• 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 Polaris architecture – GCN 4.0

Seanspeed said:
You're extrapolating release timing from the absence of evidence that it'll be sooner? Cuz that's not how it works. Might as well say there's no sign it'll come out before 2020 using the same reasoning.

AMD are also using TMSC 16nm, though we dont know for what specifically.
Click to expand...

Sorry but where in the sweet jesus did you get any of that from. The absence of evidence? Multiple companies switched from TSMC to Samsung for chips this year because they all felt TSMC couldn't deliver yields and quality required. That isn't absence of evidence, that is direct evidence.

Samsung was 6+ months ahead of TSMC in manufacturing their first finfet parts, their higher performance version of 14nm is 6+ months ahead of Nvidia's 16nm higher performance process. The industry at large has for the past 2 years believed Samsung to be 6+ months ahead. Except for the usual people making crap up(both forumites and the usual trash tech 'news' websites), there is no indication that Nvidia would release before AMD or any time particularly soon.

The news in the fab industry, from tech sites, from everywhere over the past 2 years is Samsung is a fairly large step ahead of TSMC.

The fact that this matches directly with AMD showing real chips and showing working demos of the real chip while Nvidia can't even show non working real Pascal chips only seems to agree with the general industries expectation.

Also we do know what AMD won't use TSMC for, Polaris, at least this year. They've stated for the record that Polaris is being done on 14nm Samsung/GF process.
 
Mtom said:
AMD only using GloFo/Samsung 14nm...no 16nm parts for them
Click to expand...

Pretty sure AMD announced they are using TSMC for some products.

14nm and 16nm, are actually the same process size BTW, just different marketing names. For a long time the process size name has just been a marketing name and has very little basis in physical reality. Both TSMC and Sasung process are actually more heavily based on their 20nm processes. They added FinFet technology and gave the processes a new marketing name to make it sound like a 1 node improvement. Thsi will be more and more common since the costs just go up andf up and the technology harder and harder to progress.
 
drunkenmaster said:
Sorry but where in the sweet jesus did you get any of that from. The absence of evidence? Multiple companies switched from TSMC to Samsung for chips this year because they all felt TSMC couldn't deliver yields and quality required. That isn't absence of evidence, that is direct evidence.

Samsung was 6+ months ahead of TSMC in manufacturing their first finfet parts, their higher performance version of 14nm is 6+ months ahead of Nvidia's 16nm higher performance process. The industry at large has for the past 2 years believed Samsung to be 6+ months ahead. Except for the usual people making crap up(both forumites and the usual trash tech 'news' websites), there is no indication that Nvidia would release before AMD or any time particularly soon.

The news in the fab industry, from tech sites, from everywhere over the past 2 years is Samsung is a fairly large step ahead of TSMC.

The fact that this matches directly with AMD showing real chips and showing working demos of the real chip while Nvidia can't even show non working real Pascal chips only seems to agree with the general industries expectation.

Also we do know what AMD won't use TSMC for, Polaris, at least this year. They've stated for the record that Polaris is being done on 14nm Samsung/GF process.
Click to expand...


Any links to sources?

I know there were lots of reports of the Samsung process being less power efficient, eg.g.
http://www.macrumors.com/2015/10/07/tsmc-samsung-a9-battery-tests/
although other tests show only a small advantage to TSMC.


I haven't found anything to show the inverse. Samsung was a head of TSMC to this node but I don;t think that has any bearing going forwards at all. Nvidia taped out big pascal 6 months ago without any word of issues, normally there would be news of a repsin if there were problems. Samsung being able to supply some customers a few months earlier than TSMC doesn't effect product delivery 12 months later.
 
Last edited:
D.P. said:
Pretty sure AMD announced they are using TSMC for some products.

14nm and 16nm, are actually the same process size BTW, just different marketing names. For a long time the process size name has just been a marketing name and has very little basis in physical reality. Both TSMC and Sasung process are actually more heavily based on their 20nm processes. They added FinFet technology and gave the processes a new marketing name to make it sound like a 1 node improvement. Thsi will be more and more common since the costs just go up andf up and the technology harder and harder to progress.
Click to expand...

Time for this pic again!
UA50d6D.jpg.png


Intel still comfortable ahead of everyone, especially as they have been in volume production for quite some time.
 
yep, its a great picture. TSMC/Samsung 16/14nm is the same as their 20nm offerings because they had to reuse the expensive process development. And yeah, Intel is well ahead.
 
The only problem is down to delays with the Intel 14NM process,the time difference between Samsung and Intel getting their own 14NM chips out was only like six to nine months(Core M in September 2014 and SOC in the Samsung Galaxy S6 in April 2015),and the 14LPP process which is meant for higher performance CPUs is not listed there.

If you look at the pitch difference between the Samsung 14LPE and Intel 14NM processes,its less than 10% and Intel does not even have the advantage of using Finfets now.

Nvidia is listed as a 16FF+ customer. Both 16FF+ and 14LPE are smaller than the 20NM processes listed which also did not use Finfets.

Plus,that ancient slide made in 2014,does not even list the Samsung 14LPP process.

Also,the competitive landscape has changed since 2014 drastically - Samsung only realistically made chips for its own products and Apple.

Moreover,since 2014 we had Samsung and GlobalFoundries get into bed with each other and IBM transferred its own fabs and fab technology to the new Samsung-GlobalFoundries alliance.

We basically had the three major third party fabs outside TSMC essentially merge with each other.

But with Apple trying to get away from using Samsung due to competitive reasons,it means we have another company now available for making chips. I fully expect in the future Nvidia will also consider using Samsung/GF if it suits them.

Samsung and TSMC are far closer to Intel than they have for years,and the consolidation in the fab industry means the competitors to Intel are nowhere as fractured as they were even a few years ago.
 
Last edited:
D.P. said:
Pretty sure AMD announced they are using TSMC for some products.

14nm and 16nm, are actually the same process size BTW, just different marketing names. For a long time the process size name has just been a marketing name and has very little basis in physical reality. Both TSMC and Sasung process are actually more heavily based on their 20nm processes. They added FinFet technology and gave the processes a new marketing name to make it sound like a 1 node improvement. Thsi will be more and more common since the costs just go up andf up and the technology harder and harder to progress.
Click to expand...

They haven't announced anything, it was just a rumour.
The closest thing to an announcement is AMDs press release:
AMD’s Polaris architecture-based 14nm FinFET GPUs deliver a remarkable generational jump in power efficiency. Polaris-based GPUs are designed for fluid frame rates in graphics, gaming, VR and multimedia applications running on compelling small form-factor thin and light computer designs.
 
AMD announced they were using 14nm.

They also 'announced' (according to a sourceless article) that they had multiple suppliers but that doesn't mean TMSC - in the thread on here where people claimed it was rumoured to be confirmed that AMD would use TMSC even the dodgy sourceless rumour article made no such claim.
 
Anyway,so now we have small Polaris demoed and a larger Polaris demoed behind closed doors.

So that's the two chips AMD said they were releasing this year.

So,I wonder when we will see them both launch??

I hazard a guess that we won't see the large Polaris launch before July as the Fury range was launched last year around the same time and I doubt either AMD or Nvidia would replace a high end card within a year!

The smaller one should probably launch by summer as suspected -it might be the more important chip in some ways to claw back marketshare.
 
Mtom said:
They haven't announced anything, it was just a rumour.
The closest thing to an announcement is AMDs press release:
AMD’s Polaris architecture-based 14nm FinFET GPUs deliver a remarkable generational jump in power efficiency. Polaris-based GPUs are designed for fluid frame rates in graphics, gaming, VR and multimedia applications running on compelling small form-factor thin and light computer designs.
Click to expand...

Lookign into it is probable Zen that will use TSMC
http://www.eteknix.com/amd-zen-use-tsmc-16nm-gf-14nm-falters/

Makes sense if AMD are gearing up to release a load of new products they will need multiple suppliers. TSMC process is supposedly more power efficient and that probably makes sense that AMD use that for CPUs where they have to really make it big.
 
he stated on stage multiple times that it was pascal on the board. And all of thier official photographs of the drive 2PX only show the tegra side of the board or shots with heatsinks covering the board. but it appears to be using 980M modules.

Anyone got any wood glue?

Atleast the overclockers wet dream existed as real hardware. :P
 
Not a big deal?

https://gbr.pepperdine.edu/2010/08/businesspersons-beware-lying-is-a-crime/

https://semiaccurate.com/2016/01/11/nvidia-pascal-over-a-year-ahead-of-1416nm-competition/

The room is full of press and analysts so SEC rules mandate that everything said is absolutely true, it would be illegal to do otherwise.

as an executive in a publicly traded company with press and analysts who cover and trade Nvidia in the room, he is obliged by the SEC to be truthful. The only other explanation is that he knowingly lied to the press and analysts and showed a fake card in his keynote, something that would clearly be illegal.
Click to expand...
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom