| goopy20 said: Ray tracing might just be an optional thing on pc right now for the 1% of pc gamers who own a RTX card, but it won't be next gen. |
You can do Ray Tracing on non-RTX cards on PC.
It's just the hit to performance is rather significant, hence why we have Ray Tracing cores now to minimize that impact.
| goopy20 said: On consoles they'll be able to build their games from the ground up with RT in mind, and it should be a lot more spectacular than having some reflective puddles in BF. |
That really depends how extensive the Ray Tracing cores are on next-gen, if the Ray Tracing capabilities come up short even against the RTX 2060, don't expect anything significant.
However, it's still early days and developers are still experimenting, we aren't going to have "spectacular" ray tracing effects for at least a few more years even with consoles supporting the technology in hardware.
| shikamaru317 said: You're forgetting that we haven't actually seen AMD's hardware raytracing in action yet. By the time that AMD's RDNA 2 GPU's release later this year, about 2 years will have passed since Nvidia released their first 20 series GPU. AMD will have had alot of time to study Nvidia's raytracing implementation and improve upon it to make their own hardware raytracing more efficient. It's also entirely possible that some developers will choose to make ray tracing an optional feature on consoles, allowing people to choose to take a framerate and/or resolution hit in order to experience raytracing. |
AMD is a generation behind nVidia for the most part though.
nVidia will have second generation ray tracing hardware this year, which will likely correspond with a significant increase in compute hardware to tackle the problem and efficiency gains.
| Bofferbrauer2 said: When an XBO game ran in 720p, then the PS4 title generally ran in 900p, not 1080p. After all, 1080p is twice as large as 720p, but the PS4 ain't twice as strong as the XBO. |
720P vs 900P is a difference of 56.25% which is more than the theoretical "30%" performance difference. (Which is why flops is bullshit.)
| Bofferbrauer2 said: Those 30% more TFlops would get pretty expensive. Just look at the current AMD GPUs: The RX 5500 and RX 5600 are at 4.8 TFlops and 6 TFlops respectively, yet the price difference is over $100. And It's not just the price of the chip: The cooling can be much weaker and smaller and cheaper, and thus also the casing, making the console overall also cheaper to produce. |
The price isn't in the Teraflops.
50% wider memory bus, requires more PCB layers, more intricate power delivery.
Some variants have 50% more memory as well.
The real cost is in the die size... The RX 5600 is from a die-harvested RX 5700, where-as the RX 5500 is built from it's own smaller, cheaper die that is 58.8% smaller.
So in reality... The RX 5500 should be 50% cheaper than the RX 5600 due to the roughly 50% decrease in all the hardware, but that is generally not the case... Often seeing a 75% price discrepancy between the two parts in Oceania.
| Bofferbrauer2 said: You don't seem to understand what compromises need to be made for a modern console to reach a $300 pricepoint. The big consoles will be sold at a massive loss since they are coming with over 12 TFlops and an 8-core CPU. An RX 5700XT, currently the strongest AMD GPU, clocks in at just 9 TFlops and yet costs about $400, the Ryzen 7 3700X costs about $300. AMD doesn't need the console money nearly as badly anymore as they did for the current gen and thus will also ask higher prices for their hardware. |
PC costs and Console costs tend to be a little different.
Microsoft and Sony buy in bulk which can be a significant price reduction, plus they get their own contracts done for fabrication.
There is also part consolidation, GPU's on PC's for instance have their own power delivery and RAM, on consoles that is all shared with the other components.
Plus you have the profit margins, AMD tries to retain a 64% or higher profit margin... But that really depends on how performant they are relative to nVidia and what markets they are chasing... AMD's semi-custom division tends to get lump sums with licensing revenue, totally different pricing structure.
And of course you have the die-size vs clock frequency part of the equation, having a smaller chip that can clock higher can mean more Teraflops, but also cheaper than the part it replaces, we saw this often with AMD's evolution of Graphics Core Next where they constantly re-balanced hardware every year.
| goopy20 said: There are a lot of games running at 720p on Xone vs 1080p on ps4 and not just the games that came out later on in this console generation, for example: COD Ghosts, MGS5, Pro Evolution 2015, Golf Club. Also a lot of exclusives like Quantum Break, Titanfall, Dead Rising 3 and even Halo 4 & 5 were 720p... The funny thing is that some of you commenting probably own a Xone and never even realized you've been playing games at almost the same resolution as the 360. It's because it wasn't a deal breaker for the Xone and not having native 4k isn't going to be a deal breaker next gen either. |
Allot of those games either employ a dynamic resolution or use frame reconstruction on Xbox One, so typical pixel counting isn't accurate, those technologies will be front and center for the 9th gen.
| Conina said: 960x540 is only one fourth (1/4) of 1980x1080. With perfect scaling (so if CPU, RAM, bandwith... aren't bottlenecks) you would only need a 2.3 Tflop-GPU (1/4 of 9.2 Tflops), not 4 Tflops (same architecture and everything else the same). Since scaling ain't perfect, you can expect factors of 3 - 5 when you reduce the resolution to 25%. Let's have a look at your own chart: <SNIP> 3.6x - 3.8x the performance by reducing the resolution to 25%, as expected. Yeah, you still suck at math. |
Nah. You can't just increase Teraflops and expect a linear increase in performance, not if bandwidth is the same, not if the texture sampling rate was the same, not if the geometry engines was the same, you need to scale everything else upwards.
It's why Navi with less Teraflops performs better than Graphics Core Next with more Teraflops, because *everything else* got a substantial increase in performance.
And if your design has a bottleneck, sometimes the hit to performance is larger than the performance delta would otherwise imply.
www.youtube.com/@Pemalite
