IBM has carved out a renegade path for the upcoming Power6 processor, opting to crank the chip’s GHz much higher while rivals shy away from major clock speed boosts with their products. The Power6 chip will run between 4GHz and 5GHz and has been shown to hum away at 6GHz in the lab. IBM reckons that some process technology breakthroughs have allowed it to kick GHz higher while still keeping heat and power consumption issues under control. All told, IBM claims that Power6 will be twice as fast as competing server processors from Intel, AMD and Sun.
They also claim that even with such speeds, they want the chip to consume less power ((err… forget the joke and continue writing, forget…)) than some desktop processors…
That would be interesting… a bit late, but still interesting to see.
I hope we get some (3rd party) benchmarks too.
I don’t think we should hold our breath, from the article :
“Internal IBM roadmaps once showed that the chip might arrive in 2006, although a mid-2007 delivery date now looks most likely.”
No wonder Apple moved towards Intel. This just seems like an attempt to get at Apple for leaving IBM’s bed in favour of Intel’s. Benchmarks and marketing claims are cheap, we want real life comparisons and we’ll have to wait for them but in the mean time who knows what AMD and Intel have up their sleeve
IBM, does make some go stuff, but intel has pretty good lower power stuff now. and a road map to even more products with it and future growth.
IBM saying “look we have it too!!!”
does not make apples choice wrong, there where many factors invovled in the choice of switching to intel…..
and when will the first of these ship? 2007? later?
meanwhile intel and amd lower power chips will have been out for years already and will be entering a stable and mature stage.
-Nex6
IBM has said the new line is 3x to 4x faster than the Intel solution.
🙂
IBM needs to stop talking and start doing. They always seem to have this revolutionary technology in their labs but then it never makes it to market or comes to market too late to get mass adoption.
IBM’s recent messages seem to be all over the place and often contradictory. But to sum them up:
Intel/AMD/SPARC/etc. suck because they are not Power.
AMD sucks even more because Sun is pushing it.
Intel sucks even more because Apple chose it over Power.
Solaris sucks because it is not Linux.
Red Hat sucks because they are getting too big and IBM cant buy them without alienating it’s AIX customer base.
Novell rules because Suse has 85% of the Linux on mainframe market share and mainframe sales mean big profit.
And of course if Novell started getting too big, then they would make it on the hit list also.
Anyway, I’m not sure what to make of IBM these days. I guess it all makes sense when you look at how large they are and how many markets they have their hands in… Maybe that’s the problem, IBM is just too big.
they have no vision.
or worst, they don’t have any idea what is their purpose, in what business they are. A company without a leader, just stock holder.
IBM and AMD are partners. They sell Opterons, they work with AMD on process. IBM “does.” Compared to some of the names in your list, it “does” to such a degree as to look like a giant in comparison. Simply because they expect their POWER6 line to perform better in their target market than Opteron-based servers hardly stands as an indictment against AMD.
Despite the trend here, the world doesn’t revolve around Apple computer.
“Despite the trend here, the world doesn’t revolve around Apple computer.”
One important fact-of-life you seem to be missing here. Your product is only as good as your consumers believe, in business. Apple may be a tiny share of the market, but their products (especially now, post-ipod) are VERY well known. What Apple does, as tiny as they are, shifts attention very well, and attracts attention that would otherwise not be given.
Just look at this site, and how much response there is to Apple related topics. Look at Apple’s meteoric growth lately. Don’t be surprised when Intel makes out like a fat cow in 400 acres of tall grass. No matter how you look at it, the cost of 1000 engineers and a facility as previously stated in a post is minimal compared to the advertising value they will get from Apple. IBM was blind to this, their money comes from services/support contracts. Look at all the different things IBM sells. Apple didn’t bring any of that to IBM. Intel, on the other hand, makes it money from it’s hardware. Apple has the ability to influence that market greatly. Hence, the switch. The end.
The hardware world/commidity PC world, does revolve around Apple in a fundamental way. Time will tell/show. The “hip” always win. Look at Tommy Hillfiger, Gucci, and so on. They sell shitty products for enormous amounts, making huge profit with minimal cost. All for the name of the brand. Apple has that. Intel can profit. Not only that, but it tosses their name out into anybody who leaves their home once a month’s mind. I can’t speak of a day I’ve been outside that I haven’t seen an iPod in the US, an Apple laptop, or somebody talking about Apple this or Apple that. Intel has nothing to lose and everything to gain. IBM is in a totally different market, and they gained nothing from Apple. This is simple economics at play. Just don’t belittle Apple using something unrealted as a reference.
Edited 2006-02-08 05:30
Please name the facility with “1,000 engineers” devoted to catering to Apple’s needs. Neither the world nor the computing world revolve around Apple Computer. Intel does not obtain more money from processors it sells in iMacs than it does for identical processors sold in a variety of laptops. Two guesses which sells a larger volume.
IBM needs to stop talking and start doing. They always seem to have this revolutionary technology in their labs but then it never makes it to market or comes to market too late to get mass adoption.
Did they not just finish putting 3 power-based machines on the top of the top500 list?
You people actually think IBM cares about Apple? Laugh. Newsflash: The world doesn’t revolve around Apple. Businesses don’t plan their strategies around Apple and they don’t do R&D based on what Apple wants. Steve Jobs doesn’t hold all CEO’s by the groin and CIO’s don’t jump when Jobs tells them to.
Where is all the common sense going? Apple fanboys seem to live in a dreamworld. How irritating.
[QUOTE]Businesses don’t plan their strategies around Apple and they don’t do R&D based on what Apple wants. Steve Jobs doesn’t hold all CEO’s by the groin and CIO’s don’t jump when Jobs tells them to.[/QUOTE]
I think that’s why Apple switched to Intel. IBM were doing their own thing and not catering to Apple/Steve’s demands. Intel, on the other hand, wanted Apple’s business really bad, and offered to do just that. They opened a 1000-engineer facility to cater to Apple’s needs, and even Intel’s CEO walked out at MacWorld in a bunny suit at Steve’s request.
They opened a 1000-engineer facility to cater to Apple’s needs
Yeah, you’re going to have to provide a citation for that one.
>> They opened a 1000-engineer facility to cater to Apple’s needs
> Yeah, you’re going to have to provide a citation for that one.
Had you quoted the entire sentence, you would have quoted said source: Macworld 2006. OTOH, I don’t think they said 1000-engineer, if I recall correctly they said 1000 people from Intel working with Apple’s engineers, which is obviously not quite the same.
That being said… the Power6 is/will be a server CPU, so it wouldn’t have made the switch any less desirable, particularly due to the timelines involved.
Edited 2006-02-08 06:46
Actually, I didn’t quote anything (concerning that post). Hence no quote marks, no italics, etc. Regalrdless of it being engineers or “people” (quotations for those who need them), you are absolutely correct. IBM only makes money from it’s hardware division in large scale/mainframe applications. This is in agreement with my previous post. You post simply agrees with those above you (and on slashdot.)
Cheers
[Edited for clarity about quotations]
Edited 2006-02-08 07:27
Talk about 4-5GHz is fairly useless without considering IPC. There is no way IBM will be able to get 4-5GHz while keeping POWER5’s IPC. Heck, I’d be impressed if they did that while keeping POWER4’s IPC, since POWER6 won’t be as wide a design. Moreover, 5GHz seems optimistic — Cell was rumored to debut at 4GHz, and had been reported running at 5.6 GHz in the lab, but it looks like practical implementations will come out at 3.2 GHz.
To be equitable to IBM, let’s consider the performance of a 4GHz Power6 that keeps the IPC of POWER4. POWER4 gets 639 SPECint/GHz. That means that IBM’s 4GHz chip might get a SPECint of around ~2600. That’s good for today (much faster than 2.8 GHz Opteron), but consider what x86 will have by then. It is reported that Conroe will offer 20-30% more IPC than Yonah (as it should, being a completely new 4-issue core). Even if we assume a conservative 5% improvement in integer performance, that puts a 3 GHz Conroe at the level of a 4 GHz POWER6. Moreover, Conroe will be out a year before POWER6, and be able to offer its performance without the massive expensive caches and memory busses POWER designs are tuned for.
That’s the equitable comparison. A more realistic one is POWER6 at 3.6-3.8 GHz with the IPC of a PPC970. That chip could be estimated at 2000-2200 SPECint, or in the neighborhood of the top-end Merom when it comes out in a few months.
Of course, the comparison itself is somewhat contrived, since if you’re in the situation where you’re considering a POWER6 machine, Merom isn’t even on your radar. However, the comparison has implications. It means that POWER6 will be unlikely to do anything to slow down Opteron’s growth in the low-end to mid-range market segment, nor will it do anything to counter Merom-based designs in the blade server market. Actually, it should basically maintain the current status quo, in which POWER is only a realistic competitor to x86 in extreme high-end markets running apps where its massive caches and memory subsystem come into play. Of course, its quite apparent that this is precisely the market that IBM targets POWER to, and even stating the fact seems disingenuous. The only reason I ever bother is because IBM’s recent cheerleading of POWER and power.org suggests their desire to see PowerPC in more plebian machines.
Edited 2006-02-07 22:52
IBM wants to distribute the cost of maintaining its architecture. If they don’t, then at some point in the future POWER is toast. If anything, they probably want to capture more of the embedded market mindshare. Amiga sure isn’t going to keep POWER in the game.
Speculating about the performance at this point is futile, and it’s probably best to just ignore IBM’s comments about Sun and AMD performance. Though they’re probably referring to raw floating-point performance if anything.
I can certainly understand why IBM wants more adoption of POWER. My point is simply that as it stands, IBM doesn’t offer anything that would warrent increased adoption in the workstation/mid-range server markets.
Well, what I’m saying is if they’ve shown a serious interest in targetting workstations that would otherwise go x86-64, through moving POWER to a more “open” process, they’ve hidden it pretty well. That’s increasingly a pointless market to even try to compete in, in terms of return on investment. They just don’t want their architecture to become so completely dominated by their own niche that it starves to death in mindshare, and is ultimately just replaced entirely. It’s probably futile in the long-term, since the x86 has had considerable worse-is-better results even to Intel’s chagrin.
endy: “No wonder Apple moved towards Intel. This just seems like an attempt to get at Apple for leaving IBM’s bed in favour of Intel’s.”
these chips have been in production way to long to have had any influence from apples move. also, apple doesnt use power chips, the current powerpc970 is based of the old power4.
nex6: “IBM saying “look we have it too!!!”
does not make apples choice wrong, there where many factors invovled in the choice of switching to intel….. ”
ibm isnt playing catch-up here, they are way ahead of intel. still apple doesnt have any effect on this cpu range or ibms motivations.
amigascne: “IBM needs to stop talking and start doing. They always seem to have this revolutionary technology in their labs but then it never makes it to market or comes to market too late to get mass adoption.”
please name some examples. seems to me ibm is the markedleader with plenty of revolutionary technology to show. be it the cell or the internals of the power line.
youknowmewell: “Where is all the common sense going? Apple fanboys seem to live in a dreamworld.”
yes, seems to be a trend. but you cant expect people to actually have a clue before they utter nonsense on a forum.
sanctus: “they have no vision.
or worst, they don’t have any idea what is their purpose, in what business they are. A company without a leader, just stock holder.”
no vision? 80 percent of the top20 super computers are power based, most even on the old power4 so imagine how the picture will look when power5 and later power6 gets adopted.
ibm has the most agile and powerfull cpu architecture, being able to turn it in any direction the custioners want. ranging from the supercomputers to low power mobile- no vision?
rayiner: “Putting Claims in Perspective”
comparing with conroe and merom is futile. already ibm is walking all over intel in the segment power is aimed at and no amount of conroe or merom will change that, which basicly, as you also concludes yourself, renders your comparisons and long post meaningless.
comparing with conroe and merom is futile. already ibm is walking all over intel in the segment power is aimed at and no amount of conroe or merom will change that, which basicly, as you also concludes yourself, renders your comparisons and long post meaningless.
The CPU market is getting more commoditized, not less. Opteron is making a lot of headway into markets where a proprietory RISC would have been considered earlier. It’s fine if POWER6 stays in its niche, but I think its important to point out that this niche is getting smaller, not larger. Moreover, my goal with the Conroe/Merom comparison was to address the “faster than Intel and AMD” point in the article, and the inevitable “Apple shouldn’t have switched” comments.
Funny, and all this with T1 launched, and by the time this thing comes on stream, T2 would have already shipped – 8 core, 64threads, in a smp configuration with focus on throughput rather than grand pissing competitions.
All very nice to boast about Ghz, but at the same time, if the cost vs. performance doesn’t add up, it becomes nothing more than a novel spectical in the marketplace for geeks to drool over.
Please read the ZDNet article which has more details.
http://news.zdnet.com/2100-9584_22-6035948.html
Also from the article, “IBM claims to have made major performance gains by stretching and squeezing silicon and using insulation techniques.”
It looks to me that IBM are getting desperate with their CPU line and are hoping big GHz numbers will get them some attention and hopefully some sales.
Personally I think they’ve gone the wrong way with this. They should have gone after the server market Apple were having a crack at (with the PPC Linuxes(sp)) and also started offering small discreet desktop computers (competing on price/power consumption/heat dissipation/etc).
They might also need to crank up support for the PPC OSes left if they want to keep making PPC CPUs.
How ironical it is that you mention “They should have gone after the server market” when really if you had any idea whatsoever about IBM and the computers they sell, you’d realize that the POWER processors ARE ONLY targeted at the server market. As others have mentioned, take a look at the top500 list and you’d realize this. POWER processors aren’t for desktops, they’re for 16-64 way servers with racks and racks and racks and racks of hard-drives.
It handles all your computing needs and allows you to fry and egg for breakfast! Absolute efficiency reinvented by IBM
This post is a bit off topic, and I may get ripped for this, but I wanted some input from people who are a little more in touch with the processor happenings.
It seems to me that when Intel first started the Itanium project, they must have seen something that would make them think twice about the longterm viability of the X86 platform. Regardless of what you may think about Intel, the reasons for this switch must have been significant enough to set aside the massive amount of resources for the Itanium project. However, we all know how that investment is paying off for Intel, HP, etc.
AMD was quicker to see what consumers wanted when it came to 64 bit computing, yet you don’t hear the same gradiose plans about the end of X86 coming from AMD. To this day, X86 dominates the market, and the recent introduction of the multicore X86 chips gives me the impression that the X86 architecture is far from over.
Where the shortcomings of the X86 chips that Intel found really that significant? How much more room does the X86 chips have to evolve? I understand that X86 chips are not best suited for every job out there, but I still can’t help but think that Intel must have seen something important.
As previously said, I am by no means an expert in this area, but I wonder about the future of processors (desktop in particular because will be the only ones who impact me). Thank you for your input.