Yep.
And how do you know that increasing bandwidth on polaris wouldn't increase performance ?
you are missing the point.
But you know that nextgen consoles will target 4k not 1080p so hard to say they will have better bandwidth, they should have big enough bandwidth for 4k.
there was the same 10% performance uplift in 1440p, and i am assuming 1440p/2160c for most games, not native 4k.
digital foundry said they have not yet got a confirmation its been cancelled.
Hilarious ? That would be terrible. We don't need an under powered console that would set the bar for all next gen multiplat games.
5700xt ~9.3tf with 448gb/s and xox is 6tf with 326gb/s I see 1.55x increase in tf and only 1.37x in bandwidth, console could have even worse bandwitdh per tf ratio than 5700xt so no, we could get bandwidth penalty to performance rather than increase
Brad Sams and Zhuge both said Lockhart is cancelled, and they are reliable insiders.
companies do not have to confirm or deny every rumor that pops up.
i dont see a low powered next gen console launching in the middle of the generation, how do they handle every scarlett game that launched before? do they go for CPU parity and force every developer to have a completely unlocked dynamic resolution?
Ok, but hard to say we can put some +10% because nextgen consoles will have bigger bandwidth if we even don't know what the teraflop/bandwidth ratio is ;d I would predict that ratio will be worse than on xox.
or insiders can be wrong ... We will see soon™ enough ;)
Yep, we don't know bandwidth of nextgen consoles and we can't assume they will have some advantage bandwidth/tf than xox. Future will tell.
Jesus man that doesn't mean "lokhart is alive"or they "said" its alive. This is exactly how misinformation spreads.
As I showed it many times earlier, especialy vega 64 lose petf/tf badly comparing to smaller chips and df compare chips with the same number of cores. Current gen consoles don't have many cu's so they did good work.Interesting article/video by Richard but still very muddy.
I still think a 7.9 tflops 5700 card outperforming a 10.7 vega 56 and a 12.6 vega 64 is a great indicator of the tflops efficiency we can expect from Navi cards.
I think we can easily get to 2080 levels of performance with a 10.7 tflops Navi GPU.
What are you saying? This is just classical confirmation bias... so bent on bein right that you throw reason out the window.
well... sorry. You responded to someone asking who said its still alive. By saying "DF"...
Maybe my comparison wasn't best but just read my further messages as I don't want to repeat myself.
I just interpret it to mean there are some sort of repeated/distributed structures vs. some monolithic central scheduler/interpreter for RT workflows.
