So we've moved from "easily surpass" to "decimate the 7900XTX" now have we? And your evidence for this is a leakers guess about an assumption mixed in with some Sony magic sauce of your own imagining?
That almost certainly refers to some theoretical rays per second metric. Not 2-4x the performance in a game using RT."2x-4x RT performance"
Getting pretty saucy in here.
That almost certainly refers to some theoretical rays per second metric. Not 2-4x the performance in a game using RT.
Ya, I'll believe that when I see it. Spiderman 2 for example will not be seeing a 2-4x increase in framerate when GPU limited. I mean 4x puts it close to 4080 territory, which is just pure fantasy.@Dictator as Mr. Ray Tracing, surely you're committed to discuss this on the next DF Direct? With such a massive RT boost on PS5 RT, you can't be absent from that discussion.
Nope, real world performance as confirmed by Tom (Moore's Law) and Tom (Insider Gaming).
That almost certainly refers to some theoretical rays per second metric. Not 2-4x the performance in a game using RT.
Ya, I'll believe that when I see it. Spiderman 2 for example will not be seeing a 2-4x increase in framerate when GPU limited. I mean 4x puts it close to 4080 territory, which is just pure fantasy.
Well if you look at my chart above for the 7700XT, 4x would put it closer to the 4090 than 4080.
CBR / ID Buffer was completely new to implement for most developers. Here PSSR should be much easier to do if they already do FSR or DLSS (as explained in the slides). But you are right, we'll probably still get many games still using FSR2 on their Pro mode.Except there is no guarantee most devs will use it. We saw what happened with PS4 Pro and its ID buffer/checkerboard combo. I expect we will see a very large number of games just up the FSR 2 quality setting.
In any cases it's compared to PS5 which is extremely weak here (compared to Nvidia cards). We haven't even see any games with all main RT effects on PS5 (too weak to ship like this). It's usually one effect as adding more would likely tank the framerate to ungodly low levels.@Dictator as Mr. Ray Tracing, surely you're committed to discuss this on the next DF Direct? With such a massive RT boost on PS5 RT, you can't be absent from that discussion.
Nope, real world performance as confirmed by Tom (Moore's Law) and Tom (Insider Gaming).
They probably added 1/2 RT effects on a build, seen the framerate tank to 1080p 15fps on PS5 (very poorly doing with only dedicated intersection hardware) and seen the framerate reaching 1080p 60fps on PS5 Pro. I think it's quite a meaningless metric as we'll never get the PS5 build using all those RT effects enabled (PS5 would still run SM2 at 60fps with only RT reflections). But even the Tflops and Tops metrics are more or less meaningless in many cases.Ya, I'll believe that when I see it. Spiderman 2 for example will not be seeing a 2-4x increase in framerate when GPU limited.
Ya, I'll believe that when I see it. Spiderman 2 for example will not be seeing a 2-4x increase in framerate when GPU limited. I mean 4x puts it close to 4080 territory, which is just pure fantasy.
I am so glad the foundation of their upscaling solution is one of my favorite AA/upscaling and not FSR!PSSR is machine learned-enhanced version of TAAU
This is a 3x improvement from PS5 (around 20.6 Tflops) assuming the rumored dual issue efficiency improvements are actually working. We know 1st party devs are already using FP16 extensively on their games (for ML inference) so I am not surprised it got a big upgrade.62 Tflops of FP16
This is the biggest surprise for me. I really didn't think Sony would actually do their own DLSS-like solution for PS5 Pro!8-bit integer 300 TOPS
Who is "they"?They say 4 times is a rare case. It can be only two times for example. Most of the time it is 2 to 3 times.
Yeh if he is going to bother talking about unconfirmed hardware performance he may aswell talk about how the next nvidia gpus are going to be 2-4x faster than a 4090 in RT and have 32GB memory and it will be like having 256GB because neural texture upscaling will be a force multiplier .............. /s@Dictator as Mr. Ray Tracing, surely you're committed to discuss this on the next DF Direct? With such a massive RT boost on PS5 RT, you can't be absent from that discussion.
Nope, real world performance as confirmed by Tom (Moore's Law) and Tom (Insider Gaming).
Dictator said:The multipliers on RT Performance being quoted in media need a lot more context as I think that might be confusing people or perhaps Just need more info.
RT Hardware can do a lot of different things and 2x or 4xing certain aspects of it still will not necessarily offer a 2x to 4x improvement in overall game performance that uses RT (i.e. raw frame-rate).
The RT Pipeline
1. Dispatch Ray
2. Traverse BVH
3. Trace against AABB
4a. Trace against triangle
Or.
4b. Trace against Alpha tested
5a. Shade based on simple Material
Or
5b. Shade based on complex Material
6. Denoise
Making one of these steps 2 to 4x faster for example does not speed up overall RT Rendering by that same amount, just the substep. For this reason, as an example, Ampere GPUs speeding up step 3 and 4a by 2x over Turing did not mean 2x RT performance with the same core count.
Essentially, more info is needed as to what the actual hardware has in it for RT purposes.
Potentially yeah, but no way around that other than to widen the APU's bus - then you've likely overshot your price target. I always felt the most "out there" rumour was the 320bit bus variant because of this.
Nah.. ps5 dev kits run higher than retail. The dev kit os 13 teraflops and retail is 10.. thats why people thought the ps5 was 13It seems 2180 MHz is typical frequency for a develop kit, just like 9.2TF@2GHz of PS5 dev kit.
PS5 retail product has variable frequency of 2.23 GHz max. It's likely PS5Pro also has variable frequency higher than 2.18GHz of dev kit. 2450 MHz may be the max number of variable frequency.