• 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.

Exclusive: The AMD Inside Story, Navi GPU Roadmap And The Cost Of Zen To Gamers

nVidia has nothing to succeed GTX 1080 Ti either. Which means there is a very high probability this gaming cards industry just collapses in several years.
Unless, they go to something completely different and save it.

They aren't struggling to top the 1080Ti either - they've shown off two more architectures in varying levels of detail post Pascal and even a stripped out gaming orientated Volta on 12FF would serve as a successor albeit not the most cost effective for nVidia.
 
I know. They won't release anything and "a long time" means literally. Now, let's guess if it is 2019 or 2020.
By that time, Intel will release something, and maybe AMD will be ready with something, too.
 
Depending on the software, RX Vega 64 is competitive with GTX 1080 Ti.
GTX 1080 is competitive with RX Vega 56.
GTX 1060 6GB is slower than RX 580 8GB.
...

AMD are not in a trouble either. nVidia has slight performance lead only for the tier which is bought by 2% of the consumers. For the rest 98%, AMD has RX 580, Ryzen 5 G and everything below.

What about 1070 and 1070ti ??

And where is Vega 64 competitive to 1080ti i must have missed something oO

edid quick Google and
http://gpu.userbenchmark.com/Compare/Nvidia-GTX-1080-Ti-vs-AMD-RX-Vega-64/3918vs3933

ye 32% average faster 1080ti totally same level performance LOL

and some fresh benchmark's with Driver 's from this yearzzz that deffo unlocked the TRUE POWERZ of VEGA
 
Makes sense I guess, and with all the talk of AMD going bust for the past few years it makes sense that Lisa focussed on the businesses that actually make money for them, which seems to be working.

What I do think about though is the 18 month hole this story presents until Navi 10 comes out. If 7nm Vega is sampling to partners for Instinct right now, but 7nm Navi isn't landing until the end of next year, are AMD just going to churn out RX Vega cards in their current state until then? Surely there is room for a 7nm Vega refresh to fill the gap, or is it going to cost too much money to make that shrink?



7nm Vega is not sampling to vendors until the end of the year, volume production but until next yewr. And volume production of a you selling for $4-8,000 is very different to a $400-800 consumer GPU.
 
I know. They won't release anything and "a long time" means literally. Now, let's guess if it is 2019 or 2020.
By that time, Intel will release something, and maybe AMD will be ready with something, too.

Intel could end up being the biggest competition. I reckon they will have a plan go after the biggest market segments pretty hard.
 
I thought Instinct was sampling to vendors now for release at the end of the year?

Pretty sure AMD started sampling for 7nm GPU's around last winter time and committed to a big contract at the turn of the year.

I'd take what D.P says with salt. He has a very entrenched view of things.
 
No, AMD have engineering samples, vendor samples end of year. Volume production next year. 7nm is not ready for volume production of large GPUs yet.

TSMC will be in (or at least ready for) mass production for 7nm through the whole range by the middle of this year (as in very shortly) - that includes server CPUs and gaming products range including GPUs.

Pretty sure AMD started sampling for 7nm GPU's around last winter time and committed to a big contract at the turn of the year.

I'd take what D.P says with salt. He has a very entrenched view of things.

AMD don't seem to be rushing at TSMC 7nm - more likely end of year for AMD 7nm products there and Gary Patton has confirmed similar at GF though that is probably Zen products.
 
Last edited:
Historically AMD haven't been able to afford the latest and greatest manufacturing process for more than a year after the big Smartphone companies get access to it.
Until today wccf had been blatantly over hyping 7nm. This article seems to be them now trying to pour water on their own flames.

How this affects most of us, is with the perennial decision on whether to play 'skip-a-generation' or not with the Nvidia cards later this year.
With longer generations its a game thats getting harder to play.
 
Historically AMD haven't been able to afford the latest and greatest manufacturing process for more than a year after the big Smartphone companies get access to it.
Until today wccf had been blatantly over hyping 7nm. This article seems to be them now trying to pour water on their own flames.

How this affects most of us, is with the perennial decision on whether to play 'skip-a-generation' or not with the Nvidia cards later this year.
With longer generations its a game thats getting harder to play.

I'm sure not about release times getting longer. Vega is about 8-9 months old (IIRC) and someone from AMD said they plan to have a new generation each year.
 
Wasn't the new gen each year in reference to CPU only?
If that wccf article was actually true, then Vega wont be getting superseded by something higher end, but replaced by at most another mid range. Like the 480 replacement of 290/390.
 
Wasn't the new gen each year in reference to CPU only?
If that wccf article was actually true, then Vega wont be getting superseded by something higher end, but replaced by at most another mid range. Like the 480 replacement of 290/390.

Fairly sure the generation-per-year comment was in reference to GPUs because AMD have admitted they've lost their way with consumer GPUs. The new roadmap graphic shows 14nm Vega at 2017, 7nm Vega, 7nm Navi and 7nm+ "Next Gen" at 2020. That's 4 years and 4 generations, so as a result 14nm Vega was last year, 7nm Instinct is this year, 7nm Navi is next year and "next gen" is 2020.

Given that Ryzen 2 is actually Ryzen refresh AMD are still calling it 2nd generation, so by that logic a 7nm RX Vega would be a new generation, making that 5 gens in 4 years and doesn't fit the roadmap. But unless 7nm Navi lands in Q1 of next year, we are still looking at 12-18 months until 2019 is over, which is a good chunk of time to wait for a new RX card and churn out RX Vega in its current state. But then AMD have done this before.
 
Here's a thought though: it's been suggested that one of the contributing factors for Vega not delivering was AMD weren't discerning enough with the Vega yields and literally smashed every chip they made onto a card and then set the reference power draw stupid high to cover those poor yields that should've really been thrown out.

If AMD are going to stick with 14nm Vega for RX cards for another 12 months, perhaps now they can be a lot more selective with their yields, do some proper binning and do some kind of reference revision with a much lower power requirement.

And slash at least £150 off the price so they're in the correct price-performance bracket :p
 
Do you seriously expect a company to officially announce its plans 3-4 years in advance? :confused:
Or do you expect someone to officially go on record as breaking an NDA?

No, I expect people to take anything like this written in the tech media as complete BS. As if AMD aren't going to improve on Vega for 3 yrs. They could do that with zero effort if they wanted to, they already have the parts and call them Instinct.

WCCF and every other similar site create these articles to get traffic. Next week they'll probably publish something that completely rebuffs this and flip flop til the cows come home.

If AMD make an official announcement, that is the only source anyone should believe, not some journo trying to generate money for their site.

Then again, nobody in here would be able to argue about it if we all just saw it for what it was :P
 
I'll stick this in the box of theories at this point. It is almost as soapbox as most of my roadmap theories and with almost as little tying it together. Reverse theory? Nvidia are actually going to launch sooner than expected and this is negative press on AMD having anything in the pip to soften the market.

AMD have some pretty competent Accountants as well as engineers. Everything about their CPU roadmap is the creation of a design and then rolling that design out to every single layer of tier from Enterprise, HEDT, Mobile, Desktop and Entry Level.

14nm Vega cores have been used in everything from APU to enterprise which to me shows the same thinking at work.

Do we think under the same leadership that it is likely that the GPU segment would have invested all the money required to create a 7nm Vega chip and not role it out across as many verticals as possible?

I was suprised that they have not taken the opportunity to use the 12nm process for Polaris as a final boost to that, but I would be gobsmacked if they did not use the 7nm process for all lines of business given that they have already engineered it.

As for Navi, I am completely behind it being the backbone of a semi custom chip. I very much doubt it is just the PS5 but also in the nextgen Xbox, it gives you reliable revenue to back up development costs. However the idea that it wont also come to enterprise and PC gamers seems nonsensical.
 
Back
Top Bottom