Google gets a Caffeine injection

GOOGLE has launched a new search indexing system that provides consumers with "50% fresher" web search results.

The real-time infused indexing system, Caffeine, will help users find content that has recently appeared on the Web.

Caffeine will help link searchers with current news stories, blog posts, real-time updates or forum comments that are more relevant to their search queries than the three-week-old webpage the indexing system might have returned in the past.
"When you search Google, you're not searching the live Web. Instead you're searching Google's index of the Web which, like the list in the back of a book, helps you pinpoint exactly the information you need," said Carrie Grimes, Google software engineer.

"Expectations for search are higher than they used to be," adds Grimes. "Searchers want to find the latest relevant content and publishers expect to be found the instant they publish."

Google said it created Caffeine to fill users' growing need to be connected with up-to-date, real-time content.

To provide fresher results, Google changed the way they indexed the Web, analysing webpages in small portions on a continuous basis instead of refreshing large portions of their index every couple of weeks.

For consumers, this means that Google's search results will help them find new information on existing pages plus information on newly published pages, much faster.
For web publishers, it means that their newly added content will appear on Google faster than ever before.

Caffeine, however, will not change the way in which Google ranks its pages.
"We've built Caffeine with the future in mind," says Grimes. "Not only is it fresher, it's a robust foundation that makes it possible for us to build an even faster and comprehensive search engine that scales with the growth of information online, and delivers even more relevant search results to you. So stay tuned, and look for more improvements in the months to come."

Taken from TechCentral - The Star

The Promise of HD Graphics - Intel Core i5-661

2009 was supposed to be the year we got the first processor with an integrated graphics core on the chip package itself. Sort of like what AMD promised way back in 2008 for its Fusion initiative. Things however did not go according to the roadmap and we have been informed that it won't be till 2011 before Fusion becomes reality.

Unfortunately for AMD, its vision of a processing unit merging the CPU and GPU has now become reality for its biggest rival, Intel. Broadly speaking of course, since AMD's implementation of Fusion could very well differ greatly from what Intel is launching today. But the idea of integrating a graphics core with the processor, and not onto the motherboard as is traditionally done, is the same, at least from the consumer's point of view.

In fact, even Intel was promising this integration in 2009, with Havendale the codename for this family of processors, which would be based on Intel's current Nehalem CPU architecture. However, the company decided that its shift to 32nm production was moving along so well that it could afford to skip the 45nm Havendale and jump directly to its successor, codenamed Clarkdale.

That leads us up to today's launch of six new desktop models in the Clarkdale family. (Intel is releasing new mobile processors based on the same 32nm technology called Arrandale at the same time). All of them are dual-core processors using the LGA1156 socket and thus, even more so than the previous Lynnfield, CPU launch, are squarely targeted at the mainstream and business users. Like Lynnfield, these processors are equipped with Turbo Boost that will increase the processor clock speeds by quite a few bumps when dealing with applications that are unable to use the full complement of processor cores. The Core i5 series of these Clarkdale processors also comes with HyperThreading to better deal with today's more multi-threaded environment. Below are the new models announced and some of their more relevant specifications.



As you may have noticed, while these processors are quite similar to Lynnfield (same amount of L3 cache per core, memory support and Turbo Boost), the integrated graphics core differentiates the two. Intel may have managed to shrink the processor core to 32nm, but the graphics, memory and the PCIe controllers (remember, since Lynnfield, they are now integrated with the processor cores) are still on a 45nm process. Therefore, if you ever peel open the Clarkdale processor's heat spreader package, you'll find two distinct dies as shown below:



So what are the advantages of having the graphics core integrated with the processor? The main issue here is cost, as according to Intel, moving from the three-chip package solution of the previous Intel architecture (CPU with Northbridge and Southbridge) to the two-chip package (CPU and Platform Control Hub) means a 30% savings in terms of the package area, which translates into reduced manufacturing costs.

The other issue here, power savings, has more to do with the 32nm manufacturing process used for the processor component in the Clarkdale. For those who are counting, the graphics portion too benefits from going to 45nm since the previous X4500 series integrated graphics core on the G45 chipset for instance was built using a 65nm process. So it's likely that we'll see significant power savings with these new processors, especially when the maximum TDP rating for these new models is just 87W.

Finally, these new processors all come with a new x86 instruction set meant to improve the encryption/decryption speed for AES. Applications which benefit from the new instructions include file encryption utilities like Windows' BitLocker.

As the first processor to have its graphics core right beside it, Clarkdale can be a bit underwhelming when it comes to its key feature - graphics. Of course, expectations should be lowered instantly on seeing its 0'integrated' nature, more so when the graphics architecture is a third-gen evolution of Intel's familiar Graphics Media Accelerator.

From a pure performance point of view, the new Intel HD Graphics is a slight upgrade over the GMA X4500HD, even in its fastest (900MHz) incarnation on the Core i5-661. What it does significantly better is media playback, where it certainly manages to justify its HD tag. Our Blu-ray playback may not have been very impressive, but the beta nature of both the drivers and the CyberLink application means we won't be nitpicking in this area to pass a judgment this round.

Besides, CPU utilization is not such a big deal for users once there's no discernible lag like what we experienced with the i5-661. Unless you're watching a movie and rendering some animation or encoding videos in the background, in which case we recommend a dedicated graphics card with a quad-core processor. Instead, other features like dual-stream decode, dual HDMI output support and for true HD enthusiasts, bitstreaming of audio CODECs matter more. These are all found on the Intel HD Graphics and arguably put it on par with its rivals when it comes to media playback needs.



As a processor, the dual-core nature of the Core i5-661 meant that it was rarely going to challenge the higher-end quad-core processors, even with Turbo Boost and HyperThreading. Most of its better scores were naturally recorded for system suites like SYSmark 2007 and not heavily multi-threaded applications. In fact, from some of our benchmarks
, we can't help but wonder if the smaller L3 cache on the Core i5-661 affected its performance compared to the older Core 2 Duo E8500. Plus, the memory controller located off the main CPU die might add on to latency that could show up on more intensive tests, such as our memory testing in PCMark Generally, one would rank the Core i5-661 below the Core i5-750, which would appear to be the case based on Intel's own nomenclature.

This brings us to its cost. With a US$196 price tag for 1k units, the Core i5-661 will likely be competing against Intel's own Core i5-750. And like we said, it won't surpass the quad-core. Still, we hope the new H55 motherboards are competitively prices to give these new Clarkdale processors more value. Already, unlike the i5-750, the i5-661 has a decent mainstream graphics core inside, which should be adequate for some users, including businesses. Not forgetting the impressively low power consumption figures we recorded at both idle and peak loads.

Therefore, there's certainly a place in Intel's rank for these new Clarkdale processors such as in large business organizations. Media theater enthusiasts too should be among those eagerly awaiting these processors. For consumers who have a decent desktop or a HTPC setup already, there's much more potential in the mobile version, Arrandale, which we will hopefully cover in depth once we get some hands-on with an actual unit.

For more info go to Hardwarezone

This article is taken from HarwareZone
By Vincent Chang
Category : CPU
Published by Vijay Anand on Monday, 4th January, 2010
Rating : 4 out of 5 stars

AMD and Intel Settle Their Differences: AMD Gets To Go Fabless

It’s amazing how much the world changes before breakfast.

At 6am Pacific this morning, a joint Intel & AMD press release hit our inboxes like a nuclear bomb: AMD and Intel are settling their differences. And just like that, the CPU landscape as we know it has significantly changed.

We’re previously talked about the spat between AMD and Intel, one that Intel was looking increasingly likely to lose. AMD had had leveraged some very serious accusations against Intel over actions Intel supposedly undertook earlier this decade: paying OEMs and retailers not to use or carry products utilizing AMD processors through the use of discounts and rebates. With this AMD launched civil anti-trust suits, and petitioned various governments for criminal anti-trust investigations.

Recently, the European Union’s European Commission agreed with AMD’s claims, resulting in the EC fining Intel for roughly $1.45 billion. And other governments had investigations in various states, including the New York Attorney General’s which opened up a case against Intel earlier this month. Things were not going well for Intel.

With a major loss for Intel looming, the massive cost of many more years of litigation, and the monumental fine Intel would face if they lost a case in the United States (where damages are trebled), Intel and AMD sat down and began settlement talks in earnest in April, culminating in what has happened today. AMD and Intel have agreed to settle all of their outstanding litigation, Intel has agreed to adhere to certain business practices, Intel will be paying AMD, and the two companies have signed a new 5-year cross-licensing agreement for x86.

We should note that even with this settlement, neither AMD nor Intel are admitting to any wrongdoing. Intel is not admitting to engaging in any anti-competitive actions, and AMD is not admitting to breaking their existing x86 license by not producing the bulk of their CPUs in-house. In laymen’s terms, everyone is agreeing to be happy and to stop bludgeoning each other with lawsuits.

Out of this settlement come four major things for AMD:

1. $1.25 billion in cold, hard cash.
2. Intel will stop doing things that they and AMD agree they shouldn’t be doing.
3. The right to not have to produce x86 CPUs in-house. AMD can go fabless.
4. The right to have their x86 processors fabricated anywhere of AMD’s choosing.

Although AMD and Intel don’t attribute anything to the cash portion of the settlement beyond the fact that it’s a number the two could agree on, we believe that it is intended to cover the legal costs AMD has incurred over the years in fighting Intel. Remember why Intel settled in the first place: anti-trust suits are expensive. AMD will receive the $1.25 billion in cash within 30 days.

The second concession to AMD is that Intel and AMD have come to an agreement on which business practices Intel is and isn’t allowed to engage in, which Intel will have to adhere to. This is a tricky thing, since as we stated before Intel hasn’t admitted to any wrongdoing, so we’re not quite sure what practices are in question. Intel claims that this won’t result in any changes for them, but we will hopefully have a better idea once the agreement is published by the Securities and Exchange Commission. Without any additional information, our best guess is that whatever Intel was doing specifically to drive AMD out of the market, they have or are going to stop doing it.

Moving on, it’s the 3rd concession that’s the biggest for AMD. Although the Intel/AMD x86 cross-licensing agreement is private, it’s well-known that the agreement stipulated that AMD needed to produce the majority of their x86 processors in-house. With the formation of Global Foundries, AMD was applying a new interpretation to that requirement, one that Intel disagreed with and lead to a side-suit between the companies.

As part of this settlement, the in-house production requirement has been dropped from the cross-licensing agreement. AMD is no longer required to produce x86 CPUs, and as of today AMD has begun immediate preparation to go fully fabless.


The Foundry Dilemma: No Longer AMD's Problem

AMD has wanted out of the fab game for quite some time, but the cross-licensing agreement prevented that. Now AMD is free to sell off its stake of Global Foundries to Abu Dhabi’s Advanced Technology Investment Company (ATIC), who already owns the other half of GF. The result will be that ATIC will own a foundry like they’ve always wanted, and AMD will be rid of the task of operating a foundry like they’ve always wanted.

This in turn is going to have two immediate short-term benefits for AMD. The first of course is more cash; selling their share in ATIC is going to net the company a fair bit of cash and cash equivalents. The second of which is that AMD had to absorb some of GF’s losses; they would have turned a profit last quarter for example if owning part of GF didn’t cause them to take a $100mil loss. So AMD is going to have more cash on hand, and they’re going to further resolve their long-standing cash bleed. By no means does this make them rich (they still are up to their eyes in debt), but this opens up a great deal of liquidity to invest in new processor designs. After all, they’ve gone fabless, R&D is where the bulk of their investments will be.

Finally, we have the 4th concession: along with not having to produce CPUs in-house, AMD is free to have their x86 CPUs fabricated at any foundry of their choice. One of Intel’s points of contention have been that they don’t want to share x86 information with 3rd parties. Hence the requirement for in-house production in the first place. The settlement means that Intel will not object to AMD’s choice of foundry, opening them up to build CPUs at TSMC, UMC, and IBM, along with the continued use of Global Foundries. Given the cost and work involved in staying competitive with Intel’s fabrication advancements, we’re not sure who would be better prepared than GF, but the option is there on the table if AMD wants it.

Meanwhile, here’s what Intel gets:

1. AMD is dropping all private suits against Intel
2. AMD will retract all regulatory complaints with government regulators

As we mentioned earlier, Intel’s constant battle with AMD was expensive for both sides. Intel faced a particular risk, as the fines and judgments against them should they lose in the United States would have been three-times the damages found by the courts. Basically, if Intel was going to lose, they were going to lose big.

With this settlement, all of AMD’s private suits against Intel are immediately dropped. This means that Intel can immediately stop spending money on trying to fight those suits, and at the same time need not worry about what might happen if they were to lose. The key suit among these was an anti-trust suit in Delaware, which was set to begin in March.

Meanwhile things are slightly more complex when it comes to AMD’s regulatory complaints. Since any action taken as a result of these is done by various government agencies, AMD cannot simply have these investigations dropped. What they can do is drop any formal complaints, and go to each regulatory body with Intel, where both will explain that they’ve settled their differences. Ideally, this will be enough to convince these regulatory bodies to drop their investigations. However these bodies are not compelled to do so, and may continue their investigations if they wish. However without AMD’s participation, it would be much harder for these bodies to win at their cases.

The biggest regulatory case for Intel right now is the European Commission case, which is currently in appeal. It’s unlikely that the EC will drop the case after having already ruled against Intel, so the appeal is expected to go on as scheduled. Intel likely has lost this case for good, in spite of the settlement. Meanwhile Intel stands a far better chance of the New York Attorney General’s investigation dropped, along with investigations by US government agencies and foreign trade commissions.

This may not sound nearly as substantial for Intel as it does AMD, but it’s impossible to overstate just how much Intel was at risk of losing here if every case went AMD’s way. Intel could have lost many billions of dollars, if not more. Not losing that kind of money makes for a very good deal for Intel.

Ultimately, in spite of this being a “mutual” settlement, there’s no way to construct it as anything but a major win for AMD. Intel’s actions earlier this decade have presumably come at a great cost to AMD’s position, and this agreement codifies how that should never happen again. Meanwhile AMD is finally free to go ahead with their plans to go fabless, which will complete a significant shift for the company, and for Global Foundries. With this AMD’s immediate financial situation will greatly improve from the settlement cash, the sale of Global Foundries, and from GF no longer driving AMD in to the red. Or as Anand has summarized things, AMD will have the cash needed to go develop another CPU architecture, something essential to the company’s future survival.

With the reality of AMD going fabless however, we will end on a cautionary note. While this settlement greatly improves AMD’s short-term situation, it does not change the long-term situation for the company. AMD is still faced with the challenge of being the first company to make a sustainable business out of fabless x86 CPU design. This settlement cements AMD’s plans going forward, for better or for worse. In time, we hope to see that it’s the right plan.

Taken from AnandTech