We heard rumors last week that Intel’s Ivy Bridge had been delayed and now it seems they have been confirmed. Intel’s Sean Maloney (executive vice-president and chairman of Intel China) spoke to the Financial Times over the weekend and it looks like the launch of Ivy Bridge has been pushed back. Originally set for an April launch, Maloney says June is now looking more likely.

The Financial Times cites Maloney as saying that the start of sales of machines based on the 22nm Ivy Bridge has been pushed back. “I think maybe it’s June now,” he’s quoted as saying.

However, while Maloney says Ivy Bridge is now looking at a June launch, others say that while it has been delayed, it’s nothing as significant as two months. Forbes’ Patrick Moorhead reached out to both industry sources and his contacts at Intel, with both returning confirmation that while Ivy Bridge has been delayed, it is only a short delay. Industry sources said Ivy Bridge would be available when it counted, for the back to school season, while Intel’s Jon Carvill said that the Ivy Bridge schedule had only been impacted by ‘a few weeks’ and things were still on track to be ‘in-market for spring.’

It seems that the Ivy Bridge delay is certain, but the impact this delay will have on launch is not completely clear. We reached out to Intel for comment and the company responded, informing us that Forbes has the correct information and the delay really is only for a few weeks.

“The piece on Forbes has the correct information: our 3rd generation Intel Core launch is delayed by a few weeks, and we remain on track for our Spring 2012 launch timeframe,” Intel’s Dave Salvator told Tom’s Hardware.

If Intel hadn’t specifically told Forbes that reports of an eight-week delay were inaccurate, we would be reminding you that summer doesn’t officially start until near the end of June, but considering Intel did specify that the delay would only be a few weeks (and not eight), we’re left wondering if the chairman of Intel China had his facts mixed up.

With all the buzz surrounding Barcelona’s MWC 2012, AMD quietly announced the release of two new FX processors with Bulldozer core. The company will be gradually rolling out the FX-4170 as well as the FX-6200.

The quad-core 4170 breaks through the 4 GHz mark with a base clock speed of 4.2 GHz (4.3 GHz Turbo) and complements the 4100 (3.6 GHz/3.8 GHz) on the lower end of the FX series. It is AMD’s first processor that offers a base clock speed of more than 4 GHz. The 6-core 6200 will come with a base clock of 3.8 GHz (4.1 GHz Turbo). Both processors are rated at a TDP of 125 watts. The manufacturer did not reveal the prices of these two CPUs, but mentioned that their availability “is happening on a rolling timeline, so availability will vary by market.”

With the introduction of these two new processors, AMD has also reduced the price of its FX-8120. The chip is now offered for a suggested retail price of $185, down from $205.

NVIDIA’s cooked up a few ways to describe the Tegra 3′s quad-core-with-a-spare architecture, usually by giving the extra Cortex A9 a cute nickname like “ninja,” or “companion.” Until now, the proper description was “Variable Symmetrical Multiprocessing,” or, vSMP for short. Despite how much fun (and technically accurate) some of these descriptions may have been, however, they just aren’t marketable. “Our customers wanted a name for it that’s unique and descriptive,” writes mobile business unit general manager Michael Rayfield, “A name they could put on a box or a store sign that immediately represents its value.” That official name is the 4-PLUS-1 quad-core architecture, he says, and you’ll probably see it pop up a few times in Barcelona next week if LG’s latest offering is any indication. It lacks something in pizzaz, to be sure, but we’ll admit that it is at least descriptive of the Tegra 3′s technical chops. In related news, NVIDIA promises the Tegra will be less fickle about its new moniker than the symbol formerly known as the artist formerly known as Prince.

SOURCE via Nvidia

AMD’s Trinity APU can do some remarkable things, but we still don’t know exactly what magic ingredients make its Piledriver cores superior to the tepidly received Bulldozer. Now though, a firm called Cyclos claims it’s supplying ‘resonant clock mesh’ power-saving technology for use in the new module. In speaking to the media, it’s revealed that this will help to enable a “4+ GHz” factory clock speed, which sounds high if it definitely refers to an integrated chip with low-power credentials. As for the resonant clock mesh itself, it’s a bit like KERS for processors: it recycles clock power instead of letting it dissipate and thereby enables higher clock speeds in “next generation SoCs that also require ultra-low power consumption.” We also know that the technology is financially backed by ARM and Siemens and has seen precious little implementation prior to AMD — which is fine, so long as all that resonance doesn’t make our rig hum even louder.

SOURCE via The Inquirer

According to august engineering journal the EE Times, solid state storage specialist SanDisk is working on bringing a new form of system memory known as ‘ReRAM’ to market, a fact apparently revealed in a job ad posted on the company’s website last week.

Why is this interesting? Because ReRAM, or ‘resistive RAM’, is one of those disruptive technologies like wireless power or 4G that’s been under development for years, but is suddenly looking like it might happenin our lifetime. ReRAM could, if manufacturing kinks are worked out, replace both system RAM and hard drives in PCs of the future.

ReRAM is being touted as the successor to both current system memory chips and hard drives. And it’s not just SanDisk that’s been showing an interest in this super-storage of late.

The importance of ReRAM is that it combines the read/write speeds of DRAM – the stuff on those memory chips in your PC – with the non-volatile nature of flash memory. In other words, they don’t forget stuff when you turn the power off. Potentially, it’s the best of all existing technologies: bulk storage that runs at system memory speeds.

Theoretically, a PC with enough ReRAM memory in would mean no more loading times, ever. All your data would always be kept in the equivalent of today’s system RAM.

On top of that, ReRAM is expected to require little power, and it can be manufactured on tiny manufacturing processes with high yields.

The principle of the ‘memrister‘, upon which ReRAM is based, has been around for 40 years. Memristers work by ‘remembering’ the amount and polarity of current that passes through them as electrical resistance, which enables them to store data in ReRAM. HP believes that memristers have other potential uses as logic gates and switches that “could one day, for example, act like synapses inside computer circuits, mimicking the behavior of neurons in the human brain”.

Interest in memristers and ReRAM has spiked in the last few week, moving from highly specialised news feeds to mainstream tech sites including the BBC. Up until now, ReRAM has been most closely associated with HP which has led development of the technology. In January, however, Japanese firm Elpida which put out a press release that claimed it had a working ReRAM protoype that was capable of switching state at 10 nanoseconds – the same as DRAM.

Then, at the start of this month, ‘Technology Licensing Company’ Rambus – a well known player on the patent district circuit – announced it had acquired Unity Semiconductor for $35m. Unity’s CMOx technology is another candidate for the grand unification of memory theory, which includes patents that cover memristor interfaces too.

Other familiar names who are working to commericialise ReRAM include Sony, Panasonic, Micron, Hynix, DSI and more.

Expect a lot of news about ReRAM over the coming months. Hynix and Elpida both reckon they’ll have ReRAM-based chips in production sometime next year.

While the company wouldn’t reveal specific plans, Dell did say it’s interested in entering the server arena with an ARM-based product. Earlier this year HP debuted its first ARMv8-powered server, but Dell seems to think the launch was a tad rushed. As it said during the Q&A session of its earnings call today, it’s “been waiting to make an ARM server for a year, but the software stack isn’t mature yet.” It is, however, intrigued and sees “some interesting opportunities emerging.”

It’s little notches like these that could eventually carve out a big Intel-shaped niche in the mobile universe. What you’re looking at is a prototype chip codenamed ‘Rosepoint’ that somehow crams a digital WiFi radio and a dual-core Atom CPU onto the same piece of silicon. Interference would normally make such proximity impossible, but Rosepoint incorporates new anti-radiation and noise-cancelling shielding to prevent the components from corrupting each other. The aim isn’t just to shrink everything, but also to deliver “state of the art power efficiency” by removing unnecessary circuitry. Intel even claims it can fit the RF antenna onto a chip too, but it doesn’t want to show that off just yet.

SOURCE via Wired

The cost of entry to the LGA-2011 party just tumbled by around $250 thanks to Chipzilla’s Core i7-3820 processor. It’s a proper Sandy Bridge E processor with PCIe 3.0 support and more PCIe lanes, more memory bandwidth and room for more RAM compared to older platforms, but of course it’s also lower specced than the pricey 3960X and 3930K. It ‘only’ has four cores (and eight threads), 10MB of L3 cache and it isn’t fully unlocked — as denoted by the lack of a K or an X in its title. Priced at around $300, it looks like a steal when stacked up against an LGA-1155 cousin like the $332 Core i7-2700K, which has the same 3.9GHz base clock speed, 8MB of L3 cache and none of the added benefits of Sandy Bridge E. However, once you factor in the cost of an X79 motherboard and perhaps also a new cooling solution, Intel’s pricing starts to make more sense.

SOURCE via The Inquirer

SanDisk is hauling its consumer SSD range into the SATAIII era, offering a pair of Extreme branded drives that deliver performance broadly on-par with other 6Gbps rivals. The 120GB and 240GB variants are available now priced at $190 and $400, while a 480GB version will ship “later this year” with a $750 asking price. Random reads / writes are claimed to reach 83,000 / 44,000 IOPS, with 550MB/s and 520MB/s sequential speeds.

Not into that whole $500 fuse-melting monster graphics card thing? Then good news for you: AMD has finally brought out two more affordable models in its 28nm Radeon HD range. The 7770 is priced at $169 and claims to be the world’s first reference GPU that comes factory clocked to 1GHz. Meanwhile, the 7750 comes in at $109 and boasts a low enough wattage (75W, versus 100W for the 7770) that it doesn’t require its own power connector. Both cards pack 1GB RAM and run on AMD’s Cape Verde architecture, which makes them slightly different to the Tahiti-powered 7900 series, although they do inherit key top-end features like ZeroCore Power, PowerTune and Eyefinity 2.0.

Reviewers have mixed opinions, as befits a healthy blogosphere, but the low-power 7750 generally comes off slightly better, especially for those looking to build a budget or HTPC rig. AnandTech likes the power-to-performance ratio of both cards, but dislikes the price-to-performance of the 7770, noting that the older 6850 still offers more in this respect — at least for gamers. HotHardware concludes that AMD might have “technically” priced both cards “just right,” considering how they stack up against NVIDIA’s GeForce GTX560 and 550, but in practice those NVIDIA cards deliver a lot more punch for just a few extra dollars.

Read the reviews at HotHardware and Anandtech