Carzy Zarx’s PC Gaming Emporium - Catch Up on All the Latest PC Gaming Related News

[JEMC]

Captain_Yuri said:

Rumour: Coreteks: AMD FidelityFX Super Resolution (FSR) to launch in June

https://videocardz.com/newz/coreteks-amd-fidelityfx-super-resolution-fsr-to-launch-in-june

The sooner the better!

That seems to be too soon to be true. Maybe they'll reveal how it will work, how they'll implement it and give us some examples to show how great (according to them) it works.

Captain_Yuri said:

Alright, I am gonna do some predictions based on industry trends and rumours. Lets see how many I get wrong!

#1 - RDNA 3 will get a massive improvement in Ray Tracing

I think this is obvious but because of how many games are getting RT support already and if the rumour of Navi 31 being 160 CUs is true, the improvements might even be 2x faster in RT compared to RDNA 2.

#2 - Ray Tracing capable GPU will start being mandatory for AAA games starting 2023

We already know that 4A's new engine has no Raster fallback according to DF and that all future games using this engine will require a RT GPU as minimum and they won't be the only ones for AAA titles. I think that most companies are waiting until the cross gen period is over and the new consoles are readily available to purchase which I believe will be in 2023. Then you will have 40 million if not more consoles with RT capable GPUs as well as a large number of PC GPUs going back to a 2060. RT capable doesn't mean it needs to be great at RT, just that it can do hardware RT.

#3 - Zen 4 will Roflstomp Alder Lake S and Raptor Lake in Desktop and Datacenter

While we do have a new Ceo that has a lot of promise, engineering takes time. I think 10nm will overall be a flop based on the rumours we have seen and how meh Tiger Lake is while Intel tries to fast track their 7nm.

#4 - Super Resolution will either be a bandage or delayed until RDNA 3

While Super Resolution will help RDNA 2 GPUs greatly, I don't think it will be much different than the TAA Upscaling/checkboard options we have today. The main benefit will be that it will be easier to implement as right now, every game engine has a different up-scaling solution while Super Resolution will be a single solution across multiple game engines and platforms. I think the true competitor to DLSS will be coming with RDNA 3.

#5 - Lovelace will beat the initial launch of RDNA 3 GPUs but AMD will save the zomg version of RDNA 3 until Lovelace launches

If Lovelace is delayed for months and RDNA 3 is going to release in Q1/Q2 in 2022, then it makes zero sense for Nvidia to release a product that's slower than RDNA 3. But knowing this, AMD won't release their top their GPU in the initial launch either. I think they will release a GPU that's 30% faster than Ampere in Q1/Q2 and then release their true top of the line GPU when Nvidia launches Lovelace.

#1 - RDNA3 needs to bring a big improvement in RT performance, because the first iteration has been a bit meh (roughly on par with Nvidia's first iteration, but too far from Ampere to be able to compete). The extra 2x RT cores that will come from the extra CUs will surely improve the end performance, but they also need to improve the performance of the RT cores themselves. I hope they'll be able to do it, but the result may still end behind Nvidia's next GPUs.

#2 - There will always be devs that try to push boundaries and use that as a selling point, but imho 2023 is too early for that to be mainstream. That said, some studios may try the approach of 4Games: launch a "regular" game with RT elements and later on launch an updated RT-only version.

#3 - Too soon to tell. We barely have any benchmark regarding Intel's new chips, and we have nothing regarding Zen4. The only thing we can guess from now is that AMD will still be the more energy efficient solution.

The other have too many variables to make any opinion on them.