Apple's new SOC is coming

Hopefully this along with some architecture improvements will help amd a bit more considering even current 7nm AMD cant touch Nvidias 12/16nm.

icedman:

Hopefully this along with some architecture improvements will help amd a bit more considering even current 7nm AMD cant touch Nvidias 12/16nm.

Lets hope Navi is not GCN.

labidas:

Lets hope Navi is not GCN.

Even if it isn't I doubt we will see more than 2080 performance, which still would be great for a budget gpu.

If Navi is still GCN, I'll be ready for a disappointment.

FrostNixon:

Even if it isn't I doubt we will see more than 2080 performance, which still would be great for a budget gpu.

2080 performance would be absolutely fabulous from Navi. I'd reckon 2070 far more realistic, yet I'm not sure it'll be even that. If it's GCN, then probably not. Assuming, of course, the rumours are true and an enthusiast Navi won't be appearing (it wouldn't be cheap, though).

Kaarme:

If Navi is still GCN, I'll be ready for a disappointment. 2080 performance would be absolutely fabulous from Navi. I'd reckon 2070 far more realistic, yet I'm not sure it'll be even that. If it's GCN, then probably not. Assuming, of course, the rumours are true and an enthusiast Navi won't be appearing (it wouldn't be cheap, though).

IIRC, AMD did talk few times about Navi. It should not have been GCN from 1st time it was on roadmap. But it should have similar building block structure. By that I mean 64SP (or similar units) per CU (Or similar unit). That was explained as ensuring easy backwards compatibility for existing software/games and less driver work. Then there is this last patent filled in over a year ago about quite variable "stream processors". That probably is not Navi, but Arcturus. I have strange theory about it. Navi will make base for next generation Gaming GPUs while Arcturus will be about computational/mixed workloads. Weird thing is that "Fixed" Navi statement vs. very adaptable patent. It hints that there are 2 separate approaches being cooked for long time. (Or Navi will be backwards "compatible" representation of that new variable approach.) Either way Navi should not be GCN. And considering that Navi got 50% of more experienced GPU people moment Polaris was done and Vega got other half... Then it got all attention... Then GPU division got help from CPU division... it should be New, otherwise AMD's GPU division had very long party doing very little work. = = = = As far as performance goes. It should replace RX-570/580/590, Vega 56/64. Nothing more should be expected. Only exception would be if October delay was true due to this improved manufacturing process. But even that would not create miracles, maybe 10% higher clock at comparable power consumption. I expect that higher performance level GPUs will come through MCM implementation and not through big die. And that may not be Navi, more likely Arcturus. But I really do not know since all that delay Navi has.

"ASML, which provides extreme ultraviolet (EUV) litho equipment, is looking to ship a total of 30 EUV systems in 2019. Of the units to be shipped, 18 have already been reserved by TSMC, the sources indicated ASML and TSMC are on a roll, obviously TSMC are limited currently, seems they have the orders in to up production at 7nm. I wonder where the other 12 are going? Talking about 5nm while Intel still stuck on 14.....

Andy Watson:

Talking about 5nm while Intel still stuck on 14.....

2 things: 1. Intels 10nm is very similar to TSMCs 7nm in actual transistor density 2. We'll see volume 10nm CPUs from Intel before we see anything other than low power SoC parts and one-offs from TSMC.

ttnuagmada:

1. Intels 10nm is very similar to TSMCs 7nm in actual transistor density 2. We'll see volume 10nm CPUs from Intel before we see anything other than low power SoC parts and one-offs from TSMC.

That's something I hear often but how accurate is it?? Let's find out. TSMC 7nm has higher transistor density then intel 10nm, about 14%. some(old) info from : https://www.semiwiki.com/forum/content/6713-14nm-16nm-10nm-7nm-what-we-know-now.html "Intel's 14nm process is significantly denser than the competing processes from GF/SS and TSMC, >1.5x. It has taken roughly 3 years for SS and TSMC to introduce 10nm processes that are only slightly denser than Intel's 14nm process. Later this year when Intel introduces their 10nm process they will again take the process density lead, ~1.7x denser until TSMC's releases their 7nm process with slightly better density.(old 2017 data, Intel ran into 10nm problems) When TSMC introduces their 7nm process they will have a 1.13x density advantage over Intel's 10nm process. In late 2018 Samsung is later expected to introduce their 7nm process with 1.23x the density of Intel's 10nm process, also later in 2018 TSMC will introduce 7+ with a further density improvement. Intel's 10nm is more similar to TSMC and GF/SS 7nm processes than to the competing 10nm processes. Even though Intel has a significant density advantage at each node the forthcoming 7nm foundry process will likely pass Intel for process density and maintain that yield for at least a few years.

@The Reeferman: Thanks, excellent post. I was wondering about Intel, and you made my day. Brilliant.

awesome, then we're all set for a July/August refresh of Turin on 7nm EUV.

Astyanax:

awesome, then we're all set for a July/August refresh of Turin on 7nm EUV.

There is no need. And likely no ready architecture update either. No need, because power efficiency is sufficient. Cost is likely not great for 2080Ti. But it would unlikely improve on 7nm, especially on newest iteration. And I do not think nVidia has made enough of changes and finished improvements from TU to go for another tapeout so soon.

Fox2232:

There is no need. And likely no ready architecture update either. No need, because power efficiency is sufficient. Cost is likely not great for 2080Ti. But it would unlikely improve on 7nm, especially on newest iteration. And I do not think nVidia has made enough of changes and finished improvements from TU to go for another tapeout so soon.

Agree. Plus AMD postponed Navi so they would just be competing with themselves.

Fox2232:

There is no need. And likely no ready architecture update either. No need, because power efficiency is sufficient. Cost is likely not great for 2080Ti. But it would unlikely improve on 7nm, especially on newest iteration. And I do not think nVidia has made enough of changes and finished improvements from TU to go for another tapeout so soon.

There is definitely a need, 12nm is an expensive high loss node and the die is too large, too hot and too power hungry.

icedman:

Hopefully this along with some architecture improvements will help amd a bit more considering even current 7nm AMD cant touch Nvidias 12/16nm.

Less to do with the nm, more to do with the architecture. You can have the best node processing in the world, and still have a horrible architecture that performs horribly. I'm not saying AMD's current architecture is "horrible", but i am saying that AMD needs to make their architecture better and worry about that, not worry about a refined process of 7nm technology.

I'm curious ho w 7nm Zen2 will look... tbh more so than Navi on 7nm.

fantaskarsef:

I'm curious ho w 7nm Zen2 will look... tbh more so than Navi on 7nm.

Same, though i'm looking forward to both the 7nm change and the architectural changes. If Zen2 was simply and plainly a node shrink i'm not sure how interested i'd be unless it somehow got ryzen clock speeds to unthinkable frequencies lol