Slashdot Mirror
Intel Says No SMP Support For Pentium 4
the Man in Black writes: "AMD dropping the Mustang core to concentrate on an SMP solution seemed to bode ill at the time, but it seems that this was the wisest possible decision, given the below news.
ZDNet is reporting that Intel will not have dual-processor support for the Pentium 4 at launch time ... indeed, not until the second half on next year, when the Pentium 4 is re-released with a new core."
According to this a dual board for the AMD T-bird may be out in Q1 from Tyan. Given that the following ZDNET commentary basically states that the 1.5 Willy stinks in Windoze, it may be that AMD is going to be able to penetrate big time in the high end market. Since Intel, has really ceeded the low end market to AMD, AMD may be in a nice position for the next several quarters.
I'd say it's probably because making a dual-processor system is hard and takes time. The P4 is a new processor with a new core. I suppose they could have made the decision to release single- and dual-processor systems at the same time, but that would probably mean waiting until next year for both That doesn't help Intel or their customers. Or AMD's customers, for that matter. Incidentally, how many years has it been before AMD's first dual-processor system? What are THEY smoking?
That's the short version of things, anyway.
----------
I like your idea for naming the P5 the P-P-P-P- Pentium.
It could go "P-P-P-Pick up a Pentium" and feature five dancing pengiums (a la Penguin buscuits) to replace the blue men adverts.
Why not go the whole hog and call the P5 the Penguin, then you could "P-P-P-P pick up a Penguin".
All the newbies would start to think Intel=Penguin=Linux.
Ok, so SMP is not supported at launch time. Is this the same "not supported" as Celeron SMP, which many people have in their box right now? Could a mobo maker come out with a board that allows P4 SMP, and just not have it supported by intel?
Or is this "not supported" as in not doable?
First of all I haven't seen any benchmarks yet, but the 1.2 gig tbird is probably going to be on pace with the 1.4gig P4. Add the fact that it can't dual and the tbirds soon will then there is more trouble in Intel land. So they tell us its going to be another 6-9 months to get the dualed up? Ouch. Good luck Intel, you're gonna need it.
Oh, wait. Damn.
Now AMD really will get the chance to conquer the big-end market!
And that's what we want? What we want is AMD taking over 50% of the market so we can see two giants bashing each other's heads in with us being the beneficiaries of it all. I wouldn't Intel losing some market share to Transmeta though...3 major players would be really cool...
People replying to my sig annoy me. That's why I change it all the time.
Revive Burger Kings flop Herb ads
Symbolic sheep or lemmings leaping off a cliff, but falling much faster now.
Bunny suited dancers and 'Who let the dogs out'
Something so incomprehensible you're not ever positive it's an Intel ad
A big chart illustrating how it's nearly as fast as the Piii
Co-venture ad with Alcoa, on the virtue of 1Lb aluminum heatsinks
The invisible man showing off the available motherboards against a black background (indistinguishable from your set being off)
Other ideas?
--
A feeling of having made the same mistake before: Deja Foobar
even Apple has a dual-processor system available, fer chrissakes. Soon AMD-baseds dual-processor motherboards will be available. What are the guys at Intel smoking?
Maybe they're trying to emulate Motorola's slide into desktop-processor oblivion...
I use Macs for work, Linux for education, and Windows for cardplaying.
Strangely enough, the Pentium 4 is actually targeted at the market where Intel is being hurt the most by AMD, the home computer. The Pentium 3 has been getting thrashed by AMD's chips in the home market. The home market is also where MHz (or now GHz) count. I could create a 2GHz 386 and it would probably sell better than anything else on the market, even if it was inferior in every other way. This is once again proof that the common Joe user is entirely overwhelmed by the actual differences in different solutions. I mean... the bigger the number, the better it should be, right?
The "new" Pentium 4 is already slated to be obsoleted by the next Pentium 4 chip/chipset. I honestly believe that Intel is releasing the P4 to slow the spread of AMD. On the server side, they are still pushing the Pentium 3 Xeon line... and probably will until they get the P4 going in SMP where they'll probably make a "P4 Xeon" so that they can rape companies even harder.
What I am really curious about it exactly how viable AMD's chips are going to actually be in the server market. They are going to need to make large cache versions of the chip... not to mention, does the 760MP chipset have the scalability? Can it support Quad and 8-way SMP configurations? AMD may overtake Intel on the lower-end server market by offering a Dual CPU solution, but unless they are capable of these other configurations, Intel will still be king in the server room. Remember, one of the main differences between the P3 and the P3 Xeon is the extra "glue" logic that Intel has added to the chips to allow for more than Dual CPU configurations.
I really am looking forward to seeing what AMD has to offer in this area. I also can't wait to see the great commercials that Intel puts out to advertise the new chip. Those blue guys crack me the hell up. Just think, in a couple months, we'll have Joe User going to Best Buy and picking out his 1.5GHz P4 systems out of the showcase.
For a uniprocessor system, I think that the P4 may actually reclaim the crown for fastest chip in the x86 market... I just wonder how long Intel will be able to hold out with AMD right on their back.
wolf31o2 Developer, Gentoo Linux Games Team
That's not how to burn karma. Here's how to burn karma.
Slashdot sucks, you closed minded linux zealots. Your OS does too. That's right, I said it: WINDOWS IS BETTER THAN LINUX!!! One more time, just for fun: WINDOWS IS BETTER THAN LINUX!!! Damn that's fun. If you think that Windows is so crappy, then why don't you do something about it? That's right, you have created a virtually useless "OS" that averages two lines of code per developer ("developer" == 1337 5kr1p7 K1dd13), explained on jumbled buletin boards written by people who don't know the difference between "then" and "than". You tried to make Netscape better, and I haven't seen so many bugs since Mandrake 7! Do something fun with your free time rather than rewriting software so you don't have to pay for it. Get a life!
...and if that doesn't work, you can always insult the British. Happy anti-whoring!
--
Win98 sux without these 1337 toolz !!
The Celeron isn't really an SMP-incapable chip. The microprocessor itself has full SMP support, with the only thing keeping users from using a Cel in a dual processor system is fancy packaging. That's why it was so easy to make a mobo like the aBit BP6.
With P4s, on the other hand, the processor core doesn't support it, so the closest you're going to get is a Beowulf cluster.
No, read the ZDNet article. It is literally there.
Just as long as my CPU runs at a faster frequency than my roomate's, I will be happy. NO I dont want to see how many fps my machine gets versus yours.
duh. did you even read the rest of his comment?
cpeterso
Point taken.
However, you can't automatically say that the AMD 760MP is going to suck. You haven't seen it yet. Remember, the i820 was supposed to be a BX on Steroids, not the piece of crap it is now. It could be the next BX.
Okay, even I doubt that, but if the best chipset available today is three or four years old, that has to tell you something about the sorry state of the chipset market right now.
Is this post not nifty? Sluggy Freelance. Worshi
You know, one of these days Intel is going to stop selling the BX. Intel has *not* made another chipset that can compare to the BX, and when it drops from the motherboard landscape, I imagine the AMD 760MP is going to look a lot more attractive.
I mean Hell, the VIA chipsets are starting to look better than the Intel ones, after you remove the BX from the equation. VIA!
So, by your reasoning, no SMP chipset will succeed once the BX is discontinued.
(Pretzel logic at it's finest).
Is this post not nifty? Sluggy Freelance. Worshi
Not much choice will depend on things more than just speed. I'd just as soon stick with P3 Xeons at 700mhz than a 1.4ghz chip with a small cache, poor or no (n>2)-way SMP support, and based on boards that haven't been shown to be stable. There is more to life than clock-speed. There's even more to life than reliability and stability (there are reliable and stable solutions other than Intel, and at least so-far on uni systems AMD is actually one of them...)
What is the pipelining architecture going to be like on these AMD duals? Will the SMP routing backplane be fast enough? Cache size?
How do the two instruction sets compare? Can they both do the same things in-chip? Can one do them in fewer instructions than the other - and if so, how does this wind up modifying the clock speed / real performance gain ratios?
There is also compatibility. Our first foray into using several AMD machines as single-proc servers was aborted because we had more problems with hardware and OS compatibilty than with P3's or even Celerons. We didn't give up on AMD completely, we just put the machines on corporate user desktops and vowed to try new AMD machines out one-at-a-time from now on and see if the problems get fixed...
The actual comparisons will require some research, the information for which I can't find at this time - so what's the point of making judgements one way or the other regarding the tech?
Regarding the marketplace - Intel is certainly responding to AMD in the home market making big gains with high clock-speed chips. That's fine for the home market, they can compete for clock-speed afficianados and maybe AMD's dual 1.2s will come out before Intel's dual 1.4s and there will be more competition and more scrambling by Intel to recapture home/gamer/office geek market share...
But in the server room, and in the IT Depts run by people who know their stuff, the verdict isn't in yet regarding whether or not AMD can give Intel a run for their money...
o/~ we are pissed, we are pissed, we have to resist... o/~ - ec8or
Not really. The P4 is a consumer chip. Sure, it might be used in small servers, but that's not where Intel make their money. You can bet that when Intel release a Xeon version of the P4, it'll scale to lots of CPUs. The Xeon line is what Intel expect to go in servers, not the current P4.
"The invisible and the non-existent look very much alike." -- Delos B. McKown
Near as I can tell, there are a group of people on here who will come up with any excuse for Intel.
I bet soon it will be: You should stick with Intel because they will be faster than AMD in their next generation. Or something equally stupid. Any excuse for Intel.
One is tempted to wonder where these people's heads are. Or perhaps not, since most people don't like to think about smelly, dark places that are full of solid waste.
Need a Python, C++, Unix, Linux develop
If it's big, black and you can't figure out why you need one, it must be impressive!
--
A feeling of having made the same mistake before: Deja Foobar
...anything to do with aluminium (or aluminum)!!
[Think heatsinks]
Donte Alistair Anderson Roberts - hi son!
Karma: Chameleon
personally, i don't know if i could ever go back to single proc machines. i've been doubling up since the p-pro.
screw intel. my next machine will be a double amd box.
http://kered.org
I guess this means the P4 will have DeCSS in microcode. ;)
This was well known about a year ago, when it was announved that the first P4s would be on a .18 micron process, and only after they switch to .13 micron would they release the server/workstation version of the Pentium 4.
Anyway, do you really want two 50W processors inside your case? Wait a minute, I guess that's what a dual K7 system would look like anyway.
Tarsnap: Online backups for the truly paranoid
If Intel keeps delaying the Pentium 4, they might as well just skip it and go straight for the Pentium 5. This got me thinking that the name "Pentium" was Intel's way of avoiding calling their next chip the 586. So maybe Intel is afraid of the number 5, which means they won't have a Pentium 5 after all. What will they call it instead? Penta-Pentium? Or will they do something like making the "5" an exponent, as in "Pentium to the 5th power"? Or maybe just P-P-P-P-Pentium? Or Pentiummier? Anyone else have any other stupid ideas?
And now, an unnamed company exec says, "It's possible that some of our sales force overstated the benefits of dual-capable CPU systems, unfortunately, by being overly critical of single-CPU-capable systems."
Paul Otellini, the exec VP and GM of Intel's Architecture Group, reported on the workstation performance of the P4: "The Pentium 4 processors that we're announcing Monday have the highest performing floating point of any PC processor that's out there. And, in fact, [they] compare very favorably to a lot of RISC microprocessors which for so long have been resident in things like workstations. That's one of the reasons you'll see on Monday that there are workstations also being introduced with Pentium 4." I'd like to see the quantitative results of some tests to back this up. Intel just might steal the FPU crown from AMD, and a good thing too: as it stands now, the AMD Athlon series has been shunned by C/C++/assembly programmers due to the fact that their projects will not compile with a 100% compatibility guarantee. The general rule for the last ten years has been: "If you're programming for the PC, use Intel."
"Ancillary does not mean you get to rule the world." --U.S. Circuit Judge Harry Edwards, speaking to the FCC's lawyer
It will not. I mean NOT AT ALL. No servers. Nada. Non. The reason is very simple - server also means lots of RAM. And there is no non-RDRAM solution for P4 currently available. The average server currently ships with at least 0.5G RAM, usually 1G or 2G (for those brave or stupid to run Intel on a 32 bit system in non-flat mode). The price tag on such RDRAM system puts P4 outside of the server market completely for now.
And IMHO this is the reason for Intel strange behaviour and trying to bail out of the RDRAM obligations. They got their marketing onto completely new grounds (no server release to show off and the much thinner profit margins) where they do not feel comfortable.
Baker's Law: Misery no longer loves company. Nowadays it insists on it
http://www.sigsegv.cx/
Maybe I'll just call IBM and see if I can get one of those ASCI yellow's. Or was it green? Beige? Magenta? Oh, hell, I can't remember.
Anyway, I thought that the Pentuim 4 itself was supposed to be physically huge, yes? Wouldn't haveing multiple P4's then require you to have a case the size of a coffin? Though it could be possible o heat your home with such a system...
Hi! This is the Sig, blatantly attached to the end of this comment.
"It's possible that some of our sales force overstated the benefits of dual-capable CPU systems, unfortunately, by being overly critical of single-CPU-capable systems," said the executive, who asked not to be named.
Intel usually targets new systems at servers. That way they can charge thousands for the new chips. How will that work if SMP is not available for several months?
The Mustang had some serious fab-cost problems (mainly in the large amount of cache). There's nothing wrong with this if you can sell a large volume, but AMD just didn't see the volume for the server market right now.
With Intel lacking SMP support for the P4 through next year, AMD will be ahead with a Dual [name your Athlon variation here].
At the same time, this doesn't make Intel incompetent. Intel knows that most servers are not built on the latest chip, rather, chips that have been well tested. Server CPU's are usually a few steps before the top of the line. By the time Compaq, Dell, etc. are comfortable with the P4, SMP will most likely be available.
There is no longer anything that can be done with computers that is nontrivial and clearly legal. -- Paul Phillips
That sucks, now I can't get an extra 10% out of my machine and roast chicken in my computer case... I guess it's back to toasting bread on my p3...
--
Feminism is the wild notion that women are human beings.
Okay, wait a second... a one pound heatsink??? Can anyone explain this one to me? I thought heat sinks could be made light because what you're theoretically after is something shaped so it has lots of surface area and a good heat disapation index.
So how did they manage to make the sucker weigh a pound? Does the P4 generate so much heat (a bad sign, IMAO) that it needs a one pounder?
I have to wonder if the second or third generation P4's will be any better (well, from the looks of things, they can't get worse).
Christ, I think the heat sink on my POS computer weighs a couple of ounces...
Kierthos
Mr. Hu is not a ninja.
What are most of the SMP machines around the world right now? Right, Intel P3 Katmai/Coppermine with SDRAM/ECC in them. Sure, AMD is trying to release an SMP board for the Athlon, but it won't succeed unless it can be proven to be as stable as a 440BX with 2 processors. Of course, the 31337 H4X0R5 will jump into the Athlon SMP bandwagon once it's released, but it won't be accepted by legitimate users until its stability is proven.
"Ancillary does not mean you get to rule the world." --U.S. Circuit Judge Harry Edwards, speaking to the FCC's lawyer
Molog
So Linus, what are we doing tonight?
So Linus, what are we going to do tonight?
The same thing we do every night Tux. Try to take over the world!
He returned to his desk, loaded the newly acquired pistol, took careful aim at his foot and pulled the trigger.
He is listed in satisfactory condition.
Multimedia encryption? What the hell is that? Where does ZDNet find these people it tries to pass off as writers, anyway?
I've been a long time supporter of AMD and really love their products. I'm glad that there is competion now pushing prices down and developement ahead. Unfortunately it seems like intel is rushing too much trying to keep ahead and making several mistakes. While I prefer AMD I certainly don't want to see intel loose too much of the market (I know it would take quite a few years for this to happen if it does). I'd hate to see a one processor market again.
Looks like AMD did bring the dual board in at the right time. It will be interesting to see if this stratagy and Intel's mistakes can really get them in the server market. As of right now, I don't know many people willing to give up there Pentium servers for AMD, but without dual support they may not have much choice.
I read somewhere, can't remember, but it said the 1.5 GHz P4 was actually slower than the P3 at 933 MHz. What kind of improvement is that?
Meanwhile, the Athlon continues to kick ass.
~ferich
Oh, if only there were someone so cynical as to start making 2GHz 386 CPUs!
.13micron process. They'd be low-power, low-heat chips. Toss 'em on a cheap-ass mobo with 66MHz bottlenecks.
I'd piss myself laughing.
[and, you know, I think it's entirely plausible: the 386 was a lot more simple than the Pentium-class CPUs. Combine that with
Gahd. I hope someone with a chip fab in their basement is reading this. I really want to see it happen!]
--
--
Don't like it? Respond with words, not karma.
Next time, they will announce it doesn't run windows 95 for being a 16 bit processor.