Print 38 comment(s) - last by quiksilvr.. on Feb 12 at 2:00 PM

Fabtastic news

In January, DailyTech reported on Intel's Q4 results. During a conference call that we were a part of, Intel's President Paul Otellini repeatedly stated his commitment to the 32nm development process.

Today, DailyTech has learned that Intel's P1268 32nm development process is proceeding much faster than expected. Intel is therefore announcing an accelerated product ramp, and will be introducing Westmere-based mobile and desktop processors in the fourth quarter of 2009, with volume possible in time for the Christmas shopping season.

Intel is spending $7 billion over the next two years on equipment for 32nm. For all of 2009, Intel plans to spend $5.4 billion on Research and Development across all product lines.

Fab D1D, Intel's 300mm development fab in Hillsboro, Oregon, will be the prototype line, just like for the 45nm process. Over the next six months, this unique factory will turn the 32nm process from the research phase into the production phase. It will be in Hillsboro that Intel will adjust the production line to meet its yield and cost requirements. Once the process has been refined enough, Intel can use the lessons learned to convert other fabs to the 32nm process.

D1D is already outfitted with 32nm equipment, and yields so far are good. D1C and AFO (Aloha Factory Operations), also in Oregon, will ramp up in the fourth quarter of this year.

Intel's "Megafab", Fab 32 in Chandler, Arizona, will start 32nm production in early 2010, followed by Fab 11X in Rio Rancho, New Mexico.

Fab32 was Intel's first high volume 45nm microprocessor Fab. They spent $3 billion on a brand new building and the latest equipment to produce 45nm chips with hafnium-based high-k metal gate transistors, all on a 300mm wafer line. It first opened in 2007, with 184,000 square feet of clean room space. The completed Fab 32 structure measures 1 million square feet.

DailyTech will be presenting details of Intel's 32nm product roadmap later today.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

By Mr Perfect on 2/10/2009 1:16:19 PM , Rating: 2
So are they launching mainstream Nehalem parts before Q4 on 45nm, or are they just jumping straight to 32nm? It would be nice to have some i5s show up some time soon.

RE: i5s?
By quiksilvr on 2/10/2009 1:35:49 PM , Rating: 5
I'm tired of all the names and sub names (Core 2 Can't they just have two and just call it this:

Intel (Duo or Quad): (insert speed here, FSB, and cache); (insert chipsize here)

Stop wasting time with all the weird names and just call it what it is and give us all the info we need in one quick neat line.

RE: i5s?
By Anosh on 2/10/2009 1:40:51 PM , Rating: 5
And how would you avoid confusion when new parts are introduced? Stores won't just throw away the previous gen.

RE: i5s?
By StevoLincolnite on 2/11/2009 1:25:44 AM , Rating: 2
They could always adopt a PR+ system like AMD did with the Athlon XP, where it compares the performance to the previous generation.

RE: i5s?
By PhoenixKnight on 2/10/2009 1:53:21 PM , Rating: 1
Intel Duo 3.16GHz 800MHz 4MB 45nm?

Yes, that looks so much less confusing to the average consumer. Brilliant idea.

RE: i5s?
By quiksilvr on 2/10/2009 2:00:02 PM , Rating: 5
Its better than:

Intel(R) Core(TM)2 Duo Processor P8700 (2.53GHz)

Intel(R) Core(TM)2 Duo Processor T9400 (2.53 GHz)

...which one is better? Which one is worse? They have the same speed, but what does the letters mean?

RE: i5s?
By inighthawki on 2/10/09, Rating: 0
RE: i5s?
By quiksilvr on 2/10/2009 2:23:32 PM , Rating: 2
Uh... so not only does the P8700 have no support for Intel64 architecture, it has only 3MB cache. I..I fail to see the point you're trying to make. Did you mean the T9400 doesn't have the Intel64 support despite having more cache?

RE: i5s?
By quiksilvr on 2/10/2009 2:26:20 PM , Rating: 2
And further more, it DOES tell me more than before. I see that the difference between the two is the cache.

RE: i5s?
By rudolphna on 2/11/2009 10:53:11 AM , Rating: 2
so to the average, computer illiterate consumer, which would sound better and be less confusing?
Intel Duo 4.2Ghz 1666Mhz 10MB; 32nm
Intel Core 2 Duo E9200 4.2Ghz

How many consumers even CARE about anything besides the "Intel/AMD" "Duo/Quad-X3/X4" and "4.2Ghz"? Seriously, most people I know only know Intel 3gigahertz or some such. Sure, the 5% of people that are computer literate might be happy, but I would like to say I have an Intel Core 2 Duo E7300. As opposed to verbally saying, Intel Duo 2.66Ghz 1066MHZ 3MB; 45nm. Yeah THAT sure rolls right off the toungue. Like THAT wont confuse people even more.

RE: i5s?
By quiksilvr on 2/11/2009 11:48:23 AM , Rating: 2
You can just say Intel Duo 2.66. The extra info is just so that you know what you're buying without having to Google or Wiki E9200

RE: i5s?
By rudolphna on 2/11/2009 3:12:10 PM , Rating: 2
No because in the long run there would be no distinction between the "nehalem" Intel Duo/Quad and the Duo/Quad 5 years from now. (if still even using dual and quad cores)

RE: i5s?
By quiksilvr on 2/12/2009 2:00:15 PM , Rating: 2

RE: i5s?
By Smartless on 2/10/2009 2:02:52 PM , Rating: 2
Remember back in the day.... Intel 486-DX66... *sigh*

RE: i5s?
By Proxes on 2/10/2009 3:12:37 PM , Rating: 3

RE: i5s?
By Pneumothorax on 2/10/2009 2:01:06 PM , Rating: 3
C'mon don't you know it's for Joe the plumber to think that a E8400 is better than a E7600? I miss the old coppermine days when chips were listed as Pentium III 933/256/133/1.7V!

RE: i5s?
By Alpha4 on 2/10/2009 5:07:52 PM , Rating: 2
I thought that was the Tualiton.

RE: i5s?
By StevoLincolnite on 2/11/2009 1:28:33 AM , Rating: 2

The Tualatin was basically a test chip for Intel to test a die shrink for the Northwood, it was still a Pentium 3 at heart with more cache and some cool new features.

RE: i5s?
By inighthawki on 2/10/2009 2:01:37 PM , Rating: 4
As Anosh said, if you do that, besides the fact that it would confuse most people, you would be unable to tell apart certain models from different generations. Model numbers are a far superior way to label processors, and i think AMD has a good way of doing it myself.

RE: i5s?
By TheFace on 2/10/2009 2:02:58 PM , Rating: 2
Which would completely overlook the difference between a i7 that is 2.66 ghz and a C2Q that is 2.66 ghz. Or the C2D that is 3.0ghz and is 65nm and a C2D that is 3.0ghz and 45nm. Sorry this is confusing for you, but as you're reading a TECH website and articles about processors, you might want to play catch up with the rest of us.

RE: i5s?
By TheFace on 2/10/2009 2:04:59 PM , Rating: 2
maybe I should read before I spout off. Excuse me plz.

but will they be faster?
By kattanna on 2/10/09, Rating: 0
RE: but will they be faster?
By omnicronx on 2/10/2009 1:35:15 PM , Rating: 4
efficiency > speed

Clockspeed is not everything. This is something we all should have learned from the netburst days.

RE: but will they be faster?
By Anosh on 2/10/2009 1:45:52 PM , Rating: 3
Until developers learn to program for multi-core most software will take advantage of a single or if you're lucky dual cores.

Which means even if a core is efficient it would probably benefit the end user more if they'd remove two cores and try to increase the speed of the remaining cores.

RE: but will they be faster?
By Pirks on 2/10/2009 2:26:40 PM , Rating: 2
This is why I use Wolfdale for my gaming rig instead of Phenom II. Wolfdale's two cores are insanely fast with 6MB cache and 3.6 GHz overclock, and games benefit from them much more than from Phenom's four slower cores. I'm talking only about gaming, for other things like video encoding Phenom may be pretty decent choice. But for gaming - no way AMD. Sorry :)

RE: but will they be faster?
By melgross on 2/10/2009 2:21:27 PM , Rating: 3
Clockspeed ISN'T everything. But within the same line, it is.

Besides, clockspeeds have been moving up again.

In another two years, likely less, we'll be at the fastest Prescott levels.

RE: but will they be faster?
By Master Kenobi on 2/10/2009 3:49:22 PM , Rating: 2
Most of us are already hitting 4.8GHz or faster on Core i7 CPU's ON AIR.

RE: but will they be faster?
By TomZ on 2/10/2009 4:36:45 PM , Rating: 1
efficiency > speed
Not for me - I prefer raw speed first, efficiency second.

Having "lived through" a machine with a pair of AMD Athlon MP's (a.k.a. space heater) as well as a nice, fast, hot Netburst comparison the efficiency of today's Core 2 and Core i7 is easy to live with.

RE: but will they be faster?
By teckytech9 on 2/10/2009 8:28:27 PM , Rating: 3
Speed comes with a price. Economics needs factoring into the equation. Realistically, if one jumped into the i7 bandwagon without waiting for the competition to answer, then they would have payed a premium. Less a 40% or more discount for waiting, using the same socket, is more bang for the buck. The cost of a new motherboard must also be factored in too.

RE: but will they be faster?
By paydirt on 2/12/2009 8:42:09 AM , Rating: 2
...and DDR3 RAM

RE: but will they be faster?
By amanojaku on 2/10/2009 1:53:45 PM , Rating: 3
Look, there is no ONE variable that determines system speed. One of the biggest challenges to multi-process (i.e. programs, drivers, kernels) systems is context switching. Adding more cores helps alleviate the burden of context switching. In the past this was a problem because the cores were on different sockets, which meant inter-process communication across cores was limited by the speed of the motherboard or processor interconnects. Multicore CPUs allows for process communication at the speed of the CPU die. In other words, an properly designed OS should perform better on a multi-core system than a multi-socket system, and a multi-socket system (as most of of us know) is faster than a single CPU system. As far as I know, increasing the clock speed does nothing to fix the problem of context switching. Besides, the speed of your CPU doesn't factor into things like web browsing (except Flash or having many tabs open,) IM, etc..., but all systems context switch.

Sandy Bridge
By antdays on 2/10/2009 1:52:29 PM , Rating: 2
Sandy Bridge, where art thou?

RE: Sandy Bridge
By fwkitziger on 2/10/2009 2:29:37 PM , Rating: 2
Looks like Sandy Bridge is post-Clarksdale, a hybrid design with IGP (45nm), memory-controller and Westmere core (32nm) all on a one-chip package. Sandy Bridge would be the next iteration, all 32nm. Great article here:

32nm = Better Overclockability
By Shig on 2/10/2009 2:15:45 PM , Rating: 2
Can't wait to see how westmere overclocks.

By Master Kenobi on 2/10/2009 3:50:36 PM , Rating: 2
With any luck it will be a drop in upgrade to my current i7 system.

By kontorotsui on 2/11/2009 4:47:46 AM , Rating: 3
Size matters not.

Busy busy
By Cullinaire on 2/10/2009 3:49:37 PM , Rating: 2
It's getting pretty busy here with the ramp going on - but then again being busy can't possibly be a bad thing, especially in this climate!

By crackedwiseman on 2/10/2009 5:49:35 PM , Rating: 1
technically, Intel's dual processor server chips were supposed to be moved to the 32nm process in Q1 2010. So, if things go as planned, we'll see 32nm a full quarter earlier. Yay for Intel! But wait! - this also means that the mainstream market will have to wait a solid 9 months for anything new from Intel (instead of Q3/Q4 2009, we have a late Q4 2009 launch). Oh well... we can pay a [airquote] reasonable [end airquote] sub $300 (what a steal!!) cpu if we only want to wait 4-6 months for a crippled quad core

Hey - speaking of DP server CPUs, anyone think March 29th is a really late "Q1" 2009 launch for something that was reportedly tearing through benchmarks last month, although maybe they're just being considerate, because god only knows that we need the extra time to save an extra $700 (yes, a 2.66Ghz quad DP CPU costs $1000!?) for that second enabled QPI link and ECC memory functions on our 2.66GHz quad-cores. I thought an unlocked CPU multiplier was good bang for my buck, but this just blows me away.

Also, I'm getting a strange sense of deja vu - that new chip on our mainstream processor seems vaguely familiar. Lets see here, integrated graphics, memory controller, PCI Express, DMI links to an I/O chip, one manufacturing process behind the main processing unit, separate dies... I could have sworn that that was exactly how a northbridge went about its business - but it can't be a northbridge - its on the same package as the processor, and its all "integrated" now, whatever the hell that means. I thought the meaning of the word had something to do with existing on the same piece of silicon, but what do I know - I just learned all this from hype, and we all know how meaningless that is, don't we? :P

"We are going to continue to work with them to make sure they understand the reality of the Internet.  A lot of these people don't have Ph.Ds, and they don't have a degree in computer science." -- RIM co-CEO Michael Lazaridis
Related Articles

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki