About performance, we will cross that bridge when we come to it. For the moment, its all about powerpoint battles
If Samsung can get 14LPE Exynos chips out in 1H2015, I would be impressed, they have a shot at making Exynos a lot more relevant in chip design+process. Just the switch in process alone will help them leapfrog in performance, add A57 into that and we have an impressive CPU (even though the A57-based Exynos 5433 figures remain to be seen, Exynos 6/Infinity is their real shot). But count me as somewhat skeptical; the Galaxy Note series would seem to make the most sense; the galaxy tab pro series makes sense too, but they really should just merge the two (to make larger notes), and offer the stylus.
http://www.eetimes.com/document.asp?doc_id=1321974
Based off that and the TSMC information in their most recent transcripts (+both foundries stated 28nm -> 20nm performance increased by 30%), Samsung's performance increase or power consumption reduction is lower compared to 16FF:http://forwardthinking.pcmag.com/none/322862-get-ready-for-14nm-and-16nm-chips, and the 14LPP designation has the same increase as 16FF+, so still lower performance. The density reduction for the 14 LPE process is as here:
http://electronics360.globalspec.co...balfoundries-form-14nm-alliance-to-fight-tsmc
I expect LPP to be the same density. As we've seen the density improvement from Intel's 14nm is much higher than TSMC's 16FF/+, so I anticipate Intel's 14nm to be of higher performance compared to both TSMC and Samsung.
TSMC's conference calls are so damn confusing it is hard to figure out exactly what their process perf/w improvements are, for example they state 16FF+ is 15% faster than 16FF with 16FF being 30% faster than 20nm planar, but they state that 16FF+ is 40% faster than 20nm. That doesn't make sense: 1.3 * 1.15 = 1.495, so it decreases clarity, maybe that is what they want to do in the PR wars.
Just talking about products on shelves, Intel will have products on shelves by 4Q14. Whereas Samsung said they will have 14nm in mass production by 4Q14 and on the shelves by 2Q15 (probably the Galaxy S6)
That gives Intel a 6 month lead over Samsung/GF.
About performance, we will cross that bridge when we come to it. For the moment, its all about powerpoint battles
Performance (At same power) is as follows
TSMC 28HPM - 1x
TSMC 20SOC - 1.15x
TSMC 16FF - 1.38x (1.15 x 1.2)
TSMC 16FF+ - 1.59x (1.38 x 1.15)
Area
TSMC 28HPM - 1x
TSMC 20SOC/16FF - .55x (1.9x density increase)
TSMC 16FF+ - 0.4675x (15% denser than 16FF+)
Power (at same performance)
TSMC 28HPM - 1x
TSMC 20SOC - 0.7x (30% power reduction vs 28HPM)
TSMC 16FF - 0.46x (35% reduction vs 20SOC , 54% reduction vs 28HPM)
TSMC 16FF+ - 0.322x (30% power reduction vs 16FF)
http://www.eda.org/edps/edp2013/Papers/4-4 FINAL for Tom Quan.pdf
(page 19)
http://www.tsmc.com/uploadfile/ir/quarterly/2014/1aT1b/E/TSMC 1Q14 transcript.pdf
(PAGE 4)
"Our 16 FinFET plus matches the highest performance among all available 16-nanometer and 14-nanometer technologies in the market today. Compared to our own 20 SoC, 16 FinFET plus offers 40% speed improvement. The design rules of 16 FinFET and 16 FinFET plus are the same; IPs are compatible".
https://www.semiwiki.com/forum/content/3688-samsung-foundry-explained.html
Area
Samsung 28nm (gate first) - 1x
Samsung 14LPE (gate last) - 0.55x
Samsung 14LPP (gate last) - 0.55x
Power
Samsung 28nm (gate first) - 1x
Samsung 14LPE (gate last) - 0.49x
Samsung 14LPP (gate last) - 0.40x
TSMC 16FF+ should beat Samsung 14LPP in performance and power efficiency. Density remains to be seen because Samsung 28nm gate first was more dense than TSMC 28HPM (gate last). But as always these claims have to be taken with a hefty dose of salt. Final products are the real measure of the process' power/performance/area claims. The biggest factor which trumps all these 3 factors is yields. At 14/16 FINFET the fab to win will be the one which can ramp production at good yields.
Final products are the real measure of the process' power/performance/area claims.
TSMC puts a lot of contradictory information out there. So I'll defer to those slides, the CC seems unreliable.
We'll see if Samsung can actually deliver. Their PRs promised 20nm production during the first half of 2013. How'd that work out?
Many here are very quick to want to "prove" that Intel is losing its lead, and will cite other companies' PRs (TSMC, Samsung, GloFo, etc.) but when it comes to Intel people ever the skeptics.
Why?
That gives Intel a 6 month lead over Samsung/GF.
About performance, we will cross that bridge when we come to it. For the moment, its all about powerpoint battles
But as things look now, Intel's process tech lead is more or less eradicated. Things may turn out differently to either's favor in the end. But it's pretty astonishing really.
As things look now (what is out there), Intel has a substantial lead. As to the near future (14nm) Samsung may get closer to Intel. Intel is in no danger of being overtaken. If you look at the research papers from Intel fellows, it's clear that they have the depth of talent and financial backing to be the front runner in process development and implementation.
As things look now (what is out there), Intel has a substantial lead. As to the near future (14nm) Samsung may get closer to Intel. Intel is in no danger of being overtaken. If you look at the research papers from Intel fellows, it's clear that they have the depth of talent and financial backing to be the front runner in process development and implementation.
Well, I'm looking at the release plans of actual products which is what matters in the end (assuming the plans hold). And they show that Intel's process tech lead on 14 nm is more or less eradicated. What the future may bring, nobody knows.
I have no doubt that Intel will continue to be a front runner in semiconductor process tech for the foreseeable future, but I still find it astonishing that their lead has shrunk so quickly.
What product do you think Samsung is going to release on 14LPE so close to Intel?, the only logical option for such a near release is the Galaxy S6: but that creates the issue of releasing a 14LPE version Exynos, and a 20nm pln TSMC Snapdragon 810.
I don't think they'll pull the trigger on that and even if they do we're still talking a 4-5 month delay from when Broadwell based Core M's are released, and maybe 2-3 months from Atom. If they don't do that, then we're talking a likely release in September. Plus since Samsung essentially balked at Exynos Infinity, that should at least add a bit of doubt into your thinking.
Well, I'm looking at the release plans of actual products which is what matters in the end (assuming the plans hold). And they show that Intel's process tech lead on 14 nm is more or less eradicated. What the future may bring, nobody knows.
I have no doubt that Intel will continue to be a front runner in semiconductor process tech for the foreseeable future, but I still find it astonishing that their lead has shrunk so quickly.
Well, it does seem that Samsung's 14FF will briefly narrow the gap between Intel and itself, but much needs to be seen before we'll know how much. IIRC, Intel is still creating an HP and a LP node for each process. It seems like foundries are starting with just one node and adding variations as that node matures.
Yes, in the long term, Intel's lead grows. After 20FF, it will be more than 2 years (2018) before they have 10nm aka 14FF, which is (almost) 4 years later than Intel's 14nm.
It is not astonishing at all. 14nm is difficult without EUV, which resulted in a 6 month delay. TSMC and Samsung know this, which is why they will release it 3 to 4 years later. But if they would release 14nm in 2018, they would have fallen too much behind. So in the meantime, they will quickly release an update to their current 20nm process with FinFETs instead of planar transistors.I have no doubt that Intel will continue to be a front runner in semiconductor process tech for the foreseeable future, but I still find it astonishing that their lead has shrunk so quickly.
Yes, on 14 nm likely the Exynos for Galaxy S6 and close to that Apple A9 SoC. If Intel only has a 6-9 month lead, it's really nothing much to speak of.
Check out the OP where TSMC says that they will have "10 nm trial production in 2015, mass production in 2016". If that plan is followed, Intel will not have much lead on 10 nm either. We're talking months, and definitely not 4 years.
Samsung's Galaxy S6's SoC will be made on 20nm.
Because that's the natural thing to do. There's been a discussion about this before.Source? Also, note that Samsung has previously made two versions of the Galaxy. E.g. for Galaxy S5 there is one based on Samsung Exynos 5 Octa, and one based on Qualcomm Snapdragon 801.
Look at the post above you. Node names are pretty much marketing labels, which TSMC, Samsung and GlobalFoundries now abuse to make their process nodes look better. There's no way a silicon FinFET transistor will compete against a Ge or III-V transistor. So performance will be more like 14nm. What about density?
Also looks like 14nm, although that might still change. But if it looks like a duck, quacks like a duck... and is released like a duck that has been 2 years behind for the last nodes (2014 -> 2017).
Because that's the natural thing to do. There's been a discussion about this before.
So you should give a source.