intel's 2020 GPUs can not come soon enough.
Based on the results we've seen so far, the Nvidia RTX 2080 looks like delivering between 1.15x – 1.3x extra performance compared to GTX 1080 Ti at the same price point in regular titles that do not exploit its features. Claims of 50 percent positive improvements do not tolerate control given the price difference between the two solutions. As always, all of this analysis should be considered preliminary and speculative based on generally available information, but not final hardware. It is possible that others, as yet uncovered improvements to the GPU kernel, could affect the final analysis.
Read Now: Do not Buy Ray-Traced Hype About Nvidia RTX 2080, Nvidia Announces List of Games With RTX Support, and How Nvidia's RTX Real-Time Ray Tracing Works