Associate
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD 7nm GPU News and Rumours 2018/2019
- Thread starter Beren
- Start date
Yawn, it's a cybercafé card for the Chinese market. This is the biggest non-story of the month.
Nobody is being deceived or defrauded with these, the intended customer I'm sure will be well aware of the specs.
That channel must have needed a boost to their views or something
As the guy said in the video, it is a 570 not a 580. But of course its AMD and we know they can do no wrong.
Yes hopefully it will be for china only and not for OEM as suggested in the video.
Soldato
About the same as a 2080Ti
confused ? a 2080ti takes up about the same as 1080ti which is already much lower than vega 64 which is a much smaller chip. I would think it get higher once the ray tracing is used.
Caporegime
Exciting stuff! Basically what's been rumoured for a while now, though seemingly lower price; may be wrong though, I don't see it being under $300. Does that mean that in 2020 we can expect 2080ti performance for $600ish? Tasty!
Q3 next year for something along the lines of a Vega64 but lower power consumption and $250 (read $350 with taxes/shipping in Europe). By which point the 2070 will be lowered to a similar price as nvidia make way for their own 7nm cards.
I don't understand what you mean. Do you mean the 2080 is almost near a 2080ti and once it drops in price that's going to be it, or something else?
This is just bizarre, I really don't understand how these people can come to the conclusions they do.
Ok so it is going to be a Polaris replacement, and the slide shown suggests 35% more performance.
So add 35% more performance to Polaris, we can be generous and use the already tweaked and clocked 580 variant, well that might just about bring you up to 1070 performance, and yet they reckon it is going to be a 1080 competitor.
Makes no sense to me and it is made even worse by the fact that the original article from Fudzila doesn't even mention the 1080.
Soldato
I've said before that this entirely depends on how you define "Polaris replacement": replacing Polaris architecture or replacing Polaris product stack. And, for me, given that Navi was never pitched as the "high end" card, everybody just assumes that it'll replace the RX500s - Polaris is mid-range so of course that means mid-range Navi will replace them. But what replaces Vega then?
Navi is the successor to Vega, so surely then Navi is actually the Vega replacement architecturally speaking, and we have all the talk of refreshed RX590 or RX600 cards landing on Polaris 30 which covers that bracket.
It's entirely possible that AMD won't over-promise and actually pitch Navi where it's going to perform: Navi 10 will be the mid-range segment performing between the GTX 1070 Ti and 1080 Ti, which you'd expect with from Vega's successor, leaving Polaris 30 to soak up the equivalent market of 1070 and down. And then we get Arcturus in 2020 for the actual top-end card (at least the intention thereof anyway). This shifts AMD's product stack and dvisions down a notch: navi replaces Polaris as the current mid-tier offering, and in turn Polaris is then pushed down as the entry-level offering.
But that GPU roadmap doesn't say anything about Polaris: it's Vega, Vega 7nm, Navi, Arcturus. The 7nm progression slide shows the story of investing and implementing that node, so if you assume those stats apply equally to 7nm GPUs as they do to Zen, then you're looking at 35% performance over Vega, not Polaris.
Your both right the slides don't say anything about Polaris.
It is the original Fudzilla article that says.
That's the line that throws a spanner in the works, because any 7nm Polaris replacement GPU with about 35% improvement in performance, is going to be knocking on Vega's 56 door anyway.
Are AMD still planning to bring us a 7nm instinct card this year?
If they do it will give us a good indication of what we can expect performance wise, just as the instinct MI25 did with Vega.
It might be just calculated over the nodes. TSMC 16nm was supposedly 10-15% "faster" than GloFo 14nm. TSMC 7nm process over their own 16nm is supposedly 35% "faster".
Its just my opinnion but I think Vega is going to stay but not coming to us end consumers. GNC is the way for datacenters, AI, Deeplearning + with additional cores they can come up with. Even Nvidia is going to closer the GNC route every new model. But on gaming side it isnt really efficient enough. Navi is going to replace Polaris as middrange card and AMD wont fight in high end. Money made from there isnt enough to warrant trying to fight Nvidia at the moment. They can use Navi trough out their whole portfolio consoles, laptops, igpu, graphics cards etc. I think going forward there is going to be 2 separate architectures, one for gaming, markets included above, and one for professional markets.
Soldato
Not if it's 35% improvement over Vega. I agree that line you quoted just confuses things, but a Polaris arch successor and a Polaris market segment successor are not necessarily the same thing: Navi 10 takes up the mantle of mid-tier product and Polaris 30 is relegated to entry-level. That's your Polaris market successor yet Navi is your Vega arch successor.
The 7nm Vega 20 Instinct card is coming end of this year unless plans have changed again.
Thinking about this a bit more and surely that slide is saying a 35% improvement over the previous node, not over any particular card.
So a Polaris replacement on 7nm would offer 35% improvement over Polaris just as a 7nm Vega chip would offer 35% over Vega.
It is just the hype spinning website regurgitating the same news over and over for clicks.
So a Polaris replacement on 7nm would offer 35% improvement over Polaris just as a 7nm Vega chip would offer 35% over Vega.
It is just the hype spinning website regurgitating the same news over and over for clicks.
It honestly gets boring seeing this same trend every year. Sites hype things up to stupid levels, people jump on the bandwagon then when it finally arrives its a disappointment because "AMD said" it would be this or that. When in reality AMD didn't say anything and people just believed the bs websites were pumping out which along the way somehow got engraved in stone as 100% factual info.
Soldato
The screenshot is TOO old.
Last we heard from AMD was that Instinct cards run at that, the V20 is 20.9Tflop+ card.
Now V64 at boost (1530) is ~12.5Tflop, and Instinct Mi25 is 12.3Tflop maximum.
20.9Tflop is more than 35%. I wouldn't be surprised if a well tuned Vega 20 (downvolt, overclock), wouldn't hit 24Tflop. Already we do 15Tflop with the cards we have atm.
What AMD needs to fix is the Pixel Rate (aka core speed). A Vega 64 has 98.30 GPixel/s and 12.5Tflop while a GTX1080Ti has 139.2 GPixel/s and 11.4Tflop.
That difference is down to one is working at 2000Mhz (1080Ti) the other at 1530(tops 1720 with tuning).
If Vega was running at same speed at the Pascal cards, it would have given it a hell of a kicking to oblivion.
However that could be solved next gen, as we know the AMD 7nm is on HPP for all their product, no more LPP (low power process). That is why the Vega 64 (Power Save mode) and Mi25 just sip the power (170W 1550Mhz) while overclocked by just +10% needs +90% power.
Same applies to Ryzen CPUs. At 35W running fine, but to gain the extra 15% perf, they jump three times the power consumption.