Intel Unveils 'Sandy Bridge' Architecture
njkobie writes "Intel has officially unveiled Sandy Bridge, its latest platform architecture, at the first day of IDF in San Francisco. The platform is the successor to the Nehalem/Westmere architecture and integrates graphics directly onto the CPU die. It also upgrades the Turbo Mode already seen in Core i5 and i7 processors to achieve even greater speed improvements. Turbo Mode on Sandy Bridge processors can now draw more than the chip's nominal TDP where the system is cool enough to do so safely, enabling even greater boosts in core speeds than those seen in Westmere. No details of specific products have been made available, but Intel has confirmed that processors built on the new architecture will be referred to as 'second generation Core processors,' and are expected to go on sale in early 2011. In 2012 it is due to be shrunk to a 22nm process, under the name Ivy Bridge."
pray I don't first anymore.
From the article:
"While the results were impressive, it was noticeable that as the workload on the graphics increased, the discrete processor pulled ahead in the framerates substantially. We’ll look forward to testing the new processors on our own benchmarks to see just how good the new integrated graphics really are."
So I guess discrete chips will still be around for a while...
I haven't been following AMD/ATI, what have they been up to in this area?
Old news. My 386 has turbo mode. Wake me when they add math coprocessors to this beast.
"Core 2" chips were out years ago.
They're opening a new factory in Madison county.
Task Mangler
Intel needs to dump the DMI bus and go all QPI the last thing you want is Intel video lock in and only x16 pci-e lanes.
Because this is primary motivation as no one is even coming close to maxing out an i7.
Please let me push a button on the case to enable "turbo" mode.
For serious? Their leading feature is the ability to run even less efficiency than before? Extra speed at expense of power is great for desktops - too bad people hardly buy those anymore. Some of us actually want to run these things ON A BATTERY - can you imagine? Then again, being in Minnesota, I welcome the added heat.
...I just drove over the Sandy Bridge this evening. Coincidence? I don't think so!
The CB App. What's your 20?
Better/different selection of real gpu's? Or is this just all about a slow all in one Intel on one chip gpu option?
Domestic spying is now "Benign Information Gathering"
...will it use a new socket? I just shelled out for a fresh build because my mobo's processor socket was deprecated, I really hope they don't turn around and change it again so soon.
If you aren't angry, you aren't paying attention.
3th party chip sets also apple. They can't stay core2 for ever on the mini / some of there laptops and intel video does not fit in there gpu api.
and they don't like to put full pci-e x16 video chips in there low end systems.
and Intel 2nd Generation Core Processor to specialise in media processing
yes but unfortunately they are on average 4 years too late so they look like ass
Looking like donkey compared to another product on the market never stopped Wii games from selling. Or what fundamental difference between the low-end PC gaming market and the low-end console gaming market am I missing?
branching performance usually suffers [...] Video and image processing, game geometry, and 2d rendering really belong on a GPU-like architecture, not the CPU.
I thought game geometry involved a lot of branching, especially in the cases of potentially visible set construction methods (e.g. portal casting or BSP), collision detection, and path finding. Or have these problems been solved?
Please let me push a button on the case to enable "turbo" mode.
It's not a button on the case, but several laptops give the user a taskbar control to change the power-management strategy. So you have a "turbo" setting and a "battery life" setting.
Meh, maybe I'm just an embedded person who treasures ARM above all else and thinks that 640k ought to be enough for anyone.
Then go ahead and stick to your Game Boy Advance with its ARM7TDMI CPU and 384 KiB of RAM. If it's good enough for TOD, it might be good enough for you.
Not a single word on Intel killing overclocking, eh? According to anand's article majority of new CPU's won't allow ANY kind of overclocking.
i doubt intel is ready to cannabalize its existing 'i7 extreme' lineup just yet.
specially since no apps that the average douche uses come even close to fully utilizing a 980x.
expect a powerfriendly, i5 type chip at first.
wat i wonder, is if theyll finally use a decent chipset (x58 blows) with more lanes for pci express.
i would like to see motherboards capable of using 4 or more graphics cards in full x16 setups (not fractional crap like 16x/16x/8x or 8x/8x/8x/8x)
quad channel memory configs anyone?
also i would like the integrated memory controller to be able to handle a higher voltage load, allowing higher ram overclocks
btw, executing gfx api's using the vector instruction circuits on the cpu, wont be remotely as fast as a discrete chip made just for that (gfx card).
They wont have enough transistors set aside for vector circuits to entirely handle a modern commercial 3D game's requirements.
also the instructions theyve chosen for the avx set dont help the execution of 3d api's at all.
-HasHie
drunk monkey teenager translation:
Apple also has 3rd party chipsets. Monkey want bannanas. They can't continue using Core 2 forever on the mini, because their graphics librarys are already too much of a bottleneck. Monkey like kittens. Apple doesn't like to put full fledged PCI-e x16 interfaces in their low end systems, for some mysterious reason. Kittens, tasty.
Thought the '2' in Core2 referred to the second generation already...
With the Core i3/5/7 being the third these are more like the fourth generation.
Might be time for people who make C(G)PUs to have a rethink on naming schemes... maybe even take a leaf out of the software industry, e.g
Core i .
--- Users are like bacteria -> Each one causing a thousand tiny crises until the host finally gives up and dies.
From TFA:
Of course, we are left wondering what TDP means now, if exceeding it is standard.
Ironically, I was already wondering that. It never told what TDP is, and a Google define: search wasn’t terribly enlightening.
Alexander Peter Kristopeit bought his basement from his mommy for one dollar.
Not a single word on Intel killing overclocking, eh? According to anand's article majority of new CPU's won't allow ANY kind of overclocking.
And 128 nerds cried themselves to sleep... :)
XML is a known as a key material required to create SMD: Software of Mass Destruction
http://en.wikipedia.org/wiki/Thermal_design_power
Just google for TDP to begin with.
Man who leaps off cliff jumps to conclusion.
"You'd think Intel would just accept they suck at GPUs and buy Nvidia already."
Should Intel buy nVidia? Jen-Hsun Huang, who averages about $23.02 million per year, is not the sort of person who would easily integrate into Intel, and he is important to the leadership of nVidia. Intel's CEO, Paul Otellini, makes about $14 million.
Soon Intel's integrated graphics will have mid-range speed, leaving only the high range for nVidia. The high range of video adapters is mostly bought by teenagers who want to practice being violent with video games, instead of practicing being involved with other people. That means nVidia will be dependent on buyers who are being self-defeating; eventually there may be a backlash against that.
The high range of video performance will always be needed for architectural drawing and machine design, for example, but the total demand will drop, as the nVidia stock price seems to indicate. So, maybe nVidia is not a good purchase for any company.
Should Intel CEO Paul Otellini be replaced? Another reason Intel should not buy nVidia is that Intel is generally a failure at anything besides making new CPUs and support chips. For the success of Intel and AMD in making CPUs, the world can be extremely thankful; that's enough success for any company.
But Intel in other areas seems amazingly badly managed. Intel marketing seems completely out of control. Is the product confusion at Intel a deliberate, sneaky way to sell slow processors to technically challenged customers, or just stupid?
Quote from the article linked just above: "Sandy Bridge PC processors will keep the CORE-i3, i5, and i7 designations and will be rebranded the "new CORE-i3..." That approach is likely to create confusion among customers about exactly what they're buying, given that the average user likely wouldn't be able to pick a Nehalem i7 from a Westmere i7 or Sandy Bridge i7."
Either Intel's purchase of the inferior security software maker McAfee for a "lofty 60% premium" is a HUGE mistake, or the reasons why it is not a mistake should be explained by Intel marketing. No explanation was given, apparently. McAfee has a 21.9% market share selling software often pre-loaded on a computer to technically challenged buyers.
Quote from the article: " 'We believe security will be most effective when enabled in hardware,' Intel Chief Executive Paul Otellini said in a conference call." That seems a particularly wacky statement. "Security software" is needed only because, in my opinion, Microsoft deliberately allows its software to be insecure. Insecure software makes Microsoft more money because people with infected computers often buy another computer. For example, see the New York Times article, Corrupted PC's Find New Home in the Dumpster. The Apple Mac OS, Linux, and BSD operating systems do not require "security software" because they are made to be secure.
Intel CEO Otellini does not seem to have the social sophistication necessary to running a big company. When he made an announcement in 2006 about the Intel Eduwise laptop, he seemed to be intending to have Intel compete with MIT professor Nicholas Negroponte's One Laptop Per Child (OLPC) charity program. However, Intel's intention seems to be just to make a market fo
i had a AMD 486 DX5 at 133MHz on a 386 case, after some upgrades...
i connected the turbo button to the Bus speed jumpers, so when i pressed, the bus jumped from 33Mhz to 40Mhz, overclocking the cpu to 160Mhz... i run at "full" speed when i was at home and put the normal speed when i left it idle
To my surprise, it worked really well, the PCI bus accepted that speed, the network and SCSI card never gave any error until i disconnect the computer about 6 years ago
i also tried to up the bus to 50Mhz and the CPU, RAM, the vesa local bus (for the graphic card) and the ISA bus (sound card) worked fine, but it was too much for the PCI bus and the network and scsi cards didnt work so gave up from having a 200MHz 486 CPU and fall back to the already "good" 160Mhz... relative power of the setup was about a Pentium 90-100Mhz... running at normal 133Mhz, the performance was a little lower than a Pentium 75Mhz
Higuita
sandy bridge is only going to be like 20-30% faster then what's available right now, the last two generations after core 2 duo have only had modest games in the 20%-30 range. I've not been impressed at all lately, it seems technological leaps have slowed right down for cpu's at least.
3th party chip sets
Less high power high speed chips means less power spent on interfacing between them and less complex system cooling. For most systems that is probablly worth sacrificing the ability to choose northbridges.
also apple. They can't stay core2 for ever on the mini / some of there laptops and intel video does not fit in there gpu api.
If true that sucks for them but ultimately I don't think in.
It appears (see http://slashdot.org/comments.pl?sid=1786182&cid=33570924 for caveats) that with sandy bridge intel is now beating the best nvidia chipsets for the C2D so it won't be a downgrade on general graphics performance for laptop builders to move from C2D+nvidia chipset to sandy bridge.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
s/only supported 16 modules/only supported 32 modules/
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
Isn't the whole point that the chips essentially dynamically overclock themselves based on demand and current heat profile?