AdoreTV did some calculations on the yields for a possible PS5 chip, his calculations are 83% yield after salvaging chips with redundancy which is pretty good.
That's the narrative today, but when the specs were released they weren't viewed as safe. Again, largest APU ever by AMD at the time. Heck, I'd argue that 10TF in 2020 isn't exactly going to blow anyone's mind, although it would definitely be solid. Being able to match the GTX2080 in performance (if Navi is able to bring efficiency up to nVidia level) would be amazing in a $400-$500 box., even if the 2080 is 2 years old by then.
the ps4 gpu was equvlent to a 750ti, what your saying is this time around sony is going to have a chip equivalent to a high end pc chip, and it will still be high end in 2 years, for 399? feel like thats pushing it a bit.
I wouldn't trust his yield calculations at all. Mostly because he has no industry sources or citations backing them up.
True, he is making the assumption that the yield will be around the same as 14nm had 2 years after it was made available.
safe bet they get crushed if they did, if the roumers of base scarlet are true.
he doesnt factor in, that even workable chips, atleast on the gpu side, can still be rendered not up to spec. sony would have to be targeting a decent bit above what they are using per chip to keep yelds up.
We haven't seen a 7nm GPU yet so we don't know what high end will look like on that process node. The PS4's GPU was a downclocked 7870 with 2 CU's disabled, which spec-wise is above a 750Ti but nVidia started making major efficiency gains with the Maxwell architecture. Also the 750Ti released 3 months after the PS4 came out.
I wouldnt be shocked if we see a nvidia card on 7nm by end of year, if not early next. than we should have a idea of what 7nm can do since amd is focusing on consoles.
If you mean "up to spec" is regarding the frequency, I don't think that is an issue.
We will probably see something regarding Navi during Computex at the end of May.
Yeah, definitely looking forward to seeing cards released on 7nm so we can get a better idea of what can be achieved on that process node!
sony just pulled out of e3, they arent scared, if anything they are aragent sony circa 2006, if anything that makes them play even more safe than they would. no reasson to spend more money than you have too.
If it's $400 with a brand new CPU, no one will complain. I have the PS4 since 2013. I am not buying into another ecosystem without my games.
Depends on what you consider pricey, if they went completely insane and had AMD create a APU chiplet just for them.. It would be pretty damn expensive.
The Vega 7nm card is reaching 13.8TFlops at 1800MHz and costs 800$, the only way they would reach 15TF+ would be if AMD made a Navi chip with 80CUs and 1500MHz+.
given that weird tweet from a ms exec, that isnt out of reason. the other is the pc conference, but I cant see it being big enough to reveal there. also there isnt a graphics architecture ms hasnt helped with, every gpu has to work with directx after all.
I really doubt Sony helped "design" the Navi GPU. They most likely did the same thing as when AMD was making Polaris, they made requests for specific features.
the expectation is a really good CPU, and a 10~12TF (likely on the lower part of this range) of the GPU.
8 to 10 tfps puts anywhere from a 1070 to 1080, not bad, in pratice b/c of lower overhead on consoles it gets closer to a 1080 ti in real world terms.
Beast CPU, twice the frequency and much higher performance per clock.
A 1080 would not be enough to provide a generational leap in visual fidelity over current gen games. Especially when all games are going to be 4K across the board. Not even close. One X already performs close to or better than the 1070 in some games.
good, so games will not push native 4k and instead go with smarter methods of pushing resolution like checkerboard and dynamic resolution. ;p
what are you expecting on a apu? ignore cost for a second, you will run in to thermal limits before you get enough tfps to make a massive difrence. we are also at a point where art direction gets you more than a gpu power increasse.
This is something that inevitably will happen if the next consoles do end up being around 8-10TF. I can see devs compromising IQ for higher visual fidelity and experimenting with new 3D techniques.
They can have a more efficient architecture by next year that can give significantly better performance at the same or less TDP.
We will most likely be looking at checkerboarding with better quality than Horizon Zero Dawn had, which is fine by me.
I believe their next GPU architecture is going to be along the lines of Vega, a graphics card that also does a lot of compute for enterprise.
Which isn't that bad, considering it'll be the new baseline.
What people? What are you going to say in 2020 if they return? Now E3 has significance a year later? I have seen a lot of people in the know scratching their heads, definitely more than anybody coming out and saying the show has no real significance any longer. The PR alone pays the cost of admission for Sony/Microsoft multiple times over.
amd's best gpu,which is 7nm, is only 12.6 tfpos I dont see that happening on a apu with thermal limitations.
Agreed.
The TFs of the PS4 and X1 took us pretty far game wise this gen. I'm sure the same will apply to next gen regardless of how many TFs it ends up with. The number itself doesn't really matter. The baseline being far bigger compared to this gen is enough. This will especially be apparent when the exclusives come out.
If they decided to go for an expensive console, they would go for seperate CPU and GPU.
but thats why I am saying 8 to 10, you run in to real thermal issues above that. if ether of them do it, it would be ms becouse they are willing to use vaper chamber cooling.
vapor chamber cooling isn't exactly expensive or an exotic technology. And if they use seperate CPU/GPU, cooling would be a 100% non-issue at slightly higher clocks.
thats what this company did. honestly not a awful idea, though I bet oems would scream their heads off if ms did it.
