Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Interesting graph which if true shows there will be 3 times as many smartphone users in 2024 than in 2014-2015. As 2014-2015 will be the birth of 64-bit Android Intel will be perfectly poised to be a big part of this growth going forward without any 32-bit legacy issues holding it back.
With the foundries now it's going to be survival of the fittest and I don't know which foundry and even which technology is going to come out on top eventually. Intel's process path is set and secure but it will be interesting to watch the foundry foodfight now that's about to take place below bulk 28nm which they all seemed to manage ok.
How does AMD accomplish all this at 28nm HKMG against 22nm Intel
Higher power limits and a more powerful gpu which looks more efficient as well. Bay Trail really should have had an 8 EU gpu. They guessed wrong 4 years ago but this is going to be corrected with Cherry Trail and its 8 and 16 EU gpus. Also remember some of those benchmarks are against the 1.86 GHz Z3740 which is obviously slower than the 2.4 GHz Z3770 top of the line Bay Trail. The G/F 28nm HKMG is also of the efficient gate-first variety which took much longer to get right than gate-last and is not much bigger than Intel's 22nm. AMD did a good job though improving their fanless tablet chip at the same process size and shows that design is always the most important factor of any chip.
28nm FD-SOI: Samsung & ST's Major Opportunity
http://www.eetimes.com/author.asp?section_id=36&doc_id=1322399&print=yes
http://www.eetimes.com/document.asp?doc_id=1322353
Handel Jones, International Business Strategies Inc.
The foundry-fabless industry has demonstrated excellent performance in ramping 28 nm HKMG (High-K Metal Gate) technology into high-volume production. Additionally, both gate-last and gate-first technologies are giving high yields and reliable products.
Intel has high-volume production of its 22 nm Tri-Gate products, and its feature dimensions do not require double patterning. The second-generation Haswell 22 nm processor is also demonstrating high performance as well as long battery lifetime.
The next logical technology node is 20 nm HKMG, and TSMC is projecting 20 nm will represent 10% of its 2014 revenues ($2.2 billion to $2.3 billion) and 20% of its fourth-quarter revenues ($1.10 billion to $1.15 billion).
With the capacity of 60,000 wafers per month (WPM), the average price for 20 nm wafers in the fourth quarter at $1.1 billion will be around $6,000. This is a relatively large increase in pricing compared with 28 nm wafers, which sell at $4,500.00 to $5,000.00.
If TSMC achieves its projection for 20 nm, the company will have 95% of the 20 nm foundry market in the fourth quarter 2014.
There are, however, challenges in controlling leakage and gaining high yields of application processors and modems that require low leakage. If 20 nm does not provide low leakage with cost penalties compared to 28 nm, an alternate option is 28 nm FD-SOI.
The wafer cost of 28 nm FD-SOI is comparable to 28 nm bulk CMOS, with performance potentially being 15% superior to 20 nm bulk CMOS.
Consequently, Samsung Electronics has a major opportunity with its large wafer capacity to support low-leakage products with its 28 nm FD-SOI process. Cadence Design Systems, Synopsys, and Mentor Graphics are all supporting the FD-SOI ecosystem, and the transition from 28 nm bulk HKMG to FD-SOI should be inexpensive.
16/14 nm 3D technology status
Intel, TSMC, Samsung, and Globalfoundries are trying to ramp their 3D structures. Intel planned to have the initial volume production of its 14 nm Tri-Gate structure in the fourth quarter of 2013, but low yields postponed this ramp-up.
A number of fabless companies will tape out their 16/14 FinFET product designs in the third quarter of 2014 with high-volume production planned for the second or third quarter of 2015.
Tens of billions of dollars are being committed by the semiconductor industry to have volume production of FinFETs in the next 12 to 18 months. What is the probability of this happening? And is the next-generation FD-SOI (called 14 nm by STMicroelectronics) a better option for mobile applications, which are the key driver for high-volume wafers?
The cost to manufacture a 16/14 nm FinFET wafer is approximately $4,000 at the high-volume stage, and with a gross profit margin of 45%, the selling price is around $7,270. The key issue is whether high systemic and parametric yields can be obtained from the initial designs. Based on an assessment of many variables, the probability of this occurring is very low.
FinFETs will happen, but there will be a learning process, and multiple design interactions will need to occur.
With this scenario, Samsung's adoption of FD-SOI and implementation of shrinkages are very astute decisions. If its 14 nm FinFETs ramp up as expected, Samsung's competitive position will be very strong, and if there is a delay in this ramp-up, Samsung has an alternate solution.
There are cost penalties associated with having multiple options. However, with a 14 nm FinFET fab at 40,000 WPM costing approximately $6.8 billion, it is smart to be well positioned if there are delays in volume ramp-up. Additionally, the reward for success within the 16/14 nm, 10 nm, and 7 nm environments can be very high, where the electronics industry will continue to expand and need more semiconductors.
It is also important to be able to transition the smartphone vendor base to China and meet the aggressive pricing structures of the China market, which can be done with FD-SOI. Adopting FD-SOI gives a high probability of having a cost-competitive and low-power option for high-volume mobile platforms.
While the cost of failure is in billions of dollars, spending a few hundred million dollars to support an attractive option can give good financial returns. A critical need for Samsung is to aggressively promote multiple generations of FD-SOI while continuing with its 14 nm and 10 nm roadmaps.
Samsung's culture is to win at everything (targeting $400 billion revenues in 2018 to 2019), and FD-SOI can be a key part of winning the mobile platform war in addition to the activities in 14 nm FinFETs.
http://www.anandtech.com/print/7974/amd-beema-mullins-architecture-a10-micro-6700t-performance-preview
http://hothardware.com/Reviews/AMD-Beema-and-Mullins-Mainstream-and-LowPower-2014-APUs-Tested/
http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu
http://www.tomshardware.com/reviews/amd-tablet-processor,3813.html
From the Mullins previews it appears to me they have succeeded in getting Kabini like performance out of a fanless tablet. Intel will need Cherry Trail to combat this in gpu if they ever get design wins.
http://www.anandtech.com/print/7974/amd-beema-mullins-architecture-a10-micro-6700t-performance-preview
http://hothardware.com/Reviews/AMD-Beema-and-Mullins-Mainstream-and-LowPower-2014-APUs-Tested/
http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu
http://www.tomshardware.com/reviews/amd-tablet-processor,3813.html
If you were Safran with a 20 billion market cap why wouldn't you just come in a buy them after passenger qualification.
Including debt and diluted stock options that would cost them north of $150m. It would be cheaper just to sell their ETD products at cost to undercut IMSC and make their money on EDS.
"If our technologies are not improved sufficiently and Intel keeps improving its technologies, our customers' products will lose competitiveness to those of Intel. It's horrible to imagine the outcome."
"TSMC will stand behind our customers and cooperate with them. The battlefield between our customers and Intel is where we compete against Intel," he added.
Chang says TSMC's customers should diversify their foundry sources rather than rely only on his company. "All of our customers rely on TSMC in foundry production, and Intel relies on its own foundry plants," he said
http://www.wantchinatimes.com/news-subclass-cnt.aspx?id=20120104000020&cid=1102
Exclusive: Intel CEO promises Broadwell PCs on shelves for holidays
http://www.reuters.com/article/2014/05/18/us-intel-chips-idUSBREA4H08P20140518
(Reuters) - Intel Corp's next-generation Broadwell processors will ship in time to be used in personal computers sold during the holiday season but probably won't be available for back-to-school shopping, Chief Executive Brian Krzanich said.
Speaking to Reuters at the Maker Faire in San Mateo, California, on Saturday, Krzanich said that following setbacks implementing new manufacturing technology, announced last October, Intel is on schedule to ship its Broadwell chips to PC manufacturers.
The timing of new chip rollouts is important to Intel and manufacturers looking to sell new PCs with better battery life and higher speeds.
"I can guarantee for holiday, and not at the last second of holiday," Krzanich said in an interview. "Back to school - that's a tight one. Back to school you have to really have it on-shelf in July, August. That's going to be tough."
Krzanich's comments were more precise than during Intel's most recent quarterly conference call with analysts in April. On that call, he said he expected Intel's customers to launch Broadwell-based products in the second half of 2014.
The back-to-school season around August, when parents sometimes buy new PCs for their children, has provided smaller sales bumps for PC makers in recent years than in the past, but the November/December U.S. holiday season is still viewed as a key opportunity to sell consumer electronics.
I am glad you brought some factual realism there with a great post (good touch quoting both metal and gate pitch to give the complete picture). The foundries are as late with their processes as Intel is with their mobile designs, 1-2 years, but designs are easier to fix than process and their deficiencies can be masked with good processes i.e. smaller cheaper faster dies of the same design. Once Cherry Trail and Broxton are available in 14nm Intel would have cured most if not all of their design lagging in Windows/Android.
The fundamental point is that the foundries have collectively failed their customers by not allowing them to ramp 20nm designs this year when they want to and so many have had to redesign their cpu/gpu for 28nm again. This has nothing to do with Intel and foundry fans are just using Intel process comparisons as a strawman smokescreen. The foundries all seem collectively lost as how to proceed cost-effectively to smaller processes with the current lithography limits, a problem Intel solved years ago. If Intel was not branching to better on-die gpus and low-power mobile SoCs this would not be an immediate problem to them but they are and it is. Only the fittest smartest foundries will now survive under this competitive threat to their customer base business, Morris Chang was truthful to be very worried. The big bad lithography Bear is chasing everyone and currently Intel is the fastest and most safe of all the potential victims.
"The calls from Wall Street keep coming ..."
well that explains why their analysis is generally so bad if they are relying on foundrypumper as a reliable expert source .
The canaries (semi equipment companies) are all singing that foundry finfet yields are currently in the toilet and Intel's 14nm looks like it will beat the foundry's 20nm into service at least for all the merchant non-Apple guys.
No, what needs to happen is for you to stop regularly abusing other posters and without any provocation to remotely justify it. You have had your last final warning, no more hooligan behavior will be tolerated from you. Any post not remotely in line with the TOS will be gone in future.
Thanks but are they not now claiming ~5W TDP for performance that used to be available for much higher than that ?
I must admit to not having closely followed the details of AMD's 28nm low-power refresh chips. Originally I thought the only difference between the old was the introduction of standby but according to that author, radical drops of TDP have occurred for equivalent performance. Any truth in that ? i.e. does Bay Trail have a significant AMD competitor in its power class now ?
After all these years enough is enough.
Survey: share of time spent on iOS/Android connected devices
courtesy of this very hopeful fellow http://seekingalpha.com/article/2221823-amd-mullins-is-perfect-for-tablet-gamers-a-small-step-toward-my-2016-15-price-target
FWIW chipguy has documented buys on this board well under 20 (one at the 2009 trough of 12 no less) and one sale mid 20s.
what a complete "loser".
This is an unwarranted personal attack and it is also based on very faulty information. Is there any chance you can learn to converse civilly even at 61 ?
I reckon two out the next four quarters have a good chance of being at least EBITDA if not GAAP profitable. Next missions here ...
July 1 Delta 2 • OCO 2
Aug. 13 Atlas 5 • WorldView 3
September Atlas 5 • CLIO
Nov. 5 Delta 2 • SMAP
January 13 - DSCOVR (Triana) - Falcon 9 v1.1 - Canaveral SLC-40
January - MUOS-3 (TBD) - Atlas V 551 - Canaveral SLC-41
March - MMS (x4) - Atlas V 421 - Canaveral SLC-41
When you do those WebXPRT runs is there a way to compare each iteration's time with the others so as to get some granularity on any throttling behavior ?
always gets real wonky during the middle of the test.
That's why it's a good test of mobile chips because it will give an indication of throttling effect. It will be good to see the Intel-ARM comparisons with an identical browser. Thanks for your recent efforts.
p.s. my past criticism has only been in order to make you think deeper and longer and therefore be a better analyst as you could be really good in the end and you are getting there. I have toned it down though since your PM as you obviously had trouble with its direct robustness .
Have they finished yet ? lol
Well then run WebXPRT yourself using the same browser for both.
A15 is now 2.3 GHz on 28nm and yet Qualcomm are only claiming 2 GHz for 20nm A57. Does not look ok to me.
Did they not use the stock browsers for all devices ? The real beauty of WebXPRT is that it runs forever and thus will smoke out the hot chokers. You should not be so quick to dismiss without a valid reason.
and most of the improvement in Geekbench3 is due to the new crypto instructions in ARM64. The fact of the matter is that on the same process A57 will perform about the same as A15 because it won't be able to clock as high. These ARM cores have been overdeveloped for some time now for their intro silicon process sizes.
p.s. hey "WebXPRT" rocks !!
Intel will be on the cusp of a transition to 10nm which again gives us more than a 2x shrink.
That is the crux of the ARM foundry stagnation, not really 14nm but 10nm. Where is their 14nm BEOL for them to stick 10nm FinFETS on ? Does it exist even in theory anywhere ? 2016 is judgement year for the foundries when Intel's 10nm hits the shelves and it is only 2 plus years away. 10nm Skylake and Goldmont are going to wreak a path of havoc across their competition then if the 14nm versions have not already done so.
http://techreport.com/news/26460/entire-radeon-r9-lineup-finally-available-at-list-prices
It's a little crazy that this is newsworthy, but AMD's full lineup of Radeon R9 graphics cards is finally available online at suggested e-tail prices. Various members of the family have been in short supply and selling at inflated prices since last year. Supply constraints and increased demand from cryptocurrency miners were apparently to blame, but the situation seems to have been resolved. AMD says the "madness" is over—and that the current, intended prices will remain stable moving forward ...
I assume any company that goes down the micro-server route is energy limited rather than dollar limited. If you are dollar limited you would not waste your money on exotica like that but get the best general purpose server within your budget.
It's the bizarre flaw that he insists they have the same socket so artificially slugging the x86 part. Only the dunces at AMD could think this was a good trade to make.
They seem to think that their Skybridge ARM chips will be higher performance than their x86 ones on their first in-house design.
Makes a lot of sense because after all ARM has all that high-performance related software already written for it ... oh wait . AMD ... finding new ways to shoot themselves in the foot.
http://www.fool.com/investing/general/2014/05/13/intels-silvermont-and-arms-cortex-a15-face-off.aspx
A bit more relevant than Geekbench 3 I suspect .
p.s. J1900 has 2.42 GHz turbo
http://ark.intel.com/products/78867/Intel-Celeron-Processor-J1900-2M-Cache-up-to-2_42-GHz
The K1 is at 2.32 GHz
http://www.phoronix.com/scan.php?page=article&item=nvidia_tegrak1_preview&num=1
http://www.phoronix.com/scan.php?page=article&item=nvidia_tk1_amdam1&num=1
I can see the theoretical case but in practice Silvermont only matches LV Xeons in performance/watt and that is the best of the micro-server bunch. Maybe with hyper-threading it could have exceeded it but there would not have been much in it. ARM or another RISC have to actually deliver what they promised, much better performance/watt than x86 but it is just not going to happen with the known designs on the known foundry roadmaps.
p.s. on an unrelated matter these new Haswells might spark more desktop interest now they have restored the Sandy Bridge TIM or one of equivalent performance. Haswell can now show its advanced efficiency even at the desktop level.
4.4 GHz, Intel listens to Haswell TIM criticism and unlocks more clockspeed ...
http://www.forbes.com/sites/antonyleather/2014/05/12/intel-to-release-its-first-4ghz-processor
If information uncovered by Chinese website EXPreview is to be believed, Intel's soon-to-be-released replacement for its current Intel Core i7-4770K CPU will be the company’s first that ships with a clock speed of 4GHz.
The new CPU is part of a range of refreshed chips due to be launched some time in June, with the two high-end K-series models codenamed Devil’s Canyon called the Core i5-4690K and Core i7-4790K, with the latter offering the magic 4GHz number. Using Intel’s Turbo Boost technology, the CPU will actually increase this figure to 4.4GHz when under load for extra performance.
The cheaper Core i5-4690K will apparently offer a default clock speed of 3.5GHz and will Turbo Boost to 3.9GHz. As I reported here, the new CPUs will also feature an improved thermal interface between the CPU core and heat spreader. Heat has plagued Intel’s Ivy Bridge and Haswell-based processors such as the Core i7-3770K and Core i7-4770K, with many people removing the heatspreaders and using their own thermal paste.
Combined with higher clock speeds, it’s quite possible that the new CPUs could easily reach 5GHz with a good cooler if you overclock them – something that hasn’t been widely possible since its Sandy Bridge-based processors such as the Core i7-2600K.
http://www.forbes.com/sites/antonyleather/2014/03/19/intel-unveils-2014-roadmap-4-fantastic-new-processors-for-pc-enthusiasts/
One of the major bugbears of Intel’s current and previous generation processors is the fact they run hot, especially when overclocking. Enthusiasts have blamed poor thermal interface material between the processor core and heatsink, and Intel has apparently addressed this with the refresh of its current Haswell-based CPUs.
Delidding a Haswell processor With Ivy Bridge and Haswell processors running so hot, especially when overclocked, enthusiasts have taken risky steps and removed the heatsink to expose the processor core and applied their own thermal paste
The new unlocked processor, codenamed Devil’s Canyon, will be available sooner than Broadwell in mid-2014, and Intel states it will use ‘improved thermal interface material’. This could improve overclocking and lower operating temperatures – potentially a huge deal for enthusiasts.
The final product that should be hugely interesting to enthusiasts is that of an Intel Pentium Anniversary Edition-labelled processor. The most fascinating feature with the new product is that there will be an unlocked version allowing you to overclock it to get more performance.
intel unlocked pentium
Intel hasn’t allowed overclocking on its budget range of processors for several generations and this could provide a huge boost to the enthusiast scene. At the moment, the only way to achieve significant overclocks is with a K-series CPU such as the Core i5-4670K. However, these are expensive, so in the past, it’s been very popular for enthusiasts to buy a cheaper processor and overclock it to the same speed as a more expensive model.
How far the new unlocked Pentium can be overclocked will be fascinating to see but this could herald a new age of enthusiast products from Intel that don’t just focus on the $200+ segment.
Re: http://www.fool.com/investing/general/2014/05/11/should-ibms-power-8-worry-intel.aspx
The other thing you should be aware of is that a fair amount of IBM's services/software sales are connected to their Mainframe and Power hardware (Oracle has the same deal with Sparc). The buying cost of such chips can therefore be artificially varied to keep their software ecosystems alive i.e. sold at a loss if necessary. The most immediate danger to Intel is whether Google like its performance/watt enough to use it in volume because the cost can be artificially varied to seal the deal.
Don't assume that just because Power 8 per thread performance is poor when they partition the core to do 8 threads that when they don't it will remain poor. Power 8s can be set up to have 1,2,4,8 threads per core depending on user preference and I expect when there are less threads the thread performance will go up. The Power server architecture is not known for poor single thread performance when there is just a single thread running and Power 8 is beefier than Power 7.