ARM chips in 20nm next year?

The fineness of engraving of the ARM processors should fall to the 20nm to the advance on the future of mobile processors from Intel.

Manager General of division among ARM processor, Simon Segars, announced the possible appearance of SoC with 20nm for the next year.

While fleas in 28nm start only to democratize, conversion to a fineness of engraving 8nm finer, will require at least one year of development.

The technological constraints are only part of the problem. Intel which made large research and development of its plants not to leave down so easily. The Blues already provide chips Medfield 22nm for 2013 and 2014 14nm.

At this moment at Computex, the ARM processor-based solutions and specifically laptops are more likely to be presented. The British ARM chips therefore are a real success. This trend requires first to foundries such as TSMC to invest more to increase their production capacity, but still we think that such chip with a lithograph to 20 nm is quite feasible for 2013.

For memory, since the beginning of the year, 1.9 million of ARM chips shipped 1.1 million intended to smartphones and tablets. A figure which should not only increase if announcements that we follow for the 2012 Computex in Taiwan.

Read More

NVIDIA would that Intel seriously its ARM chips

By producing his ARM chips in Intel plants, NVIDIA could take a decisive advantage, but the first person not the ear hears him.

According to XbitLabs resumed our colleagues from Tom's Hardware, NVIDIA would currently seek to push Intel to burn ARM chips in its factories. It would in any case be the desire of the NVIDIA CEO who saw an opportunity to improve the process of engraving of ARM chips while casters such as TSMC or GlobalFoundries have not yet perfectly mastered the process of engraving on 28 nm, while Intel is poised to launch its tri-gate processor engraved 22 nm.

Engraved 22 nm, ARM chips could indeed be faster, consume less energy and release less heat... but also more competition to x 86 Intel processors. However, it is known that Intel has produced ARM chips for its own account in its factories in the past. However, Intel responds in the negative to these rumours, explaining that he has well keep its advance in the matter and exploit its know-how on 22 nm to impose its own products which is logical. Especially as Intel grows its new processors Medfield, destined for competitors of the SoC ARM and mobility markets.

NVIDIA cannot a priori rely on Intel to help pass Qualcomm and its SnapDragon S4 which are engraved 28 nm. It is also the ARM chips to take advantage of the more fine etching, since the Tegra 3 (NVIDIA) and Omap 4 (Texas Instruments) are engraved in 40 nm by TSMC, while the Exynos (Samsung) and A5 (Apple) are engraved in 45 nm by Samsung.

Read More

Intel Medfield: some news in x 86 mobile chips

Intel processor x 86 ultra fast and efficient energy, what invade the world of mobility by the large door?

ARM, the processor architecture that is imposed by mass in our mobile devices, such as our smartphones or our digital shelves. Low, offering more performance, these small SoC are now everywhere, what begin to shake giants like Intel smelter which would well a place in this market (its Atom x 86 there is being did not place).

And although, according to the scientific journal of MIT Technology Review, processors x 86 dressed for the mobile age would be on the show, and perhaps even at the CES in Las Vegas next January. According to the magazine, tests seem conclusive with the Android operating system equipped with mobile devices and offer good performance over the top three chips mobile of the current market. These Intel chips, named Medfield, also have a consumption well below what is currently on ARM, without to give precise figures, still under NDA.

We therefore look forward to learn more and to know the performance of such chips integrated into mobile terminals. Processors that may also have their interest with the release of Windows 8, even if the new OS from Microsoft will be proposed in the ARM version for the first time.

Read More

What chips are hiding in the HD Radeon 7000M and 600 M GeForce?

Changes of names without change of chip generalize for the two giants of the graphics card.

Historically, AMD and Nvidia have always first to ensure that demonstrate the power of their graphics processor to then work on mobile versions of these chips. For example, as the Cypress (Radeon HD 5800 Series) followed by the Redwood (Radeon HD 5600 Series) in AMD/ATi or the GF100 (GeForce 470 and 480) then out GF106 (GeForce 450) at Nvidia. Mobile chips which often marks the end of the development of an architecture optimized to ensure that it consumes still less energy.

As indicated you yesterday, the release of the Radeon HD 7000 will intervene in the coming weeks, it is therefore logical to think that their mobile versions will arrive after. But is not the case this time, the Sunnyvale firm wishing to do the opposite for this series. Last September, a first demonstration was made on a mobile version of the new architecture incorporating a chip engraved in 28 nm. This confirms the will of AMD to be first on the market of mobile phones.

To understand this situation, he must see that TSMC's time to finalise the method of production of 28 nm chips and started only to achieve good yields (according to the skier). You do the same calculation that we, this glue not really with the AMD roadmap which appeared on the Internet, pointing the output of the HD for 7000 series in three weeks. It is therefore a question of rebadging for these new mobile chips, which integrate processor series HD 6000 to ensure this tight planning. The GPU that will be beautiful and well engraved 28 nm... which should simplify the task to the skier, able to get out of the HD Radeon 7000M in times.

What is the most frustrating here, it is that the Radeon HD 5000 have been already rebadgées in HD 6000 in some cases without any performance gain. On the training games, the next generation of top-of-range GPU at 28 nm (Tahiti) move the VLIWs (Very Long Instruction Word) while the previous architecture (Trinity) will still use version 4 of these VLIWs derived from the 6900 HD, the series Cayman. And this while at the same time some 7000 HD will retain compatibility with version 4 of these instructions to be compatible in CrossFire with the Trinity versions. Even if the version 5 of these instructions is likely to go in the trash to keep this homogeneity, these modern training games could ensure a rise of the GPU with the arrival of Direct3D 11. Even if the version 4 and 5 are and will be compatible D3D 11, the next video games more rely more on these new APIs, which could disrupt the next series of AMD, without however creating a gap as important as in the past... Finally we hope.

Table of the Radeon HD mobile our colleague AnandTech

To pass the range HD 7000M 6000M, we can expect to see land three or four models displaying a clock frequency a hair more high and operator of the not expensive GDDR5. 7500M HD can be interesting on paper, with its 480 processors flow - either as well as on the HD 7600M version, but with a 64-bit memory bus interface. What should impair its performance even if the GDDR5 on 64-bit bus could potentially be as fast as the 128-bit GDDR3. We will therefore wait to know the practical performance of these cards to bring to trial them.

To summarize, a third rebadgage of cards in a row is expected for the input and the mid range of the Mobile Radeon HD...

The firm to the Chameleon was silently for its 600 M series. We therefore discover a GeForce 600 M which replace a 500 M without so far have reviewed frequencies increase. In detail, Nvidia has renamed its GeForce GT 555 M GT 635 m... same processor, same number of texture units, same frequency, short, everything like this. Except for the GDDR5, who will have right has a small thrust on the frequency from 3140 GHz to 3600 MHz, providing a little more bandwidth.

In addition, renaming drops new maps down to leave the place to the new environment maps and high-end as the GT 640 M, or even the GT 650 M, even if nothing was announced for the moment. For the price, we also hope the descent of the entry ticket to the same card but with a new name...

Table of our colleague AnandTech for Nvidia cards

As for AMD and its Radeon HD 7000M maps, the 600 M GeForce seem for the moment that consist a marketing repositioning products to make room for new architectures. Nothing very interesting, and we are much more eager to see what future 7970 HD Radeon and GeForce GTX 680 (non-final names) were in the womb. Maps which later gave birth to real new mobile GPUS. As usual, the situation is very difficult to readable for the consumer who has printing buy - wrongly - a map of new generation. Which is always harmful...

Read More

Core i5-661, i3-530 and some chips LGA775 will be stopped

Intel program it stops the production of Celeron several of its processors, i first generation Core, a Pentium and two.

After yesterday, have formalized the decision of the X 58 Express chipset, Intel today confirms the end of view of many of its processors on socket LGA 1156-LGA 775. This is the Core i5-661 (3.33 GHz - 4 MB cache) and Core i3-530 (2.93 GHz - 4 MB cache), but also of the Pentium E5700 (3 GHz - 2 MB cache), Celeron E3500 (2.7 GHz-1 MB cache) and E3400 (2.6 GHz-1 MB cache).

The two Core-i will remain available to the command until April 27 and their delivery will end on October 5, 2012 for the OEM, while versions boxes will be sold up to of inventories.

With respect to orders for the Pentium E5700, E3500 Celeron and Celeron E3400, they will be recorded until December 30, 2011 and deliveries will take place no later than June 8, 2012. Here again, versions boxes will be sold up to of inventories.

Read More