Mannerisms

TSMC 28nm lead the semiconductor industry?http://techon.nikkeibp.co.jp/article/HONSHI/20090225/166326/fig6.jpg

Intel’s giving TSMC everything they need, including process, IPs, libraries, and design flows.----Intel would port its Atom processor CPU cores to the TSMC technology platform including processes, IP, libraries, and design flows.(the process, IPs, libraries, and design flows is tsmc's )Intel not giving TSMC anything. http://www.intc.com/releasedetail.cfm?ReleaseID=368940 Joint Intel/TSMC announcement
Message from Kevin Sellers, VP Investor Relations
----This is not a capacity outsource deal nor is this a process technology transfer. As stated above, Intel designers will port the Atom designs to a leading-edge TSMC process to allow Atom processor cores to be coupled with added TSMC intellectual property to create Atom-based SOC solutions

Intel emphasized that the MOU is not a licensing agreement of the Atom core to TSMC's customers and that Intel does not plan to transfer its manufacturing-process technology to TSMC

IT'S TAIWANESE SAID---Intel processor technology requirements are complex, more graphics, it will continue to be made of advanced process technologies,Intel had a two to three to lead on the rest of the semiconductor industry IS TURE

At IEDM, Intel 32 nm program manager Sanjay Natarajan will outline a 32 technology with drive currents of 1550 µA/ µm at 100 nA off-current for the NMOS transistor, and 1210 µA/µm for the PMOS transistor.TSMC's 28NM with drive currents of 1360 µA/ µm at 100 nA off-current for the NMOS transistor, and 960 µA/µm for the PMOS transistor,WHO will win the race to the next process node ?

Intel Maintains 32nm Lead
Intel Corp of the US was the first firm to adopt HKMG, in a 45nm-generation microprocessor, and is still in the lead when it comes to 32nm. The SRAM presented by the firm has an On current of 1.55mA/um at 1V supply for NMOS, or 1.21mA/um for PMOS (Paper 27.9), described by one engineer in the field as "...astonishingly high." Yield, according to a source at Intel, "...is improving right on schedule, two years behind the prior generation, as always." The company plans to begin volume production of microprocessors using 32nm technology in the second half of 2009.

Intel's technology is very interesting for more than just performance, because it offers a path to 22nm-generation manufacturing. Intel is the only major semiconductor manufacturer to use the "last gate" technique. The poly-Si gate electrode is formed first, then the source and drain electrodes, and then finally the gate electrode is replaced with a metal gate electrode. Most other firms use the "first gate" technique, directly forming the metal gate before the source or drain.

The first gate method is very compatible with existing manufacturing technology, and is generally expected to facilitate the transition to smaller geometries because of the simplicity of the processing.

Intel has said that it uses the last gate approach because it can deliver the maximum effectiveness for strained Si technology. The company has previously, however, admitted that the technique poses problems in fabrication ease, ease of transition to 22nm and other points. This time, though, Intel claimed that it has "...overcome the problems of the last gate technique."

Yes, RRAM, but TSMC is going for a full node at 28nm - some time before Intel will get there.

some time before Intel will get there.--because intel isn't goinging to such as 80nm/55nm/40nm/28nm.so Meaning there is no comparison

CPU complicated process than the SOC, in fact, Intel so that other manufacturers have a long history of manufacturing, in addition to communication chips, before the South Bridge chipset also so. But the core product - the processor is unable to let others foundry and process demands. Processors require too much production capacity, and technology than other types of chipsets, which is the principal reason. If Intel's chips to reach the quality requirements, and other non-core chips to enable the industry to OEM manufacturers of course we will consider it.

MARK BOHR: I think you get the latest news is not news because TSMC is now lagging far behind the technology of Intel. They do not have them to the factory 40-nm products, but also according to the report to know that they will not be 40 nm products just mentioned, the use of high-K gate dielectric and metal audit, but they clearly put forward a few weeks ago, even in 32 nm technology is not high-K and metal gate, has been to wait until the 20 nanometer generation applications can be. So look at it this way than Intel these companies leading in technology is still very deep and more.

Mark Bohr: Clearly, a significant difference between the two modes. But I think this model is to serve two different markets and different uses. Foundry model suitable for smaller enterprises, the output is not, it is difficult to self-developed technology companies. I would like for these companies, this model has been very successful. But for Intel, since the production of our high technology to develop their own it is very necessary. And it also gives us a particular advantage because we can better process and product design to match the achievement of optimal. It also allows Intel to other companies faster than the pace of the introduction of new products. I would like to Intel in the development of any technology node has been a significant leader in TSMC. However, they are still a good company. They are ambitious, and will continue to make progress. But I think they are unlikely to be the volume of products, reduce the product of similar size and technological innovation HkMG to catch up with Intel.

Well RRAM, I think I come at this from a different direction to you. I don't see it as a wonderful technical advantage to get to High K or metal gate earlier than others. I see these techniques as an unfortunate added complication which companies have to adopt to keep their process technologies effective e.g. to try and hold down leakage.

Mark Bohr:: the first I have to admit that our 45nm technology is not exactly in any dimension is 45nm. In some smaller dimensions, such as gate length of only 35nm, while in the other dimension may be greater than 45nm. But it seems that other companies, Intel and the industry for each generation technology uses a set of common nomenclature, so we are 90nm, 65nm and 45nm to show that the current generations of technology. Nevertheless, not all companies are using this name in size. Therefore, it can be said that not all the same 45nm, the size of some companies may be more lenient definition. But this common nomenclature is reflected in every generation over the previous generation of approximately 0.7 × reduction in the size, but often choose the whole throughout the five or 10 numbers that only approximate.

Do you mean there is no significance HIGH-K?

Panasonic to ship 45nm chip ahead of Intel
UniPhier video processor out before 'Penryn'

By Tony Smith • Get more from this author

2nd October 2007 14:53 GMT

Panasonic is set to become the first technology company to release products that consumers can actually buy and take home that contain 45nm chips, its parent, Matsushita, claimed today.

The Diga-brand DVD and Blu-ray Disc recorders containing the 45nm chip, a video processing part called UniPhier - short for UNIversal Platform for High-quality Image Enhancing Revolution, believe it or not - will go on sale on 1 November, albeit only in Japan. Even so, that's still more than a week ahead of Intel's first 45nm processors, assuming they're available to buy at launch.

UniPhier is a 250m-transistor codec chip supporting the H.264 - aka MPEG 4 AVC - capable of dealing with two "full HD" - by which we assume Panasonic means 1080p - images at the same time.

Panasonic also said the part can churn through 3D graphics and has the ability to number-crunch encryption alogorithms - though it didn't specify which ones.

The company sees UniPhier as the foundation for a range of products, from media-oriented mobile phones to in-car entertainment systems. There are three versions of the chip. One, for phones, contains just a basic "instruction parallel processor", while the second model, aimed at portable AV devices, adds a "data parallel processor" and what Panasonic calls a "hardware engine" - specific algorithm accelerators. The third UniPhier is designed for home entertainment products and in-car kit, takes the second model's design and builds in more hardware engines.

All these cores fit into system-on-a-chip parts that incorporate an ARM processor, a memory controller, I/O, video signal processing circuitry and, in some cases, 2D and 3D graphics processing.

Matsushita Electric Industrial Co., Ltd. started mass production of System-on-a-Chip (SoC) for use in digital consumer equipment based on the 45-nm process technology in June, 2007.

A product based on the next-generation UniPhier platform for digital appliances is now under mass production. Two more similar products will be added within 2007. The company claims that this is the industry's first mass production of 45 nm-generation SoC.

"The result of vertical integration and cooperation"

The volume production launched on the company's 300-mm line in Uozu, Toyama Prefecture in Japan. The line produces 6,500 65-nm SoCs per month. The production scale will be increased to over 10,000 pieces per month by employing the 45-nm compatible process including immersion ArF lithography.

"We are confident that the yield of our 65-nm SoC production is the highest in the world," said Susumu Koike, the company's executive vice president. "We intend to achieve the same level in the production of 45-nm SoC."

According to Panasonic, the 45-nm generation product can reduce the chip size and power consumption by 39% and 30%, respectively, compared with the 65-nm product.

Koike explained the reason why the company gained the lead in 45-nm mass production as being "the result of vertical integration and cooperation." Through cooperation, the company has jointly developed elemental technologies with Renesas Technology Corp., as was the case with the preceding generation, and utilized the development results at IMEC in Belgium.

Panasonic was the frontrunner in the mass production of 65-nm generation consumer SoC as well.

"Application software creates" demands for the 32-nm SoC

Reiterating the company's firm intention toward mass production, Koike asserted, "Demand will definitely be created through the initiative of application software," regarding the 32-nm generation SoC that following the 45-nm generation.

The company plans to reduce power consumption and enhance the design flexibility of the 32-nm generation SoC by the integration of reconfigurable logics. Mass production is scheduled to begin around 2010. The company intends to employ new process technologies such as high-k gate dielectric and metal gate.

Panasonic is leading the industry can be said? Consumer electronics giant Panasonic said it will "shortly" announce a transition to a 30-nanometer CMOS process for its homegrown digital consumer SoC platform called UniPhier (Universal Platform for High-quality Image Enhancing Revolution), a company spokesman revealed at Panasonic's booth during CEATEC Japan, a consumer electronics show here this week.
If true, Panasonic's aggressive plan for a 30-nm process at its own small Uozu fab in Japan is news. But this is not the first time for the Japanese company to pull such a stunt.

When Panasonic began in June, 2007 the world's first mass production of a 45-nm process system LSIs using 300mm wafers at Uozu, the Japanese company surprised the semiconductor industry, including Intel Corp. (see: Matsushita might not be getting the respect they deserved with a 45nm process obliterated by Intel's shadow)

Panasonic's move, however, seems puzzling -- or even counter-intuitive -- to many observers, especially at a time when most SoC vendors in the world have already moved on to a fabless or fab-lite model.

Industry analysts question how long Panasonic can keep making semiconductors at its own fab. Panasonic will need to buy advanced lithography equipment and other fab tools when the company's fab goes to 30-nm.

Panasonic's semiconductor move also draws a sharp contrast to its Japanese rival Sony Corp.

Sony in the past built larger-scale fabs for its game console chips and other ASICs. This conformed to Sony's longstanding dream, which is yet to come true, of becoming a genuinely vertically integrated company.

Today, Sony has virtually subscribed to a fab-lite model. Sony sold its high-end fabs to Toshiba and joined IBM's "fab club" for Silicon on insulator (SOI).

Panasonic, on the other hand, has always kept a low profile as a chip company. The company has kept the scale of its own fab relatively small and has remained quiet about any design wins for its UniPhier SoCs. Still, Panasonic is clearly a niche player in the chip trade compared to NEC or Toshiba in Japan.

Panasonic has Renesas Technology Corp. as its process partner. The two companies announced the deal last year. Renesas has been collaborating with Panasonic in advanced process development in terms of defining design rules and design process.

Panasonic's spokesman, however, added, "We work with Renesas, but we make our own chips at our own fab."

It's clear that Panasonic holds high hopes for its own UniPhier platform.

The platform consists of five blocks including UniPhier processor, CPU, stream I/O, memory controller and AV I/O. Typically, a UniPhier processor integrates instruction parallel processor, data parallel processor and hard engine-based accelerator.

Last week, I mentioned that Matsushita might not be getting the respect they deserved with a 45nm process obliterated by Intel’s shadow. I also should acknowledge that I was one of the doubters and expected Intel to not only get to 45nm first but with a considerable lead on second place. Well it turns out that the speculation by Engadget and others coming out of CEATEC was correct. Matsushita (or Panasonic) not only has a real 45nm logic process, but they beat Intel to the market!

Matsushita doesn’t bother with high-k gate dielectrics or metal gate electrodes at 45nm, but they achieve the transistor packing density of the latest technology node. In fact, the Matsushita process beats Intel’s tightest metal pitches. The DVD decoder chip is a complex SoC with over 300 small SRAM arrays scattered around the die. A compact die size of 68 square millimeters certainly would not be possible without a small bit cell design, and Matsushita’s SRAM cell size matches up with Intel. With slightly tighter than 140nm pitch at metals one through four, Matsushita actually has a slight edge over Intel’s 150nm observed pitch.

The UniPhier SoC is truly built to reduce silicon die area and cost. Believe it or not, Panasonic uses it in a video player with a VHS tape bay. That’s something old along with the new in the DMR-XW200V Blu-Ray player. There could be something borrowed as well, but we are still analyzing the device, and I’m not a lawyer.

Finally, let me extend a sincere apology to Matsushita for underestimating their prowess in process technology as well as a hearty congratulations for being the first manufacturer of 45nm logic technology.

No I don't mean there's no significance in high-K, RRAM, I'm saying that high-K is a way around a problem - not something which is adopted until it has to be. i.e. if your process delivers high performance and low leakage without high-K, you'll do without high-K. That's why I don't see the use of high-K as a measure of how technically advanced a company is - I see it simply just as useful technology for keeping a process useful which might otherwise have reached its useful limit.

When it comes to manufacturing Atom (tm) at State of the Art geometries, the question is how many can Intel economically afford to fabricate. That answer has to be a lot. From an accounting standpoint they're just not very profitable to produce.

Mike Bruzzone
Camp Marketing

Mike, To me it seems that a manufacturing machine built to turn out $100 parts doesn't turn out $10 parts very profitably, and that's why Intel went to TSMC, cheers, David