• 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

If they glue together two Polaris 20 chips on a single PCB via something that works in all games as a single-GPU-card, that would make it faster than GTX 1080Ti.
 
Essentially Vega64 is 2x rx570's. Scaling to 4608cu, won't get near the 1080ti. Besides Amd need to focus on competing with Turing, not the 2 year old Pascal.
 
If they glue together two Polaris 20 chips on a single PCB via something that works in all games as a single-GPU-card, that would make it faster than GTX 1080Ti.

It's been a while since they had a dual card, the Pro Duo being the last one, iirc. Not sure they could get it to work reliably enough in gaming as it is still a crossfire implementation which is kind of a dead tech now, being as it relies on driver support etc.

I think the bump in performance and perception of 'something new' will help at the target market for these cards and tide AMD over until the true replacement, in Navi, makes an appearance.
 
These no.. Navi 10 with HBM2 and GDDR6 support won't happen. And probably it will be HBM3..

I recall an interview with David Wang earlier in the year essentially saying they were hedging their bets on which memory type to use. Pro cards and data centers would always "happily" pay the premium for HBM (because they'd actually make use of it), but at a consumer level that's not so clear cut. As a result they will be looking at both HBM and GDDR6 for Navi and see where it takes them. Plus there's no way the PS5 will use HBM, so I would say a Navi gaming card with GDDR6 is possible.

Also I very much doubt a Polaris card can get anywhere near the 1080Ti. If it were possible they would have done it before now. The difference in process from "14nm" to "12nm" doesn't suddenly give Polaris enough headroom to gain 50% more perf.

You yourself say 15-20% gains, but you must surely know that wouldn't get a 580 anywhere near a 1080 Ti. It may be beating a 1060 today, but the difference between a 1080 and a 1080 Ti is >30% as it is...

I was throwing milestones around in my wee fantasy land, so it's pointless to hold me to too much accuracy. That being said, hasn't it been suggested that GloFo's 14nm was pretty ropey and as a result hampered a bit of Polaris's potential? So gain that potential back, shrink down to 12nm which is supposed to be a good process for GloFo and strap on the gains you'd get from using GDDR6 and we could possibly see Polaris 30 perform better than we'd normally think.

Also, I didn't say Polaris would beat the 1080 Ti. I said that RX680 could be snapping at 1080 Ti's heels with a chunky overclock, much in the same way an 1080 can get within Ti range with an overclock. The actual point was clocking the nuts off a much-improved RX680 could bring it within sight of the RTX2080 at half the cost, which I think everybody here would see as billy bargain of the century.

It's just a discussion point anyway, for all we know Polaris 30 will give 5% gains purely by using GDDR6 and Navi 10 still won't touch a 1080 Ti, and is still stupid money because AMD are stuck on using HBM2 in a gaming card.
 
It is probably a completely pointless exercise but has anyone graphed transistor count to performance to calculate the most efficient current chip design? The great disappointment with Vega 64 from a pure gaming perspective was that at 12.5B transistors it could have reasonably have been expected to perform at 1080ti levels - but because it was trying to be all things to all men it could not use all it's power in gaming and lagged massively behind.

At 5.7B Transistors the RX580 is actually a far more effective pure gaming card for AMD in terms of perf/transistor. A larger chip on a better process and faster Ram would potentially perform significantly better. It wont happen - but I think it was possible.
 
Is Navi actually a departure from GCN at all?

I'd also like to have replies from people who realised that Vega was just GCN, as was obvious all along.
 
I think Navi is supposed to be the last GCN architecture. New architecture in 2020.

In which case people are dreaming if they are hoping for anything but incremental gains. GCN is just not an efficient architecture.

I bet IPC will be the same (i.e. Fury X vs Vega 64) and maybe they'll get closer to 2ghz with 7nm.
 
Increased core frequencies and increased memory bandwidth with the adoption of GDDR6 will probably offer a nice performance increase over the Radeon 580. How successful it will be, depends on what performance it offers compared to Nvidia's 2060.
 
If they glue together two Polaris 20 chips on a single PCB via something that works in all games as a single-GPU-card, that would make it faster than GTX 1080Ti.

Even if you had crazy low latency, stupidly high bandwidth links, even if each card had the ability to access each other's memory as if local memory you still wouldn't be able to make two or more monolithic GPU cores work in that fashion - it needs a significant architecture change and/or developers implementing explicit multi adaptor rendering well.
 
Even if you had crazy low latency, stupidly high bandwidth links, even if each card had the ability to access each other's memory as if local memory you still wouldn't be able to make two or more monolithic GPU cores work in that fashion - it needs a significant architecture change and/or developers implementing explicit multi adaptor rendering well.
Well if they get them into consoles then that would happen automatically for the most part. That said I doubt the PS5 will be having such design, but you never know.
 

If you look at his March 2018 video he called the RTX cards.
Is this one of those new bots where you type in text and it reads it for you? Hard to tell, it sounds real, yet at the same time fake as I cannot understand why anyone would read like that with no emotion in their voice.
 
To be fair the 580 is a tiny die, they could just double it to 400mm but they also wanted that $230 price point.

If they do a 250mm die on 12nm they would get 30% more compute and also 10% higher clocks would net a significant upgrade, certainly within reach of 1070 performance. But it would be a redesign.
 
I personally think AMD will go straight to Navi and skip a Vega refresh (in the gaming space), esp if it is indeed a totally new architecture as rumoured.

I remember reading that Navi's meant to be a Polaris replacement not a Vega one?
I read that ages ago and if true Navi may only match Vega at best, a bit like Grenada and Polaris on release,
It may even be like Fiji & Polaris on release where those of us on Vega have no upgrade available in Navi,
Damn, I hope not. :(
 
Back
Top Bottom