Slashdot Mirror
Intel Discusses Future Plans
heeeraldo writes "Tom's Hardware (unfortunately known for their one-page-stretched-into-nine articles, and endless ads) attended an Intel presentation about their future processor plans. The unsurprising bit: the endless march of additional cores. The surprising part: they're already focusing on 45nm processes." From the article: "Last week, Intel held a series of presentations at its Ronler Acres campus in Hillsboro, Oregon, whose facilities represent the main pillar of product design and manufacturing. These presentations included a short tour to the top-notch 65 nm production facility Fab D1D whose specifics Intel is currently replicating to other locations. The primary purpose of this show obviously was to convince around 80 analysts and journalists of the substantial health of Intel's 65 nm fabrication leadership, which is outputting new processors in high volume for launching new Pentium 4 6x1, Pentium D 900 and Core branded (known as Yonah) processors in early 2006."
What is AMD doing in the future? It's all well and good that intel is still playing catch up, but it's much more important to know what the technology leader is doing. Quad core CPU's next year, I understand, from AMD. Intel? Who knows.
I am government man, come from the government. The government has sent me. -- G.I.R.
Intel's 45-nanometer tech mentioned in NEC article
FTFA.
they're already focusing on 45nm processes
substantial health of Intel's 65 nm fabrication leadership, which is outputting new processors in high volume for launching new Pentium 4 6x1, Pentium D 900
Now I think we all know why Apple did what they did.
Take off every sig. For great justice.
The surprising part: they're already focusing on 45nm processes.
Thats the only way to dodge their inefficiency problems. Outside of like, designing better chips.
Someone set us up the bomb, so shine we are!
If Intel starts mass producing these then people will just buy what's avalable.
I couldn't help but think that this is just the newest assult as part of a press-release war between Intel and AMD. Recently, it seems AMD has been taunting Intel about the performance of its dual core technology. So it appears Intel's reponse is to say "your manufacturing process couldn't lithograph its way out of a paper bag."
Anyone know if they intend to use DUV/193 Immersion or EUV lithography for the 45-nm node? Last I checked, EUV had a long way to go before being usable...
The article is full of problems. Whitefield was canceled a couple of months ago:
http://www.theinquirer.net/?article=27192
There is a lot wrong with the 45nm code names, and in general, it is lacking a lot of info.
If Intel gave him this info, it is blurring the lines for PR purposes, and somewhat flat out wrong. As of Friday, Whitefield was still dead, and the roadmap didn't match up with Intel's internal ones.
There is a bit of right there, but few if anything that can't be found at the usual places.
-Charlie
how is intel planning on confusing customers with their new naming schemes?
From TFA - "The introduction of the Merom design will be a turning point in Intel's product policy, because it will be the backbone for all processor families that go into the desktop, the mobile or the enterprise space. In contrast, the desktop and enterprise markets are provided with Pentium 4 and Pentium D NetBurst architecture processors while the mobility CPUs are derived from the more efficient Pentium M design"
:)
Merom being the sucessor to the forthcoming Yonah. Based on the recent AnandTech benchmarks of Yonah against desktop chips , it seems like Intel may not have to play 'catch up' for much longer. Of course, we don't know what else AMD has up their sleeve
-- Ravi
Your dates are a tad out of la-la land, but your heart is in the right place.
http://www.theinquirer.net/?article=24638
It was linked here, but I can't find it.
The short story is that Intel's new VIIV boxes are crushingly DRM infested, and can load more. There is remote key revocation and all the things MS wanted, they are playing AMD off of Intel. Don't look to AMD to be any better, they are being screwed too.
I predict massive failure and egg on Intel's face here.
-Charlie
the next generation merom will, apparently, operatate at wattages of ~0.5 watts. Maybe now they will be a bit more cool. And, maybe, now I can buy a laptop that wont heat to the point of causing infertility. Kool. This better happen soon or there will be no "geeks, the next generation"(TM)!
When the seagulls follow the trawler, it's because they think sardines will be thrown in to the sea
Yeah, thanks for the corrections. I too saw the article, and I was just trying to remember the dates off the top of my head. I still can't wait to see how AMD will fare in the end of all this (if there is an end :P)
Back in my day I had to write games in BASIC, on a 4.7Mhz computer with no hard disk and 128K of RAM. And I was grateful
How do you implement anti-music/movie piracy at processor or chipset level? At that level surely music and movies are just bit patterns and data to be moved and processed. So how, at that level, do you distinguish between music/movies and any other data?
Who CARES what process tech they are using. At one time, the Northwood at .13 looked like it was going to KICK ASS on paper. It turned out to be an underperformer - to say the least. Let's not talk 'process', let's talk about IPC and how it's going to kick AMD's butt.
"...Well, there's egg and bacon; egg sausage and bacon; egg and spam; egg bacon and spam; egg bacon sausage and spam..."
And while we're at it, could AMD explain their CPU lines better (esp. to consumers)?
Who dreamt up the "Core" brand?
It's like when MS picked "SQL server" for their SQL server product.
A: What SQL server are you guys running?
B: Oh, we're using SQL server.
A: Yes, but *which* SQL server? Oracle? Sybase?
B: No, SQL server!
A: Yes but.... doh!
Now that everyone else have been selling multi-core processors for some time, Intel chose to brand their new processor geenration, of all things, "Core".
A: What multi-core processors are you guys using?
B: Oh, we're using multiple Core processors
A: Yes, but *which* multi-core processors?
B: We're using multiple Core processors!
*doh*
Oh dear oh dear...
That seems an ok basic description of what a wireless LAN is to me... what's shady here?
You're a tad hazy on facts and references too. Care to give any?
I'd be a little suspicious of any article that is titled "Top Secret Intel Processor Plans Uncovered" even if it wasn't from a long-time Intel fan-boy site. It's hardly surprising that Intel is moving to more cores with 65 and 45 nm. AMD started doing that two years ago and just opened their newest fab to facilitate quad-core and octa-core future cpus on much larger dies. Right now, AMD has at least a one-year lead over Intel in this technology and there's no sign that Intel is doing anything that will leapfrog AMD. The entire article could be entitled 'Intel says 'me too.' It would be much more impressive if there was some meaty info about the performance of actual products rather than a lot of stuff about 'xx will do this' and 'yy will do that.' As it is, it just reads like an expanded description of a roadmap which can quickly change with future developments or non-developments.
http://www.theinquirer.net/?article=13153
This one was wrong.
http://www.theinquirer.net/?article=13171
Here is the correction.
That is the one I can think of, if you got more, I'd be glad to post them for you. You seem utterly incapable of finding them on your own, so that kind of disproves your point.
-Charlie
Have people lost their interest in GHz's ?
http://www.theinquirer.net/?article=27957
Yup, not a problem. The last two are by far the best.
-Charlie
Still Amd has been light years ahead yet it isn't seen on a tv ad. The number of users that have amd on one form or another keep growing. Can intel keep up the heat, even though challenged by to start a camp fire with one of the new procesors... ohh no wait does that say intell?. 0nly time will tell...
In the past year and a half I converted all my home machines (12) over to AMD. Then I put a sticker on my front door - "Intel Outside".
The more you regulate a company, the worse its products become.
Whats the difference between Current Plans and Future Plans? Surely if you've made them they are Current Plans, and if you haven't made them yet then they don't exist, and even when you start thinking about making them they become 'Current Plans'....
Hey, its a fairly quiet thread!
You would think they'd have learned from the xbox...
They have to name them "Core"... Don't they have "ARM" already? Let me know when the Krogoth line comes out :)
am i the only one who finds highlighted words = ads revolting ?
There is only one way to stop this horrid thing: boycott toms hardware rev and the advertisers
What ever happened to cyrix? :) I owned one or two of those, back in the day. Maybe they've got a new super-secret mega processor in the works that will jump out and take the CPU market by storm!
Or, maybe not.
http://cubemonkey.net/quotes -- fortune-mod quote generator
there's a shortage.
Perhaps it is secondary to the point of the article itself, but did anyone else find the quality of the writing, particularly the final page, to be atrocious? There were a number of spelling errors that would have been flagged by even the most basic of spell-checkers, and the style of the concluding paragraphs was glib to say the least.
Today, it is all about squeezing the current manufacturing advantage in order to conquer middle earth and lock down brave AMD into its current 90 nm shire - although it is blossoming and as green as it can possibly be....
Yet we don't believe that Intel is going to drop the elfish Pentium brand with the next generation micro architecture.
I love Tolkien as much as any geek, but I find this analogy to be way over the top.
Intel is going to focus on its reorientation towards a platform company, but it won't be able to change the simple fact that performance decides over victory of defeat - whether you simply refer to processing speed by parallelism or clock speed, or you go for the increasingly important performance per Watt benchmark. One thing is for sure, though: The upcoming years will be all but boring.
What does "performance decides over victory of defeat" mean? And I especially love the final sentence - it is like something from the old Batman TV series. "Will our heros meet a fiery doom at the hands of the Riddler? Can Batman overcome once again and save Gotham? Tune in next week for another exciting episode!" [cue the music and credits].
This rant is now over.
...for me, since my company boycotts Intel, since they started building a 45nm fab in apartheid Israel.
-I like my women like I like my tea: green-
AMD is a second rate contender. Intel is already moving to 45 nm... AMDs stuck in the stone age. I have used intel motherboards with intel processors and matched kingston ram for years. Never had a problem.
BTW Intel does not have confusion naming conventions... Thats AMD.
Do you work for the inquirer? Because it certainly seems like it; it's the only source you seem to be able to site for your "superior" knowledge about a product that hasn't been released yet.
I think I'll wait to get actual evidence over a news journal that publishes every microprocessing rumor or grumble that runs across them [not that this Tom's article is much better, though].
"Victory means exit strategy, and it's important for the President to explain to us what the exit strategy is." G.W.Bush
Of the list, one of them is mine, and at the time, ATI was going to ship when I said it was.
Go back to the Anandtech forums, you'll look smarter there.
-Charlie
Yeah, I do, and I am only linking my articles where I have first hand knowledge. It saves me from having to type it all in again, and tends to be a more complete answer than a shorter one here.
-Charlie
Pick it's and my spelling apart if you will but here's the lo' down.
Being an Ex employee that played with all of these I know some crap about 'em
First of all Intel is useing a crap load of DRM in the new chips
Phoenix Trusted Computing platform or something similar to that name
But so far it's disabled and they will boot if the chip isn't their ( in these bios's at least )
Dempsey is already shipping to developers, it's a dual core hyper threaded in each core "xeon" processor
Yonah is Dual core mobile without hyper threading ( I never did see a single core of it and it's wierd that they say single core, because yonah in itself is two dolthans side by side ( same silicon looks like a p4 size, but p3 features ). It is fast and as far as we were concerned faster than Smithfield/Cedarmill/Conroe
Sorry guys but this isn't the first time i've seen you speculating and had it wrong, I'd think you guys would have contacts all over that major company, and it doesn't take a rocket scientist to figure out what's going on.
Is it just me, or is the Big Picture non optimal?
You know, Intel's wonderful state-of-art 65 nm fab line producing dual core Opteron's would be nice:)
"Provided by the management for your protection."
Not much about Itanic. Is it finally dead then, or was it out of topic?
Being in the Enterprise server business, it is difficult to gauge a manufacturer's press hype without seeing "real" data - power consumption is a major concern to businesses running more than a few servers. This type of usable data is typically only released by the manufacturer very near consumer release date, and sometimes real-world use data can only be gathered in real-world testing, regardless of what the manufacture states in press releases and white papers.
AMD far surpasses Intel, so far, in perfomance and scale per power power consumption. Our recent power testing shows that for similar spec Opteron vs. Xeon servers from big-name hardware manufacturers, the same power use would give us either 1000 Xeon servers to rack, or 1600 Opteron servers using the same amount of power. This is a no-brainer.
A recent WSJ article hit this on the head - posted in full here, since WSJ requires login (and the article appears to now be archived...).
---
Power-Hungry Computers Put Data Centers in Bind
Newer Hardware Guzzles Electricity and Spews Heat, Requiring Costly Alterations
By DON CLARK
Staff Reporter of THE WALL STREET JOURNAL
November 14, 2005; Page A1
The University at Buffalo installed a $2.3 million Dell Inc. supercomputer last summer, hoping to bolster its image as a research institution. Instead, the big machine came to symbolize an increasingly vexing problem for data centers world-wide.
Once the machine was delivered, university officials discovered they had only enough electrical power to switch on two-thirds of the system. They have temporarily responded by throttling back use of an older supercomputer, but a $20,000 electrical-system upgrade will be needed to run both systems at full capacity.
"The calculations that were done fell slightly short," says Bruce Holm, a senior vice provost at the school, which is part of the State University of New York. "The bottom line was that they missed."
More such misses are likely. That's because, in its long-running race to boost performance, the computer industry has hit a major hurdle: The newest hardware -- particularly the servers that run most business programs and Web sites -- draws too much electricity and generates too much heat.
The power-hungry machines, along with rising energy prices, are generating enormous utility bills and forcing changes on Silicon Valley technology suppliers that are akin to Detroit's struggle to improve gas mileage. (See related article1.) Though more-energy-efficient computers are on the way, it could be years before companies replace the systems they have already purchased.
In the meantime, bringing in more electricity and cooling is expensive and difficult in some data-center buildings. Organizations face unpleasant choices that include building new facilities, putting off server purchases or leaving costly space in computer rooms unoccupied to avoid overwhelming their air-conditioning systems.
Facilities planners at the University at Buffalo, for example, originally erred because they thought an older supercomputer would no longer be needed by the time their new machine arrived, Mr. Holm says. The need for both systems caused the university to consider spending as much as $150,000 to upgrade the current data center's air conditioning, just as the university was on the verge of moving the systems to a more modern computer room. "It's that kind of juggling act," Mr. Holm says.
If planners miscalculate, servers overheat, damaging circuitry or causing shutdowns that disrupt operations. The Uptime Institute, an organization that represents data-center managers, predicts that power-related problems this year will cause four of the 20 major failures typically experienced by members annually, up from two of 20 last year. "We are headed into a territory where there is no precedent," says Kenneth Brill, the group's executive director.
For years, no one worried much about power consumption.
> It seems that the G5 outperforms the Intel lineup on the desktop right now, for Mac users at least.
You're telling me; I bought Tiger from the Apple Store and tried putting it on my P4 Inspiron, and it didn't even boot up.
Maybe those G5 zealots are right, after all.
OMG! Wau!
"The surprising part: they're already focusing on 45nm processes."
That is not surprising; it's good management. While team 2 works on streamlining the current process, team 1 should already be working on the next generation.
That makes sure that the next generation is ready for mass-production by the time the current one runs out of steam.
I expect the OS to use all 100GB. Although it would be better with another 30GB of ram. I would expect the OS to boot up, and then start caching the entire HD to ram in the background. Obiously the most often used files first. That way I can keep working, and every second my chances of having lightning fast "disk access" becomes better and better. If I leave my computer running (As many of us do) I would definitly get a massive speed boost.
- Shared-state concurrency
- Message-passing concurrency
- Declarative concurrency (synchronization on logic variables)
A post from this mailing list: Lambda the Ultimate
Peter Van Roy - Concurrency-oriented programming blueArrow
10/21/2003; 5:06:42 AM (reads: 1765, responses: 20)
Concurrency-oriented programming is a phrase invented by Joe Armstrong, the main designer of Erlang. Basically, we would like to write applications where the concurrency follows the concurrency of the problem. This should be easy (language support) and cheap (implementation support). As far as I know, there are only two languages that support COP: Oz and Erlang (I would love to be proven wrong here--please give me evidence for others, if they exist! I mean good implementations, not paper designs.)
The majority of existing languages are sequential; concurrency was added as an afterthought. This makes concurrent programming difficult for them. A typical example is Java: it has monitors (shared-state concurrency) and expensive threads. Two years ago, when I told the head of our department I wanted to teach concurrent programming in a second-year course, he exploded "That's impossible!".
The reaction of our department head is understandable: shared-state concurrency, the kind that Java has, is the hardest to program in. There are two other kinds of concurrency that are just as practical, but much easier to program: message-passing concurrency (asynchronous messages between sequential objects) and declarative concurrency (threads and dataflow synchronization added to functional programming).
The easiest is declarative concurrency (see chapter 4 of CTM). This seems to have been forgotten by almost everybody. Yet it is not new: the first article on it I found was by Gilles Kahn in 1974.
Declarative concurrency is so nice that I believe it should be the baseline execution model for functional programming. (Not strict or lazy evaluation, which are both sequential.) This leads to many good things, for example here are two. (1) All the usual functional building blocks become concurrency patterns. For example, Map is a broadcast that collects results, and FoldL is the heart of a concurrent object with internal state (it accumulates an internal state from a stream of messages). (2) I/O becomes very simple: program input is a stream and program output is a stream. This is a perfectly good solution to the problem of declarative I/O.
When you compare the three ways of concurrency programming, shared-state is the worst and declarative concurrency is the best. Declarative concurrency prevents the most bugs and makes you waste the least amount of thought effort on concurrency. There are implementations of the different kinds of concurrency:
- Message-passing concurrency:
- Declarative concurrency (synchronization on logical variables):
All these languages are open source and free but they are functional programming languages. The difference betwee
- -- Truth addict for life.
Did I just miss it, or were they absolutely no references to the clock speeds of any CPUs in the article?
The cpu speeds have hit a plateau for the last couple years. (Yes, I know about the P4->PM core changes, but even considering that, clock speeds have been stagnant).
Will the changes to 65nm and later 45nm enable 4GHz++ clock speeds? Or, is all this multi-core talk implying that Intel will be "building out rather than building up"?
...what do you use your 12 machines for?
Ooooh, touched a sensitive nerve there, did I? Got too close to the revenue stream and the political support base then?
-I like my women like I like my tea: green-
Hey, look at that other major issue not mentioned - pricing. If a 65nm, dual-core chip rated at 3.2GHz on the inexpensive netburst architecture with only 2MB of L2 costs $1000, how much is a quad-core based on a 45nm process with 4+MB of L2 and all-new architecture going to cost?
Sure, therse new Intel chips are going to be really, really fast...but you could buy a dozen AMDs for the price of one!