CPUs
Latest
Cedar Trail may be delayed, new Atoms gone 'til November
If you've been waiting to catch a whiff of some Cedar Trail freshness, looks like you're just gonna hold your breath a little bit longer. DigiTimes is reporting that the next-gen Atom chip has been pushed back from its anticipated September launch to November. Apparently Chipzilla is having issues with the graphics drivers and has been unable to pass Windows 7 certification. The new low-power CPUs should still be ready in time for the holiday season though, and will likely find their way into plenty of netbooks that almost nobody will buy.
Will Intel's Core i7 Sandy Bridge E CPUs ship without fans or heatsinks?
Over the past few months, we've peeked Intel's roadmap more than once, but all told we've been treated to scant few details about its high-end desktop line, Sandy Bridge E (that's "E" for enthusiasts). Today, though, VR-Zone is reporting that the next generation of these CPUs will ship without fans or heatsinks -- a tacit acknowledgment, perhaps, that Intel's home-brewed cooling system will be inadequate in the eyes of hobbyists anyway. We reached out to Intel for comment, and while the company stayed mum on the topic of cooling, it did go out of its way to clarify another point the folks at VR-Zone made in their report. The outlet had said that the forthcoming 3820, 3930K and 3960X CPUs will be rated at 130 watts, but will consume closer to 180W and draw up to 23 amps from the 12V2 supply rail -- all without overclocking, mind you. An Intel rep writes: "TDP expectations for the 2nd Generation Intel Core i7 processor family for socket LGA-2011 are in line with previous generations of high end desktop products." In other words, built-in cooling system or no, the TDP should be in line with what we've seen from other Extreme-branded processors. As for the cooling, it's unclear when, exactly, we'll get the full spill -- the CPUs are rumored to launch before the end of the year, with the quad-core 3820 arriving after the six-core 3930K and 3960X.
Researchers boost multi-core CPU performance with better prefetching
Piling on cores is one way to boost performance, but it's not necessarily the most efficient way -- researchers at North Carolina State University have developed a new prefetching technique for processors that could boost performance by up to 40-percent. As you may know, any data not stored in a CPU's cache must be pulled from RAM, but as more cores are added they can create a bottleneck by competing for memory access. To counter this designers use prefetching to predict what information will be needed and grab it ahead of time, but guessing wrong can hurt performance. Researchers tackled this problem from two fronts: first, by creating a better algorithm for divvying up bandwidth, and second, by selectively turning off prefetching when it might slow the CPU. Full PR and an abstract of the study being published June 9th are after the break.
Intel goes ULV for laptops to combat the oncoming tablet horde
Intel has been talking up its x86-powered smartphones and battery-sipping Atoms for tablets quite a bit recently, but the company hasn't forgotten its roots in traditional PC form-factors. At an investor event in San Francisco, CEO Paul Otellini announced a significant change to its line of notebook CPUs -- ultra low voltage will be the new norm, not just a niche chip for high-end ultra-portables. The baseline TDP for future CPUs will be in the 10 to 15 watt range, a huge drop from the 35w design of the mainstream Core line and lower than even current-gen ULV chips (which bottom out at 17w). The company also plans to make NVIDIA eat its words by putting the pedal to the metal on die shrinks -- releasing a 22nm Atom next year followed by a 14nm version in 2013. That could mean our fantasy of true all-day battery life in a sleek and sexy laptop will finally come true. Don't crush our dreams Intel!
IBM says graphene won't fully replace silicon in CPUs
As you may have been able to tell from the flurry of research that's occurred over the past few years (which has even resulted in a Nobel Prize), there's plenty of folks betting on graphene as the next big thing for computing. One of the big players in that respect has been IBM, which first opened up the so-called graphene bandgap and has created some of the fastest graphene transistors around, but is now sounding a slightly more cautious tone when it comes to the would-be demise of silicon-based CPUs. Speaking with Custom PC, IBM researcher Yu-Ming Lin said that "graphene as it is will not replace the role of silicon in the digital computing regime," and further explained that "there is an important distinction between the graphene transistors that we demonstrated, and the transistors used in a CPU." To that end, he notes that unlike silicon, "graphene does not have an energy gap," and that it therefore cannot be completely "switched off," which puts it at quite a disadvantage compared to silicon. Intel's director of components research, Mike Mayberry, also chimed in on the matter, and noted that "the industry has so much experience with it that there are no plans to move away from silicon as the substrate for chips." That doesn't mean that there still isn't a bright future for graphene, though -- Lin gives the example of hybrid circuit, for instance, which could use graphene as a complement to silicon in order to "enrich the functionality of computer chips."
AMD sees a tablet chip in its future, and an end to the core-count wars
AMD told us that it wasn't terribly interested in the iPad market, and would wait and see if touchscreen slates took off, but CEO Dirk Meyer changed the company's tone on tablets slightly after reporting a $118 million net loss (on $1.62 billion in revenue) in a Q3 2010 earnings call this afternoon. First revealing his belief that tablets will indeed cannibalize the notebook and netbook markets, he later told investors that he actually expects AMD's netbook parts to start appearing in OEM slates in the next couple of years, and that AMD itself would "show up with a differentiated offering with great graphics and video technology" when the market becomes large enough to justify an R&D investment. Elsewhere, AMD CTO of servers Donald Newell prognosticated that the number of individual CPUs on a chip won't go up forever: "There will come an end to the core-count wars," he told IDG News. Just as the megahertz race was eventually defeated by thermal restrictions, so too will the number of cores on a chip cease to increase. " I won't put an exact date on it, but I don't myself expect to see 128 cores on a full-sized server die by the end of this decade," he said. So much for our Crysis-squashing terascale superchip dreams, we suppose.
Leaked Intel roadmap details Sandy Bridge CPUs, expands SSD lineup?
If a set of supposedly leaked slides are legitimate (and they sure look convincing to us) then the shape of Intel's dirt-brown 2011 plans has finally been revealed. As you can see immediately above, Chipzilla's 25nm flash process is just about ready to double the size of the company's award-winning consumer SSDs, bring up to 400GB of "enterprise-grade" multi-level cell memory to the enterprise space, and create a series of netbook-sized mini-SATA drives with the remains of the 34nm silicon. On the processor front things are a little more iffy, but it seems safe to say that the naming scheme has changed, as the silicon wafers you'll slot into a Q67 Express motherboard will have an extra digit (and often a letter) affixed to the end. Instead of a Core i7-870, you'll see the likes of Core i7-2600, i7-2600S, i7-2600K and i5-2500T, with the K (as in the past) affording you an unlocked multiplier for overclocking and the S equaling reduced power consumption, or vastly reduced for the T models. German publication ComputerBase -- which found and subsequently pulled the slides -- somehow managed to dig up nearly full specs for desktop and laptop CPUs as well, and though we can't verify their legitimacy, you're welcome to marvel at the idea of a 3.5GHz processor running on just 35 watts by visiting the source link below. %Gallery-99426% [Thanks, Aristo]
New Core i7, Core i5 mobile processors to debut in HP Envy?
Intel's churning out the mobile chips like nobody's business -- no sooner does it admit the existence of Core i5 and Core i3 ULV CPUs, the chipmaker finds itself with more explaining to do. That's because dedicated Hewlett-Packard fans just found references to three new standard-voltage Core 2010 chips in the service manuals for the new HP Envy 17 and Envy 15 gaming laptops. The i7-840QM appears to be Intel's new top-of-the-line quad-core CPU, sporting a 1.86GHz clock that turbos up to 3.20GHz (or 3.20MHz, if you believe the above screencap) and 8MB of L3 cache; the i7-740QM is two steps down with 1.73GHz / 2.93 GHz clocks and a 6MB L3 cache; and the i5-450M appears to be indistinguishable from the existing i5-520M with 2.4GHz / 2.93 GHz clocks and 3MB of L3 to help it along. Given the megahertz typo above and the fact that none of these new processors appear anywhere else in the documentation, we wouldn't be surprised if these specs weren't rock-solid... but if they are, the new Envy may well live up to its name. Full PDF available below. [Thanks, Reznov]
NVIDIA VP says 'Moore's law is dead'
NVIDIA and Intel haven't been shy about their differing respective visions of the future of computing in the past year or so, but it looks like Team GPU just upped the rhetoric a little -- a Forbes column by NVIDIA VP Bill Dally argues that "Moore's law is dead." Given that Moore's law is arguably the foundation of Intel's entire business, such a statement is a huge shot across the bow; though other companies like AMD are guided by the doctrine, Intel's relentless pursuit of Gordon Moore's vision has become a focal point and rallying cry for the world's largest chipmaker. So what's Dally's solution to the death of Moore's law? For everyone to buy into parallel computing, where -- surprise, surprise -- NVIDIA's GPUs thrive. Dally says that dual, quad- and hex-core solutions are inefficient -- he likens multi-core chips to "trying to build an airplane by putting wings on a train," and says that only ground-up parallel solutions designed for energy efficiency will bring back the golden age of doubling performance every two years. That sounds fantastic, but as far as power consumption is concerned, well, perhaps NVIDIA had best lead by example.
Samsung's ARM roadmap lays coordinates through 2013: Aquila, Venus, and Draco (oh my)
Ah, leaked company presentation slides, they have a clarity only Mr. Blurrycam would despise. EETimes got a batch of them from Samsung dated November 2009 making the rounds, but more important than revealing its equal love for both Roman and Greek mythology, we get a glimpse at its then-planned ARM chip roadmap (yeah, another one) through 2013. In a nutshell, for the Cortex A9 crowd we've got the 800MHz dual core "Orion" due for mass production in Q1 2011, a 1GHz single core "Pegasus" for Q4 2011, a 1GHz dual core "Hercules" for Q1 2012, and for sometime in 2012 / 2013, a 1.2GHz dual core "Draco" and quad core "Aquila." Fear not, Cortex A5 fanatics, you've got gifts as well, in the form of 600MHz single core "Mercury" and dual core "Venus" chips, slated for 2010 / 2011 and 2012 / 2013, respectively. We don't expect the nomenclature to extend beyond internal usage, but frankly, who cares -- it's the devices that count, and unfortunately all we can do is doodle our future gadget hopes and dreams onto scraps of paper while we wait.
Intel's 2011 CPUs require new motherboards, start saving those pennies now?
Hoping Intel's 32nm Sandy Bridge CPUs will unify your computing world next year? Bit-tech reports they may also drain your pocketbook. Anonymous sources told the site that Intel's spiffy new CPU / GPU / memory controller integrations will use two new sockets, LGA-1155 and LGA-2011; if true, you'll need to buy a new motherboard to match. Aside from packaging the aforementioned GPU and memory controller on the same die, LGA-1155 rumors don't hold any surprises thus far, but the enthusiast-grade LGA-2011 chips will supposedly sport a quad-channel DDR3 memory controller (like Nehalem EX) and the long-awaited PCI Express 3.0 for 32 lanes of graphics-gobbling bandwidth. We also hear new southbridges will have native SATA 6G. Honestly, these rumors are so tame as to be entirely credible. Let's dream up something ridiculous, like Intel actually supporting USB 3.0.
Intel putting some ultra-thin laptop CPUs on hold to gauge demand?
Rumors of Intel delaying processors to avoid stepping on some of its other processors' toes? Stop us if you've heard this one before. This latest instance concerns some unspecified ultra-thin laptop CPUs that Intel has reportedly now put on hold until the third quarter of the year and, according to DigiTimes, a launch then will still depend on what demand looks like at that point. All of this is supposedly at least due in part to Acer's decision to shelve some of its ultra-thin laptops that would have used the processors, and instead focus on thin-and-light Calpella-based laptops. No word from Intel, of course, but past history suggests we could see some further back and forth before things finally shake out.
Intel profits recover to $2.3 billion in Q4 2009, company describes it as 875 percent jump
Yo Intel, when your 2008 fourth quarter was one of the worst you ever recorded, it's slightly, just slightly, facetious to go trumpeting an 875 percent improvement in your 2009 fortunes. The self-appointed chipmaking rock star has clocked up $10.6 billion in revenues for the last quarter, which filters down to $2.3 billion in pure, unadulterated, mother-loving profit. That's good and indeed technically nearly nine times what the company achieved in the same period the previous year -- we'd just appreciate this to be represented as the recovery it is, rather than some major leap forward in the face of a global financial meltdown. Either way, the Santa Clara checkbook is now well and truly balanced, even if it would've looked fatter still but for the small matter of a $1.25 billion settlement reflected in last quarter's results.
Intel forks over the $1.25 billion settlement to AMD, apparently had it 'just laying around'
Remember that time you owed your buddy for the take out and then you found $20 in the couch? This is kind of like that. Except instead of "take out" we're talking "accusations of anticompetitive practices and stolen IP," and instead of "$20" we're referring to the $1.25 billion check that Intel just shot into the arm of AMD, as per agreement. Intel certainly isn't out of the woods yet with this anticompetitive stuff, but with the biggest CPU monkey off its back and some fancy patent cross-licensing between the companies, we should hopefully see the benefits of this in better and faster chips from both chip giants in the somewhat distant future.
Intel's Silverthorn becomes the Atom, Menlow the Centrino Atom
Another brain-shattering announcement that you've all been waiting for is here. Intel has officially dubbed the Silverthorn and Diamondville chips "Atom," and the Menlow platform has become the "Centrino Atom." We know, it's hard to believe the day has come, but it's finally happened. Obviously, the 45nm Atom chips, and Centrino Atom technology will be targeted for MIDs, UMPCs, and all manner of small, internet-centric devices. Undoubtedly an unending vista of total awesomeness awaits us now that they've got some fancy new names.[Thanks to everyone who sent this in]
