And this footage was captured from a PC running FSR in performance mode, quality mode would look even better
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
And this footage was captured from a PC running FSR in performance mode, quality mode would look even better
Associate
Potentially 4080 perf priced below £500? AMD have an opportunity here.
But we all know what they say about AMD and opportunities...
Also had an interesting random theory... what if AMD's delays are software/driver-related?
E.g. the regular raster/RT perf is sorted, but FSR4 and new AI frame-gen stuff isn't ready yet?
Cos they probably would want to show any new tech/software when they 'officially' announce the cards.
It's less of an impact if they show regular raster numbers before showing fancy FSR4 boosted numbers.
Also would give them an opportunity to show framegen FPS beating a 4090 and beating Nvidia at their own mischief
But we all know what they say about AMD and opportunities...
Also had an interesting random theory... what if AMD's delays are software/driver-related?
E.g. the regular raster/RT perf is sorted, but FSR4 and new AI frame-gen stuff isn't ready yet?
Cos they probably would want to show any new tech/software when they 'officially' announce the cards.
It's less of an impact if they show regular raster numbers before showing fancy FSR4 boosted numbers.
Also would give them an opportunity to show framegen FPS beating a 4090 and beating Nvidia at their own mischief
FSR 4 looks impressive. If these benchmarks are genuine, the 9070 XT could be a real dark horse this gen: https://videocardz.com/newz/amd-rad...080-super-in-timespy-and-4070-ti-in-speed-way
AMD just need to price it right...
AMD just need to price it right...
Soldato
- Joined
- 14 Aug 2009
- Posts
- 3,192
Fingers crossed it's also easy to implement and AMD helps and pushes that rather then the usual "here it is, do what you want with it, don't care".
Soldato
Maybe AMD are gona let reviewers show it all off as AMD themselves seem to mess it up when they do it LOL
Associate
- Joined
- 5 Apr 2021
- Posts
- 45
Daniel is convinced at least.
Last edited:
IF = to 4080 in raster and = to 4070 Ti in RT $480 starting price (£460) is good, tho it also depends on how good the 5070 is, $550 starting price for that is very aggressive pricing.
Last edited:
It does look impressive and yes it says "Developed for RDNA4" it could be backported to RDNA 3 and again yes AMD are looking at that.
Also i agree the wording, again suggests that any game with FSR 3.1 already in can be pushed through the FSR 4.0 algorithm through a driver toggle, FSR 3.1 is similar to Nvidia's DLSS 3 where the .dll can simply be swapped out for a later DLSS version, AMD do the same with FSR 3.1 so there is no reason, as with Nvidia and DLSS why you can't just have a driver toggle for it.
Last edited:
The thing about RT that Daniel is missing, not just him lots of people seem to be missing this.
RT performance is not linier on AMD GPU's, this is because of the difference in approach with how AMD vs Nvidia do it, AMD's RT architecture is Integrated, what this means is AMD's RT component is on the shader core, Nvidia have a dedicated RT cluster on the GPU, the benefits of AMD's approach is it is faster in latency terms, the down side is the more the GPU needs to work to produce a Ray Traced Image the more bogged down it gets because its sharing the rescores between RT and Raster given they are separate functions on the same shader, with Nvidia because it has a dedicated RT cluster it is only limited by how fast that cluster is, its independent from the Raster performance.
AMD can improve their penalty by making their RT ever more efficient and powerful, so that it just takes more to get it bogged down, this is what AMD are doing RDNA 4 vs RDNA 3, so a game that is not demanding of the RT hardware would not necessarily gain performance 'Relatively' but in games where the RT is brutal, like Cyberpunk RDNA 4's RT is that much more robust so it doesn't suffer the heavy RT penalty and with that you get an improvement in performance.
So that Speed Way Benchmark really doesn't tell us anything, what we need to see is how it does in Cyberpunk.
Part of this explained from a Developers point of view in the video and an Example of this in benchmark charts below.
RT performance is not linier on AMD GPU's, this is because of the difference in approach with how AMD vs Nvidia do it, AMD's RT architecture is Integrated, what this means is AMD's RT component is on the shader core, Nvidia have a dedicated RT cluster on the GPU, the benefits of AMD's approach is it is faster in latency terms, the down side is the more the GPU needs to work to produce a Ray Traced Image the more bogged down it gets because its sharing the rescores between RT and Raster given they are separate functions on the same shader, with Nvidia because it has a dedicated RT cluster it is only limited by how fast that cluster is, its independent from the Raster performance.
AMD can improve their penalty by making their RT ever more efficient and powerful, so that it just takes more to get it bogged down, this is what AMD are doing RDNA 4 vs RDNA 3, so a game that is not demanding of the RT hardware would not necessarily gain performance 'Relatively' but in games where the RT is brutal, like Cyberpunk RDNA 4's RT is that much more robust so it doesn't suffer the heavy RT penalty and with that you get an improvement in performance.
So that Speed Way Benchmark really doesn't tell us anything, what we need to see is how it does in Cyberpunk.
Part of this explained from a Developers point of view in the video and an Example of this in benchmark charts below.
Last edited: