The next big thing in processors is hitting today — Intel’s Core i7, otherwise known as Nehalem, is finally hitting the ground. If you don’t know that it is, check out our past coverage. This first batch probably doesn’t have the processor you want to get (the more consumer-oriented ones be coming out a little later, good price comparison at the bottom of this page), but it’s a good preview of the performance we’ll all soon have in our desktops. Check out the coverage at: HardOCP Tom’s Hardware Anandtech Tech Radar Guru3D Or your favorite hardware review site. I can’t wait to get one of these for my PC, and I’m hoping Apple is looking at making them available as well. → Read More
The fine people over at APC Mag got their hands on Intel’s Core i7 Nehalem processor (codenamed Bloomfield) and have bestowed upon us the following information: Intel has moved the memory controller onto the chip itself, which has made the chip larger, which means we’ll be getting a new socket size (socket LGA1366). So you’ll need a new motherboard, basically. The front side bus (FSB) has been completely replaced by what Intel is calling QuickPath Interconnect (QPI). QPI can outhustle FSB handily and facilitates direct communication between the CPU and RAM (read more here). Bloomfield adds level three cache. It’s 8MB shared amongst four cores. The new motherboard chipset is codenamed “Tylersburg” and will integrate Intel’s X58 chipset. Mobo’s available at launch will be from GIGABYTE, Asus, MSI, and EVGA. Intel is bringing back Hyperthreading – these new chips will be able to handle two threads per core. X58 motherboards from Intel will initially include four RAM slots. Boards from other manufacturers can and will include six. That’s a lot of RAM. We should start seeing these new CPUs and motherboards this month. → Read More
The always excellent IEEE Spectrum has a great article right now that is both a primer for current chip manufacturing techniques and a look at what the challenges are that chipmakers are facing. As we’ve talked about before, Moore’s law has continued to apply only through the utmost efforts and desperate “cheats” by the engineers who design the chips and the process to create them. But we’re running into some fundamental barriers, many of which have been circumvented before only to have them crop up again, more formidable this time. What is to be done? Double-patterning lithography is the current stopgap solution. Their very elegant explanation of the idea is of taking two picket fences with the maximum possible post density (the posts representing transistors) and aligning them, one behind the other, so that there are no gaps between the posts. What they’re planning on doing is printing two different patterns at the limit of density right now that are complementary with each other, then essentially putting them on top of one another. They’re really MacGyvering things here but hey, whatever works, right? → Read More
This is pretty cool. I’m all for tweaking your setup a little bit, but the guys at this Gigabyte-sponsored Championship in Taipei take it more seriously than I ever could. Maybe that’s because there was a $5000 prize at stake. Hmm… The pictures are great if you’re a system builder: all cables and pipes and thermoses (thermi?) filled with liquid nitrogen. They had 11 full canisters for the participants to work their way through; now that’s a sign that this is a serious competition. There’s a lot to the story and if you’re an overclocker yourself, it’s definitely worth a look. For the record, a Finnish team called Sweden-2 took top honors but the Americans took home the “freestyle” prize. We tied with the Russkies for 2nd place overall, but they had a higher 3DMark score so they took the silver. Next time, Boris. Next time. → Read More
If that headline doesn’t get excited, may as well move on to the next story. Nothing to see here, folks! Except for a bad-ass processor that’s about to drop on November 17. If you’ve been following the Nehalem story, you know that it’s Intel’s microarchitecture “tock” to the 45nm process Penryn “tick.” You may have heard it referred to as “Core i7″ as well. Basically, it means you’re going to be getting the most out of the silicon starting next month, or more likely early next year, since the ones coming out next month are super high-end, not really for end users like you and me. Still, it’s time to start checking under those cushions for spare nickels, because upgrade time cometh! [via Expreview and ZDNet] → Read More
Oh my. Facing mounting debt and increasing costs, AMD is splitting into two parts: one to design chips and one to make them. The move is made possible by a $6bn infusion by two Abu Dhabi investment firms, an amount which will go a long way toward relieving debt and covering the cost of establishing the manufacturing arm (“Foundry” for now) as a workable partner. They’ll be splitting the debt and building a new chip fab in upstate New York (with some help from the state — where’s that money coming from?) If this desperate move doesn’t succeed, AMD will surely fall, and then Intel will reign supreme. And those would be dark days indeed. → Read More
These little netbooks that we hear about day in and day out are about to get a little zippier, thanks to Intel’s dual-core Atom chip – the 330 – shipping out to vendors. The September launch seems to coincide with what we’d heard earlier, so that’s good. The 45-nanometer Atom 330 chip will feature the same 1.6GHz bus speed as the single core model, will have 1MB of level 2 cache, and will support DDR2 667 memory. → Read More
Yeah, I know, it sounds a bit dry; how could a chart of microprocessor stats be awesome? Well, just look at it! It’s glorious! And huge! It’s so very complete and monolithic that I just have to stand in awe of it. There’s just so much… information! It’s organized by date, by core, by what have you, and having it visualized is a revelation. I like the color scheme, too. There’s lots of them and they are interesting to look at; click below to see ‘em, or go to where they come from. → Read More
Well! IBM and AMD are certainly proud of themselves today! They’ve been working on a 22nm SRAM cell, two generations ahead of the current 45nm process. The devices developed and manufactured by AMD, Freescale, IBM STMicroelectronics, Toshiba and the College of Nanoscale Science and Engineering (CNSE) were built in a traditional six-transistor design on a 300 mm wafer and had a memory cell size of just 0.1 μm2, which compares to Intel’s 45 nm SRAM cell size (the test chip that was used for today’s 45 nm processors) of 0.346 μm2. Bad news for Intel, right? Well, kind of. But they’re unfazed. An Intel spokesperson downplayed the advance, in my opinion successfully: “A single SRAM cell has 6 transistors in it. Intel’s 32 nm SRAM array, which we announced back in September, has 290 million cells or bits, and a total of 1.9 billion transistors.” Oh, daisies! Basically Intel is saying “yeah they’ve got a single weak-ass 22nm cell, who cares? We could do that any time we want, but in the meantime we’re miles ahead of you in the sector that matters.” → Read More
Intel’s “tick-tock” process of alternating between shrinking the die and updating the microarchitecture has done them well so far, and apparently they see no obstacles to continuing it through 2012. Unfortunately, that’s when things go a bit pear-shaped and you start having to deal with quantum properties (10nm is edge-of-comprehension small) and doing drastic things like stacking chips and carving canals into them. The commentary is all in French, but you can smell what Intel’s got cooking if you go through the slides here at Canard Plus. I assume you either want all the details or don’t care at all, so I’ll let you read them yourself. Image credit: Canard Plus → Read More
The next round of Intel processors have officially been named Intel Core. The very next round of processors, previously codenamed Nehalem, will go by the name Intel Core i7. As always, model numbers will be used to differentiate the different chips. Yes, there will still be an Extreme Edition, which is the one on the right up there. Useful, perhaps, for those of you trying to play Crysis at decent speeds. These chips should go into production in the fourth quarter of this year, with up to four cores of processing goodness. via Boing Boing Gadgets → Read More
What a ridiculous project! But how awesome would it be to be the hardcore system building nerds they asked to do this? A couple months ago, a French Tom’s Hardware-related superteam got together to overclock an Intel Core 2 Quad Q6600 2.4GHz as far as it would go. They just put up the pictures and everything yesterday. They used liquid nitrogen cooling and a pretty serious-looking compressor to suck the heat right out of the thing, and ended up more than doubling the cycles. For reference, it’s generally safe to overclock your stock hardware about five percent, and even the real pros get maybe an extra thirty percent — and at that point you’re risking a lot of errors, artifacts, and so on. If you’ve ever wondered what liquid nitrogen cooling looks like in motion, check out the video (en Francais). So congratulations to the team, but watch out for this guy (two years ago). → Read More
We’re all hearing a lot about Intel’s Atom range of processors these days, likely because of the tiny laptop revolution (or whatever you want to call it). While VIA’s Nano is not competing directly with Atom as, say, GeForce does with Radeon, they are in the same ballpark and when it comes to low-power, general-purpose setups for things like tiny laptops or tablets, people are going to talk about Nano and Atom as competitors. HardOCP took a look at the two to see how they’d do in various situations. With a bare bones setup (just a compatible mini motherboard and a gig of RAM) they compared the two in both synthetic and real-world applications. Intel’s been ruling the CPU roost these days so it came as a surprise to me when their touted Atom platform got its ass handed to it by pretty much every measure. Pure number crunching, desktop environment, video encoding, it got whupped pretty much all over town. These were essentially desktop setups and it’s difficult to extend the testing to the market, where demands on the systems are so varied. But it’s fun to know that Intel is getting schooled somewhere. → Read More
If you’re reading this website (you flatter us) you’re probably one of the more tech-savvy people in the family. So doubtless you get tapped occasionally to fix Aunt Estella’s printer or de-spyware your parents’ machine. And every once in a while, someone will want to buy a new computer. While there’s no better time than the present, they will certainly be mystified by the array of possible configurations, and none more confusing than the processors. How many cores? Front side what? Celer-who? This article is a good primer for what the story is with multi-core CPUs. There’s really nothing new in it, and it won’t tell you which is the best of course, but in terms of explaining just why the selection is the way it is (if you don’t already know), it’s a pretty good read — suitable to forward to Aunt Estella. → Read More
Not content to let Intel steal all the glory with its new low-power Atom chip, AMD appears to be developing a similar chip of its own. It’s currently only known as “BGA” but will hopefully get a much cooler-sounding name when it’s ready to go. AMD’s current low-power Geode processor has served as the company’s netbook CPU of choice, but it tops out at 500MHz and has been around for more than five years now. It’s been speculated that this new BGA chip could actually be AMD’s “Bobcat” chip that the company announced last year. As you can see from the above slide, this new CPU will clock in at 1GHz and will use eight watts of power, making it ideal for netbooks but a little too greedy for smaller mobile devices, as CNET’s Tom Krazit points out. No word on when we’ll start to see devices using this new chipset, but it’d better be soon for AMD’s sake since Intel, NVIDIA, and VIA seem to have a good head start. via Electronista → Read More
Chips these days are pretty much flat. And I’m not talking about tortilla chips. They call those silicon things wafers for a reason: all those little gates and channels are lying flat in a single layer. Chipmakers know that you can stack chips on top of each other and multiply your computing power due to the decreased distance the signal has to travel (I’m kind of at the edge of my knowledge here, someone correct me if I’m wrong), as well as the fact that you can fit more computing power into a given area. The problem is, all those electrons flying around end up generating heat, and if your chip is layered, there’s less space for the heat to escape to. Without a proper cooling solution, the stacked chips would melt from the inside out almost instantly. Well, IBM’s Zurich Research Laboratory is looking into ways to have their cake and cool it, too. They’ve cut tiny canals into the chips that, once the chips are stacked, water can flow through without risk of leaking into the paths of the gates and channels. Cool water goes in one side, sucks up the heat, and comes back out to be cooled by whatever they’ve got rigged up, a desk fan or something. I don’t really see this getting into your laptop any time soon, but it’s cool to know. → Read More
Those lucky bastards over at AnandTech got their mitts on a couple totally unofficial Nehalem chips. For those of you not in the know (a group which included me until about 15 minutes ago), Nehalem is the second part of the Intel processor cycle that started with the 45nm Penryn shrinkage all those months ago. Just as the Core microarchitecture (and the phenomenally popular Core2 Duos) were the second step of the 65nm shrinkage. The new architecture (and pinout) will show what 45nm can really do. In the tests they did, the Nehalems annihilated their Penryn Core2 Quad Q9450. Even at artificially limited clock speeds and lacking the optimizations surely forthcoming, it trashed Intel’s current flagship like it was nothing. Media encoding, memory access speeds, 2D and 3D graphical tasks, everything was improved by huge margins. I’m pumped for this. I may actually have to overhaul my system. → Read More
AMD has officially taken the plastic-wrap off of its new high-performance notebook platform, formerly known as “Puma”. The platform consists of the new AMD Turion X2 Ultra Dual-Core Mobile Processor and graphics powered by ATI Radeon HD 3000 Series chips, along with 802.11n and 3G wireless connections. We’ll see notebooks outfitted with the new platform from Acer, Asus, Fujitsu, HP, MSI, and Toshiba. The new Turion processors that’ll be found in these notebooks will feature “AMD Independent Dynamic Core Technology, a new mobile-optimized memory controller, and power-optimized HyperTransport 3.0,” which should help out with stuff like high-definition video content and better battery life. → Read More
NVIDIA’s on a roll with their graphics hardware and motherboards, why shouldn’t they find success in the mobile and PMP market? They have the technology. Graphics-intensive applications like 720p video recording and playback are a breeze when you design the whole package from tip to tail with your own technology instead of piecemealing it from various parts. It looks like it’s an actual, three-dimensional product too(it’s at 3:00), not just a scribble on a webpage. With other full-touchscreen phones coming out from like every provider (Instinct, Diamond, Vu), it’s got a lot of competition, but this is probably more a “look, we can do this” phone and not intended to dethrone the iPhone. People may notice that NVIDIA’s Tegra Platform is more efficient for this kind of architecture than Intel (if that is the case) in which case NVIDIA may make more of a splash by powering others’ phones instead of making its own. In either case, I can’t wait to try out one of these little beasts. → Read More
San Francisco, CA
