[Sweclockers] Tonga is going to do more than just replacing 750 Ti this summer

[NTMBK]

Well GlobalFoundries has one of its 22/20 nanometer class nodes in the 28nm realm.

GlobalFoundries 22-nm/20-nm class nodes;
- 28nm SHP
- 20nm LPM
- 20nm SHP(renamed to LPM+)

We are already seeing 22nm SKUs; Kaveri, Beema, Mullins, etc.

No, Seronx, just no.

 

[ams23]

Your post sounds like a heavily slanted advertisement.... Just saying.

The thread title is nonsensical. Only NVIDIA can "replace" one of their existing products in the lineup. AMD already has products that can "compete" against GTX 750 Ti in terms of perf. and perf. per dollar, but as of now they do not have anything that can "compete" with Maxwell v1 in terms of perf. per watt.

 

[Cloudfire777]

TSMC 28nm HP (All GCN GPUs are made with this process)
vs
GlobalFoundries 20nm LPM
=
15% less power for 20nm LPM according to GlobalFoundries own documents

------------------------------------------------
This could be true for TSMC as well because they only use 1 process there as well (LP) because the other processes didnt gain any significant gains vs LP according to their own research.

16nm FinFET from TSMC begins mass production late this year. Samsung and GlobalFoundries are working together and are ready to begin production on 14nm FinFET also this year.

 

[Mondozei]

16nm FinFET from TSMC begins mass production late this year. Samsung and GlobalFoundries are working together and are ready to begin production on 14nm FinFET also this year.

Yeah, we we won't see those for 2 years. A leaker at the OC forums posted these slides:

It was the same leaker who did later showed the HBM slides that AMD has been talking about generally for quite some time now. Time will tell if it's accurate.

He also wrote that we should see 14 nm GPUs come in 2016 but only in the low- to midrange segment. I hope he's wrong. Waiting until 2017 for high-end 14 nm would be a re-run of this year and next, where the low- to midrange cards are on an old node and the new node gets reserved for the high-end, much later.

This is what we get from the diminishing returns of node shrinks. Graphite, where art thou.

 

[Cloudfire777]

That looks awefully a lot like roadmap for VRAM/HBM for Hynix?
It also looks like HBM is ready this year. So I`m not quite following you why 16nm FinFET with HBM isnt ready for AMD already next year.

According to TSMC, 16nm FinFET yields are already better than 20nm SOC (current production) and they have 15 tapeouts this year (2014) for 16nm FinFET.
Another interesting aspect is that it delivers 40% increase in power consumption over 20nm SOC so this will be very interesting to see the GPUs built on this node.

I don`t know about 14nm progress but could go fast there as well.
All I know is that many recent articles said fabless companies could skip 20nm and go directly to 16nm/14nm due to the progress they made on the production and the performance gains.

Makes you wonder if AMD/Nvidia will stick to 28nm for this year and then move on to 16/14nm in late 2015 or something. I know AMD recently said they would do 28nm this year and then move on to 20nm next year, but was that for all products or maybe just APUs?

Graphite would be nice yes. Long due coming to ditch silicon and just crank up the power with graphite
Too bad research seems to take foreeeeeeeeeeever.

 

[NostaSeronx]

@Mondozei;

I didn't leak anything everything was already on the internet.

Also, I could be wrong with this:

28nm TSMC Q4 2013 : High-end GPU, VI 1.0
28nm GlobalFoundries Q4 2014 : Mid-end GPU and Low-end GPU, VI 2.0
28nm GlobalFoundries Q2 2015 : High-end GPU, VI 2.0
20nm GlobalFoundries Q4 2015 : Mid-end GPU and Low-end GPU, PI 1.0
20nm GlobalFoundries Q2 2016 : High-end GPU, PI 1.0
14nm GlobalFoundries Q4 2016 : Mid-GPU and Low-GPU, PI 2.0
14nm GlobalFoundries Q2 2017 : High-GPU, PI 2.0

This roadmap to me already looks faulty when Tonga is a high-end part just slightly below Hawaii. The final blow to this roadmap is if Tonga is 20nm LPM.

If Tonga is 20nm, it becomes this;

28nm TSMC Q4 2013 : High-end GPU, VI 1.0
20nm GlobalFoundries Q4 2014 : Mid-end GPU and Low-end GPU, VI 2.0
20nm GlobalFoundries Q2 2015 : High-end GPU, VI 2.0
14nm GlobalFoundries Q4 2015 : Mid-end GPU and Low-end GPU, PI 1.0
14nm GlobalFoundries Q2 2016 : High-end GPU, PI 1.0
10nm GlobalFoundries Q4 2016 : Mid-GPU and Low-GPU, PI 2.0
10nm GlobalFoundries Q2 2017 : High-GPU, PI 2.0