It’s the Consumer Electronics Show in Las Vegas, so it’s time for new stuff to spend your hard-earned cash on. This time around, Intel is making headlines by officially launching its Sandy Bridge line of processors, which, up until now, were totally next-generation.
Sandy Bridge is just the codename, of course, and now they’re known simply as the 2010 Intel^A(R) Core^a"c Processor Family. The biggest innovations here are the 32nm manufacturing process on the company’s second-generation high-k metal gate transistors (which sounds like a line from a Mass Effect weapon description, but allright, I’ll roll with it), and Intel HD graphics integrated onto the 32nm die.
“The all new 2010 Intel Core processor family is the first to integrate graphics into mainstream PC processors. With Intel HD Graphics, the processors deliver stunning visuals and smooth high-definition video playback,” the press release reads, “It’s also the industry’s first integrated solution to deliver multi-channel Dolby TrueHD and DTS Premium Suite home theater audio.”
The new processors (the Core i7 and i5 variants, at least) also come with Intel Turbo Boost Technology, which allows cores to temporarily run at a higher operating frequency than the base one, but only “if it’s operating below power, current, and temperature specification limits”.
Expect to see these things in your local laptop and desktop population soon enough.
I guess he means desktop? I thought laptops are the mainstream these days, and desktops are for enthusiasts…
They mean x86/x86-64 processors.
Ah, so he means integrating on the same die. pinetrail already had gpu in the same cpu package iiuc.
They really do mean x86. ARM, at least, has had CPUs with integrated GPUs for a while now.
Edit: actually atom has had integrated GPU for a while now too. So yes, by mainstream they mean desktop I guess.
Edited 2011-01-04 08:51 UTC
Intel’s GMA HD, which replaced the GMA 4500M HD, was integrated in the CPU. The Intel HD Graphics 3000 goes a step further and is now directly on the die. This is the “tock” phase of Intel’s “tick-tock” release schedule. In other words, it’s an entirely new microarchitecture targeted at PCs. Speaking of which, I’m always stunned to find that people rely on Wikipedia as a primary source for information. I’m even more stunned when a Wikipedia article has more accurate information than an average person on a pretty common topic. For example: “A personal computer (PC) is any general-purpose computer whose size, capabilities, and original sales price make it useful for individuals, and which is intended to be operated directly by an end-user with no intervening computer operator.” Desktops and laptops are just PCs, and yes, Apple sells a lot of PCs despite it’s “Mac vs. PC” ads.
It’s a catch up to AMD play.
I think that even AMD will have something to suffer from it. AMD Bulldozer will have strong integer performance for threaded apps (almost unseen for now, excluding databases and virtualizations in my experience). Anyway, about integrated video card, is fairly enough good to not think on: “integrated intel graphics”, even is not the “Crysis” CPU/GPU, is for certain the Win7 CPU. If you play any game that IE9/FF4/Flash will give, go video decoding, and so on, I think Intel still will be a winner. The last scarry thing, is that AMD with their quad double core module (aka 8 cores) will likely have lower frequencies per core, so they will be as Phenom was some years ago: a lot of cores, but bad performance in small threaded workloads. So for me Intel did at least halfly his job by byting lower end graphics, right now AMD hopefully will find the next marketing term of their CPUs to prevail.
AMD has indicated that operating frequencies will be equal to & greater than currently available.
Top shelf should be nearing the 4GHz mark on Turbo at release. Speeds should step from there.
The x6 line of phenom II shows that AMD knows how to make it happen. The longer pipelines on the Bulldozer indicate a design more suited for higher clocks. I just hope we don’t see a repeat of Intel’s Pentium 4 mistake.. except made by AMD this time…
Doubtful, performance is claimed to be at least 12% faster per clock ( to match i7 clock-per-clock ). No figures for the clock capabilities, which is something that can really only be surmised until qualification production begins.
I think AMD’s Bulldozer architecture is a brilliant move, but I fear the FPU performance may be below intel by too great a margin…
Oh well.. I’m rambling… or not.. I can’t tell… me so tired…
–The loon
But it all depends really on how optimised middleware, compilers and operating systems are for these new extensions. If Intel has the hardware edge but AMD does a better job working with existing technology which requires no additional optimisations then it’ll be an interesting to see AMD pull ahead in terms of market share.
How so? Sandy Bridge will work well on existing codes without recompiling or rewriting.
They’re introducing new extensions and applications will have to be optimised to use them. Why is that so difficult to understand?
it appears this baby is packing drm in hardware, the only place where it matters.
Behold, citations!
http://www.reuters.com/article/idUSTRE7020FG20110103
Are there any performance numbers available yet? Also, does anybody know what socket this is? I have an I5 650 and was wondering if the I5/I7 variants will be direct drop ins for people currently using them.
the integrated graphics mean nothing to me, but if I can get a solid performance boost from my processor I may look into getting one if it works with my motherboard.
This looks enticing for an HTPC.
Whenever my Original Xbox decided to give up and die I will have to build one and this might be the processor for it.
So Intel makes this huge announcement and the stock dips, even on a big Up day in the markets. Didn’t make any sense, until…
Until I saw the date on the linked press release. INTERNATIONAL CONSUMER ELECTRONICS SHOW, Las Vegas, Jan. 7, 2010
This story is a year old.
This isn’t the first time that’s happened on this site…
This is on the frontpage of Ars and Sandy Bridge is indeed Intel’s latest architecture, so the story is new. Thom must have linked to the wrong announcement.
Most — if not all — benchmarks posted on the web state the Core i5 is significally faster than the Phenom II. This does not reflect my own experience. Or to be more precise, it is only a part of my experience.
I am architect of a .NET project with about 90MBytes of Code. Mainly C# code, but low level “drivers” have been coded using C++. The build script is a nant script, that is purely sequentiell. Building that project from scratch takes ~340s on a 2.66GHz Core i5 core, but “only” 230s on a 3.0GHz Phenom II core. Per GHz the Phenom II is ~25% faster. An 1.6GHz Atom N270 takes ~1025s to do the trick. But that core is another league.
Also in my experience buildung Java projects is significally faster on the Phenom II. The same is true for commercial workloads.
All these workloads share a high number of unpredictable jumps and a large active code size.
In performing 3D graphics, image manipution and multimedia de-/encoding the Core i5 is faster. Significally. But these workloads are completely different. Mainly tight loops with a priori known number of iterations and a small active code size. In each iteration the same set of operations are performed.
There are of benchmarks available for the SandyBridge. But only 3D graphics, image manipution and multimedia de-/encoding. The bright side of the Core i5 core.
Any numbers for commercial workloads ???
pica
Edited 2011-01-04 12:43 UTC