IBM Releases Power7 Processor

Dan Jones writes "As discussed here last year, IBM has made good on its promise to release the Power7 processor (and servers) in the first half of 2010. The Power7 processor adds more cores and improved multithreading capabilities to boost the performance of servers requiring high up-time, according to Big Blue. Power7 chips will run between 3.0GHz and 4.14GHz and will come with four, six, or eight cores. The chips are being made using the 45-nm process technology. New Power7 servers (up to 64 cores for now) are said to deliver twice the performance of older Power6 systems, but are four times more energy efficient. Power7 servers will run AIX and Linux." And reader shmG notes Intel's release of a new Itanium server processor after two years of delays. The Power7 specs would seem to put the new Intel chip in the shade.

Comments

4.14GHz?

[Anonymous Coward]

What happened to the "3GHz ceiling"? Why can IBM go above it but Intel, AMD and VIA are stuck below it?

Re:4.14GHz?

[MichaelSmith]

Different architecture maybe? A little bit like how RISC could clock faster than CISC back in the day.

Re:4.14GHz?

[Totenglocke]

You mean how you can buy a 3.4 GHz Phenom II X4 from AMD? That 3.0 GHz ceiling?

Re:4.14GHz?

[the linux geek]

Its still that way. POWER6 could actually go up to 5GHz, but IBM sacrificed out-of-order execution to get there. POWER7 brings it back with a slightly lower clock speed and more cores.

Re:4.14GHz?

[wizardforce]

There probably were better ways of increasing computational speed using multicore processor designs than just increasing the clock speed. Kind of like going from a V4 engine to V6 being a better option in terms of power than increasing the individual piston HP of the V4 from 25 to 30.

Re:4.14GHz?

[alvinrod]

Actually both AMD and Intel have chips currently clocked over 3 GHz. Some of the newer Intel chips also have something they call Turbo Boost where the chip essentially overclocks itself if it's not using all of its cores. It also looks like AMD has a 3.6 GHz Phenom II X4 chip slated to be released soon. It would appear that the companies found solutions to whatever ceilings may have existed. VIA doesn't target the high-end of the market so I don't think that they're producing any chips that would run at those clock speeds.

Re:4.14GHz?

[RightSaidFred99]

First, there is no 3GHz ceiling, so you're begging the question. Second, these processors use specialized cooling - not run of the mill cheapo barely-enough heatsinks. If AMD or Intel spent $20 more on their heatsinks, they'd easily be selling 3.4-3.8GHz processors. But the profit margin isn't there. Third, power usage hikes as you increase voltage high enough to hit those speeds. Most people running nuclear explosion simulations on a 4GHz processor don't care, people running 30,000 machines in a design center...do care.

Re:4.14GHz?

[Hurricane78]

What stops you from doing *both*? ^^

Re:4.14GHz?

[Hurricane78]

If AMD or Intel spent $20 more on their heatsinks, they'd easily be selling 3.4-3.8GHz processors. ...
Third, power usage hikes as you increase voltage high enough to hit those speeds.

You’re contradicting yourself. The reason they can in fact not easily ramp up the CPU speed, is exactly this increase in voltage. Which increases temperature at a cubic speed relative to processor speed. (See the Pentium 4, for what that results in.)

And because your bring not a single actual argument to why you think there is no 3GHz ceiling (actually it’s a gray area above 3 GHz), I call your argument... busted! ;)

Re:4.14GHz?

[Gerzel]

Fairy dust! IBM has captured several pixies and uses them to craft its magical chips.

Re:4.14GHz?

[RightSaidFred99]

There's no contradiction. Power usage is less of an issue on higher end "enthusiast" chips. They could easily sell 3.6GHz chips in this space with better heatsinks (as evidenced by...people running them at 4GHz easily on air cooling).

In the commodity space, even with better cooling, the power usage increases disproportionately as voltage goes up. There is a sweet spot, and it isn't currently >3GHz.

Finally, I didn't point out why there is no 3GHz ceiling because it takes 30 seconds of googling to see that there are currently chips selling at > 3GHz, and there have in the past been x86 CPUs up to 3.73GHz.

Busted my ass.

Re:4.14GHz?

[Anonymous Coward]

Don't forget that one of those 4GHz CPUs probably costs over 10 times as much as an equivalent Intel or AMD part. A decent PC costs as much as a car payment -- a decent POWER machine costs as much as a car.

The price is old, but a couple years ago a 5GHz Power6 CPU cost $15k for a dual-core module (with 4 threads) plus $30k to activate each core. That means you'd pay $75k total to use both cores of the CPU module. I'm sure Intel would have no problem supplying 5GHz CPUs at $75k each, but it's unlikely that they'd have many takers, so you're stuck with CPUs that are only 3GHz (but go almost as fast as IBM's 5GHz parts).

dom

Re:4.14GHz?

[wizardforce]

Cost. Spending an extra 500$ to double the power makes sense. Spending 5,000$ to increase the power a measly 20% is rather foolish either way you look at it by comparison.

Re:4.14GHz?

[afidel]

Intel did 3.73 Ghz as the top end for Netburst (Xeon 5080) but it was a fairly poor performer on a MIPS/Watt basis. In fact the 5160 running at 3.0Ghz did about 33% higher Specfp and run at 80W instead of 130W for the 5080 (35.2 specfp_2000/watt vs 15.5).

Re:4.14GHz?

[TubeSteak]

There probably were better ways of increasing computational speed using multicore processor designs than just increasing the clock speed. Kind of like going from a V4 engine to V6 being a better option in terms of power than increasing the individual piston HP of the V4 from 25 to 30.

Back in my day, manufacturers used to slap a turbo button on the front of the case.
And we liked it that way.
Now get off my lawn!

Re:4.14GHz?

[amorsen]

Spending 5,000$ to increase the power a measly 20% is rather foolish either way you look at it by comparison.

Not if your work load doesn't scale with additional cores. Then $5000 for 20% extra speed can be worth it.

Re:4.14GHz?

[wizardforce]

The cost of producing these higher clock speeds appears to be very cost and technologically prohibitive. The operations that can't be parallelized are apparently not important enough to justify higher clock speeds.

Re:4.14GHz?

[amorsen]

Well it gives IBM a niche to fit the POWER in -- a niche which Intel would have liked to fit the Itanium in as well. Most people won't spend $12k per CPU though.

Re:4.14GHz?

[Anonymous Coward]

3.4 is still 3.

Re:4.14GHz?

[hairyfeet]

The reason why AMD and Intel don't push the high end chips is frankly there just isn't much of a market for them ATM. Since most of the games being released are for console first and PC second, they simply aren't being bottle-necked by the CPU. This is also why AMD and Nvidia are having to push multiscreen and GPGPU, because frankly a less than $100 card will play a good 80%+ of the games out there.

Second for the jobs the average Joes are doing, web surfing, music/video, maybe the occasional video conversion, even the lower end chips are well past "good enough" for them. I have been selling a lot of low end AMD dual and triple core machines lately, and all I hear from my customers is "how fast" they are, and how they never seem to slow down. With hardware acceleration on the motherboard these 2.4GHz-2.8GHz duals and triples are frankly overkill, with most of the time the CPU twiddling its thumbs. I myself bought a 925 quad when they got so cheap, but a good 90% of the time the chip is barely above idle.

So it isn't that AMD and Intel can't make them, because we have seen in the past they can, it is just there really isn't much of a market for them. To get faster than 3.2GHz you really start cranking up the heat and the power, and that equals higher electric bills most folks don't want, not to mention having fans that sound like a F15 taking off isn't very pleasant. With the new 95w like my 925 the chips rarely get above 83f idle and so far mine has maxed at 109f doing video transcoding. And the 65w duals are so quiet I have to watch when setting them up I don't turn them off when I mean to turn them on, because i simply don't hear any noise. Folks nowadays seem to care much more about that than the MHz race anymore.

Re:4.14GHz?

[Your.Master]

You learn something new every day.

Re:4.14GHz?

[porl]

hahaha i remember back in the day trying to argue with a guy at school that my 'xt' machine was faster than his '386' one because i had a turbo button and he didn't. it was even called 'hal 2 turbo' from memory, which, after having seen 2001 space odyssey is even *more* awesome than i thought back then haha

Re:4.14GHz?

[Anonymous Coward]

And that's why pi is pretty much still 3. It's a cunning result deduced by that same excellent mathematics school of thought. You know, the one that the ancient egiptians said "fuck that shit" almost 4 thousand years ago.

Re:4.14GHz?

[couchslug]

"Back in my day, manufacturers used to slap a turbo button on the front of the case.
And we liked it that way."

Noobs...
Back in MY day, I used to wax the strings on my abacus to lessen bead friction.
We LOVED it that way.
Now get off my peat bog!

Re:4.14GHz?

[DarkOx]

I am going to call BS on that statement. If all the only major barrier to increasing the clock speeds for AMD64 chips from Intel or AMD right now heat sinks I'd be buying different servers. None of the credible server manufacturers HP,IBM,DELL? use heat sinks looking anything like the ones attached to a retail packaged microprocessors. None of these guys are shy when it comes to noise and fans either.

Customers want as much power from a single chip as they can get with the virtualization targeted products. The system builders would simply say to Intel and AMD "Give us the 5ghz chips and tell our engineers how much heat they have to dissipate and a how fast they will find a way."

Re:4.14GHz?

[Machupo]

Most people running nuclear explosion simulations on a 4GHz processor don't care, people running 30,000 machines in a design center...do care.

People doing real nuclear blast simulations (and more importantly, effects of long-term nuclear decay) would be using 10k's if not 100k's of these processors and most certainly care :)

Re:4.14GHz?

[LWATCDR]

Why can't we get a -4 terrible metaphor tag?

Re:4.14GHz?

[drsmithy]

Customers want as much power from a single chip as they can get with the virtualization targeted products.

Why ? For starters, virtualisation is nearly always limited by IOPS and/or RAM, long before it is by CPU. Further, it's a textbook example of a situation where adding more cores will (typically) deliver more benefit than adding more performance per core.

When building virtualisation infrastructure, individual CPU core performance is about the last thing that matters.

Re:4.14GHz?

[drooling-dog]

Even our compilers were "turbo" back then.

Re:4.14GHz?

[dwiget001]

I have been in the software field (sales, first line and higher lines of support, programming, product management, certification) for 23 years now.

There was a time, late 1980's to about mid 1990's where I had a number of clients tell me:

If the product has the word "turbo" in it's name, management told me they will not authorize purchase for it.

Re:4.14GHz?

[drooling-dog]

The high end of just about every market is ruled by enthusiasts, and their demands often have a lot more to do with how they choose to distinguish themselves from the masses than any objective need for performance. It's the same in CPUs, cars, wine... you name it.

I love 'em, though. They help keep things cheap behind the wave where "good enough" is good enough (and better all the time).

Re:4.14GHz?

[Anonymous Coward]

Computer nerds: You mean inline-4 engine, not V4. There is no V4, boxer-4 maybe. Stop saying V4... ...and all are inferior to the gorgeous balanced inline-6 and V12!

Re:4.14GHz?

[imgod2u]

There is no 3GHz ceiling; but 3GHz is the sweet spot right now. The current i7's turbo-clock themselves to near 4GHz (3.9 in some of the Lynnfields I believe) but is limited by the power envelope of its target applications (desktops).

Frequency will not be scaling much more in CMOS beyond this as transistors are not going to be getting much faster.

Re:4.14GHz?

[muckracer]

And back in MY day, I used to hit nerds like you severely with a huge wooden club. Then we played football with your heads. We LOVED it that way! Now get off my woman!

Re:4.14GHz?

[Carewolf]

There probably were better ways of increasing computational speed using multicore processor designs than just increasing the clock speed. Kind of like going from a V4 engine to V6 being a better option in terms of power than increasing the individual piston HP of the V4 from 25 to 30.

There is no such thing as a V4. A typical 4-cylinder engines are inline engines. The 'V' in V6 and V8 is the configuration of the engine two parts not a unit of measurement. Note that at first only V8 made sense because it was just two ordinary 4-cylinder engines in V-configuration. The V is important because there are both 6-cylinder engines in V-configuration and inline-configuration (inline-6 and V6).

Re:4.14GHz?

[i.r.id10t]

Actually, the Honda Sabre has a V4 in it.

And I'd bet that the majority of 4 cylinder motors out there are boxter style (Porsche 356/912, VW Typ I and III).

Re:4.14GHz?

[hairyfeet]

While this is true, I was simply pointing out we seem to have reached a plateau when it comes to CPU/GPU design where the work folks have to do for them simply doesn't keep up with the incredible power they possess. Look back just a decade ago-I personally went from 400MHz > 733MHz > 1.1GHz > 2.5GHz > 3.06GHz > 3.6GHz. Those are some pretty big leaps with the performance to match. You could really feel when you made those jumps, by the increase in performance and speed.

Now let us compare to the end of the decade. As a lifelong Intel + Nvidia man who while not on the cutting edge was rarely more than a gen or two behind, I went AMD. Why? Because after building some low end AMD duals for customers I found their machines to be VERY snappy, and didn't sound like jet engines taking off. Then I went from a 7550 Kuma Dual to a 925 Phenom II Quad. Could I tell the difference, like I did with those earlier revs? Maybe, when I am using a multi-threaded transcoder, but otherwise? Not so much. I simply don't have enough heavy duty lifting to stress the CPU/GPU when compared to the single core MHz race, and the amount of power even in the low end AMD duals and triples is frankly extreme overkill. My cheap but good ECS board had an onboard GPU that not only hardware accelerated all the major codecs out of the box, but I played games like Bioshock and FEAR on it for over a month until I got around to getting a discrete.

So I was simply pointing out that with today's super powerful machines, there really isn't a point in pushing the MHZ envelope. I have a few enthusiast friends, and while some have built Core I machines, they went with the cheaper I5 over I7 and the machines they replaced weren't bottle-necked in any way either. Two of them after playing with mine decided to go AMD quad, simply because the BE quads were so cheap and with a good GPU there really wasn't any job they could come up with that would make the increased expense worth it. While AMD and Intel have reached truly ludicrous speeds in amount of work per cycle, we users simply haven't kept up. When my GF's PC fried its mobo instead of building her a new as I would have from 2000-2005, I simply dug out an LGA775 spare board I had and fixed it. Why? Because for the work she has (surfing, webmail, Facebook games) her P4 3GHz is frankly overkill, and with speedstep enabled it is cool and quiet with as little as she is hitting the CPU.

PCs have just gone beyond "good enough" for her and millions like her, which is why AMD and Intel really don't push the MHz anymore.

Re:4.14GHz?

[jon3k]

^ mod this up, that's almost universally true. Every VMWare installation I've seen is limited by RAM or I/O long before CPU. Why do you think Cisco found a way to get 384GB of RAM per socket in their new UCS line?

Re:4.14GHz?

[speederaser]

V-4 engines certainly do exist.

For cars, both Ford and Saab have built V-4s. V-4s are more common for motorcycles though because an inline engine makes the engine too wide (motorcycle engines sit sideways in the chassis). Honda, Yamaha, Suzuki, Ducati and Aprilia all have V-4s in their motorcycles.

http://en.wikipedia.org/wiki/V4_engine

There have also been a number of V-2 engines. The only V engine I have not heard of below 16 are V-14s.

Re:4.14GHz?

[bill_mcgonigle]

You mean how you can buy a 3.4 GHz Phenom II X4 from AMD? That 3.0 GHz ceiling?

If you constrain yourself to PPC, there was a famous pledge for a 3GHz PPC 970 but that materialized about three years behind schedule in the Power 6.

Actually, that's worth noting - the Power7's aren't trying to out-clock the Power6's.

Re:4.14GHz?

[BikeHelmet]

VIA's C8 (Nano) processors were rumoured to have included a design tradeoff for the FPU that capped them at around 2ghz.

But they're remarkably efficient, and perform way better than Atoms. They aren't comparable to desktop chips, but it seems like they made a decent tradeoff, if true.

Re:4.14GHz?

[BikeHelmet]

Absolutely correct on so many points!

Most games out now will play beautifully on an Athlon II X2 with a 9800GT. That'll set you back perhaps $150. Toss in a motherboard, RAM, HDD, PSU, case, and you have a cheap gaming PC that gives lots of enjoyment. Or you can go with a quad for barely $40 more.

I too have gotten plenty of such comments. However, it's worth noting that new HDDs are helping a lot as well. I have done system upgrades to Athlon X2's in the past, and until this latest batch of HDDs, they didn't feel quite as snappy. I suppose I've fallen in love with WD blacks.

I suspect ARM will be trying to work their way into this market. In a few short years they'll be caught up with the "fast enough" CPUs that 95% of people can settle for. Audibility and power consumption are starting to become more important, and that's one area where ARM already excels.

I ran my Athlon II X2 245 rig through a Kill-A-Watt. The dang thing is overclocked to 3.5ghz, but still only consumes 95 watts at idle, and 130 watts when video encoding or gaming. That's quite a bit less than my old Athlon XP system!

Re:4.14GHz?

[walshy007]

V16's are completely balanced engines, unfortunate there are so few of them. V12's are very smooth but far from perfectly balanced.

Re:4.14GHz?

[QuantumRiff]

I have a 3.2 GHz Pentium 4 Extreme Edition.. One of the last models before moving to the Core Architecture. It was an "engineering sample" and is one hell of a space heater, but also many years old.

Re:4.14GHz?

[hairyfeet]

That is why I love my 925, it only uses 95w, is whisper quiet, and takes any job I throw at it with nary a complaint. I mated it to an ultra cheap 4650 1Gb GPU and so far every game I've thrown at it has been nothing but high framerates and pretty picture, and you can't beat $700 for a PC with 8Gb of RAM, a Tb of HDD, dual burners and a quad. if your board can take one I highly recommend the 925 to future proof your machine. That 8Mb of total cache makes for a damned good video transcoder. Of course if you don't transcode you can get a Propus 2.6GHz for a cheap $99, and after building a couple for customers I can say for most jobs they are truly scary fast and quiet.

As for "speed and snappiness" even on the smaller AMD chips, have you thought about showing your customers one of these? With the free Junction Magic you can set the My Docs and program files to be on a larger HDD and leave the SSD for the OS. These puppies will max out a Sata II connection, and by moving the two most overfilled folders to a large cheap HDD you can get nice long life from the SSD and still enjoy the crazy OS responsiveness you get from SSD without having to break your wallet.

After setting a couple of these up for customers and seeing the truly insane speeds they were getting on even the lowest X2s and X3s I'm probably gonna pick up one for myself. I already would have but Valentine's day is right around the corner and that is one of the "Thou shall not" along with thou shall not forget her BDay or Xmas, and Thous shall not answer honestly if she asks if she looks fat in that dress. But I doubt ARM will get more than the "browser in a box" tiny niche because of the No Windows problem. Everyone just seems to love the new Windows 7 and folks that buy Win 7 for their new machine end up coming back and having me upgrade their second box to Win7 as well. Not having Windows will leave ARM to the cheap netbook niche IMHO.

Re:4.14GHz?

[BikeHelmet]

That is why I love my 925 [newegg.com], it only uses 95w, is whisper quiet, and takes any job I throw at it with nary a complaint.

It probably consumes far less most of the time. Run your PC through a Kill-A-Watt, and prepare to be amazed.

Phenom II's are more power hungry than Athlon II's, but I'd still bet on your PC being around or under 150 watts.

As for "speed and snappiness" even on the smaller AMD chips, have you thought about showing your customers one of these [newegg.com]? With the free Junction Magic [rekenwonder.com] you can set the My Docs and program files to be on a larger HDD and leave the SSD for the OS. These puppies will max out a Sata II connection, and by moving the two most overfilled folders to a large cheap HDD you can get nice long life from the SSD and still enjoy the crazy OS responsiveness you get from SSD without having to break your wallet.

I have tested that particular SSD. It was not significantly better than a second 64MB cache HDD. Don't get me wrong - it was better - but the space you lose for a bit of extra performance wasn't worth it.

I'm Canadian - the cheapest SSD I saw was an Indilinx 30GB drive for $85 after MIR. SSDs are approaching the sweet spot for prices, but aren't quite there yet.

Re:4.14GHz?

[hairyfeet]

Oh I don't doubt it is VERY conservative on power. The new Cool & Quiet really drops the CPU when not in use, and even with desktop being offloaded to the GPU on Win 7 the ATI 4650 barely turns the fans. And for when I am just surfing the net I have a Sempron 1.8Ghz a customer sold me for $50 (He got one of those "Best Buy Vista Specials" and of course it sucked) and with XP Home and cool and quiet you can't even tell it is on most of the time. Gotta love those new cheap AMDs! When I told a customer today how much a fully loaded dual would set him back he was shocked they were so cheap.

As for the SSD? That is why I suggested Junction Magic, as it makes it feel like you have a huge SSD even when you don't. You see I set up My Docs and Programs to D:, but thanks to Junction Magic all programs install and act like it is on C:. so everything is seamless. you just act like you normally do and the SSD stays for the OS only, while the bloated music and programs folders invisibly end up on D:. But I added an 8Gb flash using Readyboost to my dual WDs and frankly it is already faster than me. With 8Gb of RAM the new Windows 7 caching really learns my habits and has my programs ready to go in RAM, so the speed is just incredible.

I used to swear by Intel + Nvidia but after Bumpgate and the frankly crazy prices for the new Intel CPUs (not to mention mobos. Have you seen the prices? Ouch!) I can build my customers truly scary duals and triples for about half what it would cost for Intel + a decent GPU. The new Radeon mobos are just awesome with hardware acceleration for everything. When I started out we measured RAM in bytes, so the fact that I can get this much power for under $700 just blows my mind. I can easily see this new quad lasting me for a decade, it is just that fast and quiet. Sorry about the Canada prices though. Considering it is ass deep snow here in AR I bet y'all are miserable there, huh?

Re:4.14GHz?

[BikeHelmet]

Considering it is ass deep snow here in AR I bet y'all are miserable there, huh?

25C, and pissing rain. Just great for the Olympics.

Heh... I don't even have Cool & Quiet enabled, and it's using that little power. ;)

I hear you when it comes to cheap AMD systems. They pack a lot of punch for their price.

Re:4.14GHz?

[hairyfeet]

That is like 77f, yes? How in the hell are y'all that warm and we here in normally balmy AR are at 27f and dealing with snow? That is all kinds of fucked up.

As for C&Q? You really should enable it and watch your power usage drop off the map. The new C&Q is REALLY fast and efficient, so fast that even on this 1.8GHz Sempron I don't ever notice a slowdown or feel the switch. It is so quiet I have the Sempron less than 4 feet from my bed and never hear it, and when surfing on my quad it will drop it down to as low as 600MHz from 2.8GHz and thanks to the distributed workload I never would know it if I didn't fire up CPU monitoring just to check it out once in awhile.

And I totally agree on the AMD love. As a former exclusive Intel + Nvidia man I simply couldn't ignore the incredible prices that allow me to build scary powerful machines for my customers for such little cash. With OEM duals running at the $50 mark or even less I can build some truly scary Windows 7 HP duals for less than $450 and still make a nice profit for myself. And the new Radeon onboard GPUs not only accelerate just about every codec anyone uses, but I was even playing Bioshock and SWAT 4 on mine for a couple of months until I got around to buying a discrete. And the hardware acceleration makes even the most HD video smooth as butter with the lowest AMD duals, and cuts power usage as well. You just can't beat that!

Direct comparisons are bad

[the linux geek]

POWER and Itanium are architecturally so different that kdawson's snide "put this new Intel chip in the shade" comment is kind of nonsensical. Itanium is superscalar to an extent that POWER doesn't come close to, with each core being able to execute up to six instructions per cycle. While its possible that POWER7 is faster, its also more expensive to get a reasonable configuration and the performance difference between the two is not as clear-cut as our illustrious editor is trying to suggest.

Re:Direct comparisons are bad

[Anonymous Coward]

The Itanium is more that just superscalar, it is explicit parallelism. You can accomplish the same feat with superscalar and out-of-order execution but it takes far more silicon and it tends to have some odd corner cases.

POWER and Itanium are both pretty slick architectures but Itanium is definitely a generation later in design. If only Intel were willing to bet the company on it, about 10 years ago, we would all be using it today.

Re:Direct comparisons are bad

[ZiakII]

I'm not sure what your trying to say here could you use a car analogy?

Re:Direct comparisons are bad

[Wesley Felter]

POWER and Itanium are architecturally so different...

That doesn't matter; they both address the same market (high-end Unix) and thus they are competitors.

Itanium is superscalar to an extent that POWER doesn't come close to, with each core being able to execute up to six instructions per cycle.

Yeah, POWER7 can only execute... six instructions per cycle. And you might indeed say that an in-order Itanium at 1.7 GHz doesn't come close to an out-of-order POWER7 at 3+ GHz.

While its possible that POWER7 is faster, its also more expensive to get a reasonable configuration...

Since no Tukwila servers have been announced, we don't even know how much they will cost.

Re:Direct comparisons are bad

[WilliamBaughman]

You're totally correct: a direct comparison between POWER and Itanium is difficult because the architectures have different purposes. That said, do you know of any benchmarks where the new (or old) Itanium chips clobber POWER6 (or 7) processors? I'm not trying to be snarky, I'm really curious.

Re:Direct comparisons are bad

[Anonymous Coward]

Until they are superscalar and vectorial they can only go to their bed and cry.... damned programmers that failed linear algebra!

Re:Direct comparisons are bad

[the linux geek]

http://www.tpc.org/tpcc/results/tpcc_perf_results.asp - I wouldn't say "clobber," but they're roughly at par on performance and Itanium has an edge on price/performance.

Re:Direct comparisons are bad

[GenP]

Itanium: vrooom!, vrooom! POWER: putt, putt, putt

Re:Direct comparisons are bad

[teg]

I'm not sure what your trying to say here could you use a car analogy?

Ferrari vs volvo Truck? Straight out speed, vs. load capacity.

Re:Direct comparisons are bad

[Bert64]

Had Intel bet the company on it, AMD would likely be the dominant processor vendor today...

It doesn't matter how good the IA64 architecture is, customers want to run their existing applications, most of which are compiled for x86 and don't come with source code. That leaves you with emulation, which i doubt Intel could make faster than native... Intel can't move to a new architecture because they are held back by all the millions of closed source applications out there.

Re:Direct comparisons are bad

[MrNemesis]

Since no Tukwila servers have been announced, we don't even know how much they will cost.

As a sysadmin for a company with POWER5 and 6 equipment, all I can say is if you have to ask you can't afford it. Part of the reason why jumping ship to RHEL + Oracle running on a VMware cluster is looking increasingly appealing to managment.

Re:Direct comparisons are bad

[xZgf6xHx2uhoAj9D]

That leaves you with emulation, which i doubt Intel could make faster than native...

If only you could go back in time and convince Intel of this! The first generation of Itaniums actually did x86 emulation in hardware. A brilliant idea: the only problem with it was that it was actually slower than software emulators (which themselves were pretty slow).

Anyway I don't think Itanium was every supposed to replace x86. This was before x86-64 existed and Intel thought it would be their only 64-bit chip.

Intel can't move to a new architecture because they are held back by all the millions of closed source applications out there.

Ahh but they did! It's called x86-64!

In the end it wasn't backwards compatibility that was the problem. x86-64 has the almost all the same backwards compatibility problems that Itanium has: software developers are forced to release two binaries for their code these days, an x86-32 binary and an x86-64 binary. Obviously x86-64 took off like gang-busters, though, even though it came out like 5 years after Itanium. The reason developers are happy to develop x86-64 binaries but not Itanium binaries is that there still doesn't exist a decent compiler for Itanium.

Okay, okay, I know x86-64 trivially does x86-32 "emulation" very efficiently, which helped it out, but I think the existence of extremely good compilers is what helped it more.

Re:Direct comparisons are bad

[imgod2u]

That whole philosophy went out the window when Intel couldn't make a compiler good enough to make Itanium work well in all situation; which to this day -- despite having more software engineers than silicon guys -- they still don't.

Scheduling things beforehand will only get you so much. It sounds good on paper in a "look how much silicon we save" kind of way but the reality is, explicit parallelism and static scheduling simply aren't good in this day of variable memory latencies, multi-tiered caches and people no longer programming in assembly.

Re:Direct comparisons are bad

[pslam]

The Itanium is more that just superscalar, it is explicit parallelism.

Except it isn't, really, and that's what precisely the problem with Itanium. It was intended to be Explicitly Parallel Instruction Computing (EPIC), but then Intel remembered they were Intel and instead designed Extremely Complicated Parallel Computing (EPIC FAIL). Yes, the instruction bundles do explicitly indicate dependencies, but they still have to worm their way through a huge reorder and retirement buffer, and just to top things off there's STILL register renaming. These are supposed to be things that EPIC allows you to remove from a design!

It's a shame, because it could have been a very straightforward design, with a shorter pipeline, and most importantly smaller and cheaper. Or just tons of cores. I think Intel just had POWER envy and wanted to be reclaim the crown of King of Fucking Complicated back from IBM, which is funny because at least IBM ditched the complexity when they saw it wasn't working so well (POWER6).

Re:Direct comparisons are bad

[John Whitley]

If only Intel were willing to bet the company on it, about 10 years ago, we would all be using it today.

IMO, Intel borked the IA-64 nee EPIC architectural effort in the design stage. It's informative to compare the IA-64 architectural approach to that of the TI C67xx series of VLIW DSPs chips. There's certainly some apples and oranges in there, but the TI VLIW chips showed the possibility of a parallel ISA that is simple, effective, and targetable (i.e. both humans and compilers could generate good code) -- and at a similar point in time to Intel's own efforts. Cell would be another interesting point of comparison with the above two architectures.

At the time, IA-64's design implied "we're Intel, we can make things as complicated as we want!"... and this hubris persisted without regard to implementation concerns such as design time due to said complexity (which effectively killed Itanium) and the enormous die-area suck for the massive hardware controller. It's also unclear to me whether proper concern was given to good compiler-chip synergies in this architecture, which are generally key to good performance. Frankly, I think we're *much* better off with the current multi-core path than we ever would have been with the conceptually bloated mess that was the Itanic.

Re:Direct comparisons are bad

[w32jon]

whoops, accidentally modded you down, posting to remove mod

Real question

[LordoftheChmod]

That's all fun but it doesn't answer the real question : Can it run Crysis?

Re:Real question

[skine]

I think you clicked the wrong link on Tom's Hardware.

The question here is whether it can run Linux - followed shortly by a debate on how terrible Ubuntu is.

Re:Real question

[ducomputergeek]

I'm not sure, but their pre-lease benchmarks on Duke Nuke'em forever was off the chart. No. Really. True story.

Re:Real question

[FooAtWFU]

I think the big-iron IBM philosophy response to that question is "Eh, throw that shit on an expansion card and we'll virtualize it fir ya."

Re:Real question

[amorsen]

The question here is whether it can run Linux - followed shortly by a debate on how terrible Ubuntu is.

Yes, it can. And approximately <------> this horrible.

Re:This is Bad News

[RightSaidFred99]

Right, because the Power series is such a global moneymaking powerhouse, right?

These chips (and Itanium) are niche. In fact, the commodity chips are getting so powerful across the entire CPU segment from embedded to HPC computing that they will start eating into the market of even these niche chips. Why buy a Power7 when you can buy 3-4 Nehalems, be twice as fast, and spend 1/2 the money? There are exceptions, but they're becoming fewer and fewer.

Ah, AIX

[wandazulu]

AIX....the last Unix you can't just "get" a copy of, but need to actually buy the hardware (a la the Mac). We had a Power box at work with AIX for awhile, but its configuration tools was quite ... unique among Unix flavors (though I was told it was pretty straightforward IBM) and I had a horrible time getting GCC to work with it; most every F/OSS package I came across either straight up wasn't tested on AIX (because no one had the hardware), or it had a whole separate setup (I believe one of the standard lines running ./configure is "Is this an AIX system?").

I recall the box being wicked fast when we were running Oracle on it; it was a "small" Power machine but it still could handle a monster database with hundreds of millions of rows with no trouble. Frankly, I was sort-of sad to see it go; I really did want to get more familiar with it, but apparently the maintenance costs IBM was charging made it a non-starter. Plus, ultimately, it seems that it just wasn't very OSS friendly; xlc is apparently an amazing compiler for the PowerPC, but they wanted $6000 for a license per developer. Plus, and I'm sorry if this is nitpicking, but to have the C compiler called xlc and the C++ compiler called xlC was just, well, insane.

What I really wanted to do was get Linux on it, and Oracle even has a Linux-on-Power version of their database, but there seemed to be some grumbling from the IBM salespeople (according to my boss) that they discourage people from running Linux on Power....I guess you (according to them) need AIX to unleash the real "power" in the PowerPC.

Sigh, okay, whatever. back to Linux on x86-64.

Re:Ah, AIX

[raftpeople]

Minor note: PowerPC was the line of processors used in the Apple computers. POWER (as in Power7) is the server line of processors with it's roots in the as/400 servers back in the 90's. IBM didn't do a very good job of making that distinction clear.

Re:Ah, AIX

[argent]

AIX....the last Unix you can't just "get" a copy of, but need to actually buy the hardware (a la the Mac).

Don't forget HPUX.

Re:Ah, AIX

[the linux geek]

And UNICOS, Super-UX, and a few other niche systems.

Re:Ah, AIX

[Junior J. Junior III]

And Mac OS X.

Re:Ah, AIX

[oldhack]

"[AIX] unique among Unix flavors"

Hah. From my foggy memory, I am thinking VMS seemed more similar to typical UNIX than AIX. :-)

Re:Ah, AIX

[afidel]

With Darwin you can get the Unix part of OSX free for download.

Re:Ah, AIX

[FooAtWFU]

You can walk into a store and buy something with Mac OS X for $500, though. If you want to get something running HPUX or AIX you're probably talking to a sales rep over the phone and getting a quote somewhere in the low five figures (or worse).

Re:Ah, AIX

[FreakyGreenLeaky]

Fond memories... I remember having a similar experience wrt SUN's C compiler - the licensing was just stupid. I needed an extra lic for another machine, but management balked at the price. So,... gcc it was. Once I got it going (this was circa '96) gcc was a pleasure - the binaries were smaller, they ran faster (by several percentage points IIRC) and gcc compiled faster.

I also had a few years of exposure to several power boxen running aix - quite zippy, but the UNIX flavour was just weird.

Re:Ah, AIX

[red crab]

They made the beta version of AIX 6.1 available on their site some couple of years back. IBM is just paranoid about AIX/Power; they want it to remain in a sort of exclusive aka privileged domain.

Re:Ah, AIX

[Guy Harris]

Minor note: PowerPC was the line of processors used in the Apple computers. POWER (as in Power7) is the server line of processors with it's roots in the as/400 servers back in the 90's.

The first POWER processor was in the first RS/6000 machines, back when the AS/400's were still IMPI machines. In any case, POWER/PowerPC/Power Architecture/etc. are all close cousins....

Re:Ah, AIX

[argent]

And Mac OS X.

http://opensource.apple.com/

Re:Ah, AIX

[SEE]

Hmm? You certainly can "just 'get' a copy of" OS X. Amazon keeps offering me it for $24.99, with free shipping on an order over $25. Nobody checks to see if you own a Mac before they ship it to you.

Re:Ah, AIX

[raftpeople]

Yes, on review correct. I was thinking of when they created the 64 bit server processors.

Re:Ah, AIX

[Ed Avis]

Plus, and I'm sorry if this is nitpicking, but to have the C compiler called xlc and the C++ compiler called xlC was just, well, insane.

That by itself makes me want to get my hands on an AIX machine.

I used to have a pimped-out PS/2 Model 80 (56 megs RAM, mostly on an MCA card, twin SCSI controllers each with its own two megabyte cache on 30-pin SIMMs, SGI IrisVision hardware 3D acceleration and all sorts of other weird shit) that I could have run it on (instead of Linux)... sadly once I stopped being a student I no longer had the time to fiddle with it, so it went to the tip :-(.

Re:Ah, AIX

[Rexdude]

I work for IBM's Java Technology Center, and we develop IBM JDK for 12 combinations of platform/architecture - 32/64bit each of Windows,Linux* and AIX, z/OS(31/64).....and Linux on p-series(32/64) as well as Linux on System Z (31/64).

* - 64bit Windows/Linux refers to AMD64 not IA64.

I administer a POWER5 dual CPU box used for development that runs RHEL 4. IBM does provide Linux on its own hardware for compatibility/ease of use etc.
You can directly download IBM Linux JDKs here, but for Windows and 32bit Linux you have to get it bundled with the Eclipse developer kit, as those platforms are the same as Sun, and licensing forbids us from directly offering it for download on those platforms other than as part of another product.

Re:Ah, AIX

[argent]

Hah. From my foggy memory, I am thinking VMS seemed more similar to typical UNIX than AIX. :-)

"SMIT happens."

Re:Ah, AIX

[armanox]

Solaris has been freely availible since 2006

Apple skunkworks?

[bobdotorg]

I'm curious whether or not Apple is maintaining a parallel dev. of OSX for this line of IBM chips the same way that the Intel version of OSX was lurking in the dark from 2000 until 2006.

Re:Apple skunkworks?

[RightSaidFred99]

Lol. No, they're not.

Re:Apple skunkworks?

[hitmark]

i find it more likely that they are looking into running osx on a multi-core cortex-A9 or later, with some special sauce from PA semi added on top.

Re:Apple skunkworks?

[Wesley Felter]

You're confusing POWER with PowerPC. They are not the same thing.

Since POWER3 in 1998 they are the same thing, actually. http://en.wikipedia.org/wiki/Power_Architecture#History But don't let me stop you from showing off.

Re:Apple skunkworks?

[Khyber]

LOL You're in no position to know.

I am, however. But my NDA forbids me from saying anything.

Re:Apple skunkworks?

[bonch]

If you'd read the article you're linking to, you'd see that PowerPC was "a modified version of the POWER architecture" and that POWER3 "was only used in IBM's RS/6000 servers."

Re:Apple skunkworks?

[Khyber]

Actually, I just checked said NDA, as it has been at least five years since I worked for them, and my NDA is over, so....

Support has NEVER been fully dropped and never will unless IBM becomes non-viable in the marketplace. On top of that, long-term contracts Apple has with some companies pretty much ensures that they keep some minimal amount of POWER support active, at least into the next decade.

Oh, I'm sorry, did I break your bullshit reality bubble? Get a real job in the industry and maybe you'd have half a clue.

Re:Apple skunkworks?

[H0p313ss]

They will most certainly never return to Power.

And 640k should be enough for everyone. The world crazy place... who'd have imagined the impact that such diverse things as the PS3, ARM processors and Android would have had on the industry. If the mobile world goes off x86 (the death of the laptop) would power pcs be so hard to imagine?

Re:Apple skunkworks?

[Bert64]

But is your information now 5 years old?
5 years ago, PPC machines were still being sold...

Re:Apple skunkworks?

[LWATCDR]

Makes sense. Apple Supports the Kernel on ARM and Intel publicly and was maintaining support for the Power line in Leopard so some development must have been on going.
I think Apple is too smart to get so lazy as to not spend at least a little effort in keeping OS X portable. And if OS X is going to be Portable the most likely targets are ARM, Power, and Intel. Frankly I wouldn't be shocked if Apple also had an Itantium version in house just in case that system ever makes it into the mainstream.
I don't think Intel has given up on the Itantium yet. Let's face it some day we will have to move off the X86 ISA so keeping you options open is just smart.

Re:Apple skunkworks?

[Anonymous Coward]

Apple is married to X86

Apple is married to x86-64, not IA-32. It is unlikely that the successor to Mac OS X 10.6 will support IA-32, and it has been several years since Apple has produced hardware that is IA-32 only.

Apple is also married to ARM.

As a result of at least this bigamy, Mac OS X will be kept portable across instruction sets, byte sexes, and other architectural aspects.

With portability as a current goal, porting to another architecture should be straightfoward; it ought to be even more straightforward to port to an instruction set and architecture that is very similar to PPC.

Indeed, SSE4 has obviously been incorporating ideas from Apple Velocity Engine, making the vector part of Intel's x86-64 processor line look a lot more like those in the Apple-Intel-Motorola collaboration on PPC (~ AVE, AltiVec, VMX). The A4 chip likewise.

Finally, Apple has substantial in-house CPU design talent. The company's key problems historically have been competitively cost-effective moderate-scale manufacture of custom CPUs in the most modern processes, and the risks associated with using the most modern processes in the first place. (They still have the latter problem in their "Pro" systems, although switching to Intel has largely fixed the first.).

Power 7?

[bigplrbear]

I'm still waiting for my G5 powerbook

Re:Power 7?

[oldhack]

Due to Apple Awesomeness (TM), Gx is equivalent to P(x+2), so just wait a bit.

Commercial sales?

[CAIMLAS]

I have to wonder why IBM is (at least, as of now) limiting these processors in their own hardware.

I can understand the initial economic advantage: they'd gain more profit from server sales, and would be able to sell Linux servers at a fairly non-trivial mark-up (on base hardware cost, to them).

But what is gained there is probably trivial compared to commercial marketing of the chips/boards (OEM sales). I suspect it might also avoid scrutiny from antitrust lawyers more easily. Why wouldn't they do this? I'd certainly love a processor like that; it'd be incredible. 1/4th the power envelope of the Power6, and twice the performance (assuming it means core clock)? That's incredible: the 3.2GHz Power6 is rated at under 100W TDP.

Such a processor might just sway Apple to go back to the Power architecture, I'd think. Linux will run on them, obviously; the only thing you couldn't run on them is Windows (and even that might be possible down the road with only a little work on MS's part).

The only two reasons I can imagine are 'exclusivity' and 'insufficient fab capacity'. That second one would certainly do it on its own.

Re:Commercial sales?

[the linux geek]

IBM is a low-volume high-margin company. They don't see a lot of gain from doing the R&D so they can make these chips at a reasonable price and then taking the much smaller profit margins from them when they can make billions in profit from selling their own hardware. In addition, they want to be the only viable upgrade pathway for their existing customers - a while ago a company (Platform Solutions) tried selling Itanium mainframes with an emulation layer for running Z/OS and they got sued into the ground by Big Blue.

Re:Commercial sales?

[Wesley Felter]

A non-IBM POWER7 system would end up looking pretty much like an IBM POWER7 system, and you can bet it wouldn't be cheaper, so what's the point? If you want POWER7, buy it from IBM.

Re:Commercial sales?

[Anonymous Coward]

I don't know what a 3.2GHz Power6 costs, but last I checked a 4.2GHz Power6 cost $12k! Somehow I don't think Apple will be swayed to using them unless IBM can sell them at 1% of their current price.

dom

Re:Commercial sales?

[nxtw]

I suspect it might also avoid scrutiny from antitrust lawyers more easily.

Antitrust laws don't apply to big companies just because they're big companies...

I'd certainly love a processor like that; it'd be incredible.

Significantly more incredible one or two quad-core AMD or Intel x86-64 CPUs?

Such a processor might just sway Apple to go back to the Power architecture

Why? Does IBM have anything that compares in power usage and performance to a mobile x86-64 CPU, such as the 2.53 GHz Core 2 Duo CPU in my laptop with a 25W TDP along with a low-power chipset and GPU, such as the nVidia 9400M? Apple doesn't sell any systems that benefit from the huge bandwidth that makes POWER7 so great for systems with many sockets; at most, Apple systems are dual-socket Nehalem (XServe and Mac Pro). I doubt that Apple will even make any systems using the Nehalem-EX.

Is there an untapped market for running OS X on systems with 32 sockets?

Re:Commercial sales?

[RAMMS EIN]

Why do you suppose being able to buy similar systems from multiple vendors wouldn't drive down the price?

Re:Commercial sales?

[amorsen]

1/4th the power envelope of the Power6

I don't read it that way. I read it as 1/4th the power/performance. I bet they'll be 100W+ like the Power 6, just 4 times as powerful.

Re:Commercial sales?

[Bert64]

The ability to run windows (either under a vm or natively) helps sell quite a few apple machines these days, it gives people a fallback if they don't like osx or an option if they have a few windows apps they require.
Also, Apple sell a lot of laptops, a POWER7 would not really be suitable.

Re:Commercial sales?

[NighthawkFoo]

Actually, PSI was acquired by IBM

Re:Commercial sales?

[rtaylor]

You don't spend $10,000 per CPU then put in a small amount of crappy ram and a single tiny SATA harddisk.

Re:Commercial sales?

[Rexdude]

Power series is hardly aimed at the consumer desktop market anyway.

Uh, did you look at your link?

[raftpeople]

Power procs=32, cores=64
Itanium procs=64, cores=128

So double the Itaniums almost gets you to where Power is.

Re:Uh, did you look at your link?

[RightSaidFred99]

What's your point? All that matters is price/performance. If processor A can get the job done more cheaply with twice the processors, why would I care?

Re:Uh, did you look at your link?

[raftpeople]

Because this statement was made: "but they're roughly at par on performance", which isn't correct.

Your point is valid, but that isn't what I was responding to.

Re:Uh, did you look at your link?

[Khyber]

Because there's more shit on the die to fail and go bad, quickly rendering your cheap solution useless and slow.

Also, it needing more processors to get the same job done means it's likely a weak POS to start.

Re:Uh, did you look at your link?

[Anonymous Coward]

Point taken, though one could argue that for any sane, meaningful comparison there is always an elliptical "for a given price". If CPUTech sells a $10,000 chip that's only twice as fast as MicroTech's $500 chip, is there even any point in discussing how CPUTech's chip is "faster"? Who would buy it?

Re:Uh, did you look at your link?

[afidel]

Look at $/transaction, they are almost the same, and that's what matters.

Re:Uh, did you look at your link?

[Khyber]

I do indeed. I work with semiconductor fab quite often. It's part of my business selling horticultural LED lighting.

And LEDs are prone to the same failures as any other semiconductor-based technology.

Go suck it. :P

Re:Uh, did you look at your link?

[Tycho]

I'd also add that depending on the task, the cheap solution would be slower if the task had serial parts that could not be separated into threads. For instance if a task takes 1,000 cycles and all of the instructions must be done in a precise order, a quad-core processor running at 2.0 GHz would be slower and be of lower utility than a single core 4.0 GHz processor, assuming all other things are equal. The quad-core ends up working at half the speed of the single core and the quad-core also has the penalty of three idle cores draining electricity.

I would also imagine that these newer POWER7 processors carry over the decimal floating point units present in the POWER6. Yes, floating point units that operate in base-10 as opposed to base-2. Not necessarily of much value for scientific purposes, but great for preserving accuracy in financial calculations. One gets to avoid the base-10 to base-2 conversion and the conversion back that can severely hurt accuracy with only a binary floating point unit. One also gets a nice speed up by doing decimal math in hardware as opposed to the other option of software decimal math.

Re:Uh, did you look at your link?

[amorsen]

If processor A can get the job done more cheaply with twice the processors, why would I care?

Because most actual work loads scale worse than TPC-C. For most actual loads you're better off with half as many cores which are twice as fast. Also, the Itanium is a horrible power hog, so you'd likely lose out on your electricity bill too. If your workload scales nicely, go for Xeon or Opteron, nothing in the Itanium or Power space can beat those for performance/price.

Re:Uh, did you look at your link?

[Calinous]

You pay some licensing costs PER PROCESSOR (or PER CORE, or a combination thereof). As such, it might be cheaper to buy a $100k, 32 processors server than a similar performance, $50k, 64 processors server.

Re:Uh, did you look at your link?

[Bert64]

Because performance per watt is also important, if you have twice as many physical processors and twice as many cores you will also need all the supporting infrastructure (ie sockets for those processors to go in) etc... Not to mention the extra space required.

And more power consumption also equals more cost.

Otherwise, why use powerful machines at all, why not a cluster of the cheapest machines you can find?

Incidentally, something which performs half as well needs to be considerably cheaper or it just won't sell at all.

Re:Uh, did you look at your link?

[Nerdfest]

With IBM, you even have to pay for MIPS in some cases, on top of the cost of hardware.

Re:Uh, did you look at your link?

[DarkOx]

Price / Performance is not all that matters. If it was we would probably be running tens of di-shrunk 486 cores in PCs today. Complexity matters as well. Nobody wants to deal with hardware nonsense in software. In the PC world you can't even get people to write threaded apps; and you're going to tell me its ok to ask developers to deal with 2x as many cores to get the same amount of computing done? That is going to complicate you application a great deal!

Re:Uh, did you look at your link?

[Bonzoli]

Because an Oracle LISC per Core costs you 18k to 40K x 64 more cores.
Anyone care to do the math? Wait till Oracle only runs on Sun for 254 Cores.
Uncle Larry needs a new Island!

Re:Uh, did you look at your link?

[TheNinjaroach]

I do indeed. I work with semiconductor fab quite often. It's part of my business selling horticultural LED lighting.

Interesting. I've heard lots of hype about LED for horticultural uses, but I'm not sure I believe it. I will be checking out your LiveJournal, thanks.

Re:Uh, did you look at your link?

[Sique]

Someone who needs twice the performance, where double the space costs more than the money saved on MicroTech's offering, and where the processor is only a small fraction of the overall cost.

Re:Uh, did you look at your link?

[meosborne]

You've obviously never paid a software license that was based on # cpus, otherwise you might actually care very much.

Re:Uh, did you look at your link?

[Khyber]

There is a shit ton of hype - and it's all from fools that never took any real horticultural science classes. They're all about marketing, not about facts. that's why you still keep seeing 90w=400w instead of 400w LED kills 400w HPS. They're trying to market nonsense as a selling point, and it's working and simultaneously killing the damned industry before it even gets off the ground.

Everyone quotes 7:1:1 ratio from NASA as being the best - NASA learned LONG AGO that was just for trying to act as a supplement replacement to HPS, not sunlight. If you want a GROWING PANEL, you need to totally overhaul the train of thought and mimic the sun as much as possible. These people are totally clueless about the relative power output levels of the sun during different seasons. They've never measured it themselves. They rely upon what they've gleaned from other slightly less-clueless people, and then try to make a quick buck.

I replaced 216w t5HO with 20W LED. I'm still getting the same results, with less power usage, and far less water consumption.

Keep checking it out, I'm going to have loads of stuff coming in soon so I can really showcase my designs.

Re:Uh, did you look at your link?

[QuantumRiff]

Some Database Software is $5k or more PER Processor (some even price it by the core) so i could see a very, very significant cost savings from buying the much more expensive Power system.. I once got quoted $20k per core for a database replication tool for a disaster recovery plan. That software got installed on a 3 year old, single core machine..

Power + Linux = DEAD END

[Anonymous Coward]

The average mere moral will never get their hands on a power7 machine. There is no desktop option. I guess IBM could have one under development but people speculated the same thing about power5, power6 and nothing happened. Across the entire power landscape there aren't any machines which can compare to the average x86_64 desktop.

As a result, who from the open source landscape is actually going to be able to put time into Linux on Power?

Look at the distro story on Power,

RHEL: still there, but costs $$$
Fedora: I think so
Open Suse : gone
SLES: still there, but costs $$$ .. but you gotta wonder how long since Open Suse dropped power
Debian: wilting
ubuntu: gone

I guess IBM can shell out the bucks to get Novel and Redhat to support power, but is that really fostering a community? Can a community be a community when the price to enter is seriously expen$ive hardware and most of the people working on it aren't doing it because it's their passion but because it's a job?

Things were great when the Apple's G5 hardware hit... but nothing has been put into it's place since. Seems to me that IBM is just playing Linux lip service on power, specially reading all the posts about AIX here. It's been my experience as well.

Re:Power + Linux = DEAD END

[nxtw]

Is there anyone who can afford POWER7, but can't afford a RHEL or SLES subscription?

Re:Power + Linux = DEAD END

[RAMMS EIN]

This is pure speculation, but I wouldn't be surprised if IBM were supporting Linux on POWER because AIX has become or is becoming too expensive to maintain. They're still working on AIX and still shipping it, and it may still work better than Linux on POWER, but in the end, one IBM can't compete with a system that has both the open source community and a lot of major corporations backing it. So they've done the wise thing and started supporting Linux, so that they can eventually axe AIX and get a lot of work done on their new OS for free.

From your link

[raftpeople]

"The PowerPC 970 is derived from POWER4. It lacks some server oriented features, but does have an AltiVec unit. The 970 and its descendants are used by Apple and IBM and some high end embedded applications."

LPARs

[Gothmolly]

IBM gear gets you LPARs, with a real hypervisor that is laps ahead of all the other stuff.

Re:LPARs

[nelvinboy]

IBM gear gets you LPARs, with a real hypervisor that is laps ahead of all the other stuff.

You're absolutely right. We're just finishing migrating our data warehouse Database, ETL, and BI systems from smaller x86 boxes over to LPARs on a mid-upper-range POWER6 box. The performance and the on-the-fly configuration that our AIX admin can do on this box is SIMPLY AMAZING. We have lots of boxes running in VMWare environment, too. But the capabilities there performance-wise don't even touch IBM's visualization.

Re:Query

[Wesley Felter]

Anyone have data on how these compare to x86 and Intel's latest creations? Presumably, one could write an efficient algorithm for a variety of common computing tasks and port it to the different chips to get a cross-architecture performance estimate.

That's called SPEC CPU; here are some results: http://www.realworldtech.com/forums/index.cfm?action=detail&id=107244&threadid=107238&roomid=2

Re:Query

[scotch]

Holy shit, this guy is on to something. You could write these common computing tasks as a sort of "bench" suite of tests. Then on each architecture, you would get different "marks" against the "bench". Let's call them "benchmarks" for brevity. These "benchmarks" would give allow clear and unambiguous comparison of these various chips. Foolproof and brilliant!

Wouldn't it be cool?

[kurt555gs]

Imagine a PowerMac with a couple of these in it, and assload (actual technical term for large quantity) of RAM and a big display?

Oh, I forgot, the new improved Apple has told us that the Intel chip give us, the users, better performance.

I actually think Apple started it's slide into evildom with switching from Power to Intel.

Oh well, we can dream.

Re:Wouldn't it be cool?

[the linux geek]

Quite a bit more than $20000, if the prices posted on IBM's website are anything to go by, since the lowest-end, single-processor POWER7 server costs $34152.

Re:Wouldn't it be cool?

[Lunix Nutcase]

Imagine a PowerMac with a couple of these in it, and assload (actual technical term for large quantity) of RAM and a big display?

Yeah and it would sell miserably at around $75,000 a piece.

Oh, I forgot, the new improved Apple has told us that the Intel chip give us, the users, better performance.

Considering that you could buy around 30 of the highest end i7s for the same price as a single Power7 I would think you would get far better performance per dollar.

Re:Query

[afidel]

On a $/FLOP basis they get slaughtered by Nehalem-EX, but if you need flat out performance the Power7 system will be superior thanks to 2x more memory bandwidth per core and ~3.5x more interprocessor bandwidth. The basics for this type of comparison are Specfp_base, Specint_base for CPU performance and usually either SAP, TPC-C or specjbb for business logic comparisons.

Re:Query

[Jeremy Erwin]

I like your folk etymology. It neatly excises surveying from the discussion.

Re:Query

[RebootKid]

oh for mod points

Uh... Power7 also executes 6 instr/cycle

[uarch]

Itanium is superscalar to an extent that POWER doesn't come close to, with each core being able to execute up to six instructions per cycle.

Hate to break it to you but POWER7 can dispatch 6 instructions per cycle as well.

That little fact was revealed last year during the Hot Chips 21 presentation.

Re:This is Bad News

[FranTaylor]

There are exceptions, but they're becoming fewer and fewer.

That's not necessarily true. As the top end moves up, new opportunities and markets are created that were not there before.

Re:This is Bad News

[Anonymous Coward]

You'd buy a Power7 because it comes with 63 other Power7 friends in a single box and runs an operating system specifically designed for the ridiculous number of cores and capable of handling even the most data intensive legacy applications.

I agree that the high end server market is becoming smaller and smaller as time goes on but in reality there's still a huge backbone of legacy applications that require the sort of processing throughput only a single whopping great server can provide. The kind of applications that draw $150,000 3 month contracts for developers because nobody knows a damn thing about them, general public included.

Performance Comparison & More

[markus_baertschi]

I've ran through the performance numbers announced by IBM and what I found at spec.org (specint_rate & specfp_rate) of the other CPU's and roughly the following picture (give/take 20%):

• Power6/Power7: about 30 spec_int/fp_rate/core
• Intel Core i7: about 30 spec_int/fp_rate/core
• Sparc: about 10 spec_int/fp_rate/core
• Itanium 2: about 12 spec_int/fp_rate/core

So it looks to me that performance-wise Power and x86_64 are similar. Both seem almost three times as fast as Itanium/Sparc. However. in the commercial world scalability matters and I there are not many big (>4 socket) x86 systems around. Big Power, Sparc and Itanium servers scale to hundreds of cores and are built like mainframes with excellent RAS features. I see high-end kit from both sides, x86 and Power and the margins in the x86 world are not good enough to pay for the engineering it takes to get to the same levels.

If you compare Power and x86_linux with cars:

• You are read to spend some money to drive a nice car with excellent performance and stop at the dealer for inspections regularly then you well of with a Range Rover (=AIX Power server).
• You are going to cross Africa, will be on your own (and have the truck full f spare parts) and are ready to get your hands dirty then you want a Land Rover (=Linux x86).

This picture is far from complete, but shows what the choice is quite well.

Markus

The CPU for Playstation 4?

[Per Wigren]

Rumour has it that this baby is going to be the CPU of the Playstation 4 in 2012.

Re:The CPU for Playstation 4?

[Viol8]

Lets hope it looks better than that hideous mock up photo. My mum has a kitchen radio that looks similar to that.

Why can't Sony just build a playstation that looks like a nice sophisticated piece of stacking hi-fi instead of either
something that looks like it escaped from toys-r-us (PS1 & 2) or looks like something Bauhaus would have in their
living room (PS3) ?

Re:The CPU for Playstation 4?

[LeadSongDog]

Rumour has it that this baby is going to be the CPU of the Playstation 4 in 2012.

..so you can reverse-simulate from your bootleg of Avatar and watch Meryl Streep act?

You don't need to maintain such thing

[Ilgaz]

Apple doesn't need to do skunk works, all they need (and possibly do) is make sure they don't use any X86 specific stuff to the point of not being able to release it for any other CPU arch. Who would be that stupid? Well, Adobe. Adobe couldn't release their half ass Premiere for PPC along with another half ass audio editor making Premiere a further joke until Apple switched to X86.

As Apple maintains OS X for ARM Arch right now (via iPhone/iPad OSX), they aren't really doing the mistake of relying to X86 architecture. Who does such mistake right now? Well, Google Chrome to begin with.

As a PPC owner (G5 Quad, Mac Mini G4), let me tell you the sad thing. Once your users got the taste of running Virtual or real Windows and have Windows option, you can't really go back to anything. Perhaps AMD for certain cheap stuff later but still X86.

Even such an amazing enterprise CPU's resellers will have tough questions like "What if we want to run some enterprise Windows?"

And as a last thing, Apple never used the real, big POWER chips. The G5 (PPC970/SP/MP) is actually a POWER4 Lite. Now you can imagine what kind of power these enterprise monsters are.

ARM was Archimedes, a failed/expensive desktop

[Ilgaz]

You should talk about the history of ARM, what a sadly failed British Amiga like Desktop's CPU before they made wise choice of becoming a pure R&D house.

People talking about processors and thinks they are educated enough to the point of comparing enterprise Unix processors should start with Wikipedia information.

Imagine talking to someone early 1990s and show that Psion weird handheld and tell that weird OS will be powering 40% of smart devices in the future.

People doesn't even know that there is 1990s Apple, right at the beginning of ARM Holdings.

Re:ARM was Archimedes, a failed/expensive desktop

[LWATCDR]

I wouldn't dismiss the idea of an ARM desktop or even server CPU in the future.
Right now we have reached a point where CPUs are really powerful enough for just about normal PC task.
The tasks that could still use more power are those that traditionally where done on Workstations and Gaming.
Just how many average users need a quad core today? Not many because you see people buying a lot more Notebooks and Netbooks than Desktops.
The one bottle neck for the average users is HD video. nVidia and Broadcom have shown that it is better to move that off the CPU and put it on a dedicated GPU or just a decoder.
So in pure usability which is better a Tegra2, OMap4, or an Atom? Odds are the ARM solutions will be cheaper, faster, and use less power.
The one gotcha is software but I think Apple has shown that if you do right people will flock to develop on a new platform and people will buy new software if it is cheap enough.
If I could get cheap ARM motherboards with an OMap or Tegra 2 with PCIe slots and SATA ports I would buy several of them today.
At my office we have a NAS running an Intel CPU, two firewall Routers boxes, email Server, phone server, database server, and a intranet/CVS server. They all run Linux and we don't run to visualize them because of security and redundancy.
Each of those servers except the phone server could be replaced by an ARM based system.
Our database server which supports around 50 users is running on an old PIII 600Mhz system with only 512MB of ram. I am sure that a Tegra2 like system with a Gig or more of RAM would run it just fine and use less power as well.

Can I get a Microchannel CPU card?

[that this is not und]

I have a Power 1 RS/6000 box. The Power chipset is on one of the Microchannel cards. Maybe I can get a processor upgrade in the form of a Power 7 chip on Microchannel card?

No?

CPU speed vs memory bandwidth, I don't get it

[egnop]

I can understand why you would get a Power chip for pure number crunching.
But having a lot of data to chew away, I use p-threading for the larger jobs and let the rest of the jobs over to the os.

I was always under the assumption that data has to be delivered to the cpu fast, very fast and since the Power6 rs6000 only supports ddr2 I don't get it.

We recently bought a new rs6000, which has the Power6 in it(still has to be delivered), but the memory is 'only' ddr2, can someone enlighten me why this machine would run faster than my dual Xeon 5560 with triple channel ddr3?

The Xeon box only costed 1/2 of what the rs6000 costed

thanks in advance

Re:CPU speed vs memory bandwidth, I don't get it

[asciiRider]

IF you bought an RS/6000 there is your answer about why your dual Xeon is faster. I don't think they have made RS/6000's for like ten years man.

Re:CPU speed vs memory bandwidth, I don't get it

[egnop]

You are totally right, my bad, it's a blade js23

But still I don't understand why it still uses ddr2 and should it be faster...

Re:CPU speed vs memory bandwidth, I don't get it

[imgod2u]

Not all computational workloads are memory-bound. Some are, some aren't.

Re:CPU speed vs memory bandwidth, I don't get it

[uarch]

You might expect a Power6 to be at a bandwidth disadvantage when compared to the x5560 but when you look at the memory bandwidth specs they're surprisingly close. Once you move up to Power7 you begin to see a big difference.

Xeon x5560 memory bandwidth/chip: 32 GB/s [1]
Power6 memory bandwidth/chip: 32 GB/s (??? This may be wrong. I see references to 50GB/s but I believe that's peak)
Power7 memory bandwidth/chip: 100 GB/s [3]

Re:CPU speed vs memory bandwidth, I don't get it

[stevesliva]

Not sure about the blade you purchased, but it's common for each Power6 processor chip to share a package with a 32MB L3 cache chip. The bandwidth between the processor and L3 is huge.

Re:This is Bad News

[Anonymous Coward]

You buy it for the hypervisor, massive IO, and capacity upgrade on demand. Forget what you know about virtualization from xen and VMWare. The POWER hypervisor lets you add (or remove) ram, buses, and processors from a running server. You can even set the memory and cpu to pull from a shared pool (with set priorities and limits). The internal 10Gb network doesn't hurt either.

Try scaling your xen system when you are IO bound to disk. POWER offers physical and (fast) virtual IO, giving each partition "big iron" IO capacity.

POWER is made by people who understand scaling. Commodity boxes are made for people who like big numbers printed on the side of the box and don't understand why high CPU and memory numbers are useless if your disk array can't keep up.

Why Did Apple Switch?

[Doc Ruby]

When Apple switched from PowerPC to x86 for Macs, Steve Jobs said it was because Intel's energy efficiency was on a much better curve than IBM's. But the Power7 is 2x as fast at 1/4x the efficiency. I don't think Intel's performance:efficiency has improved as much, and indeed IBM might already be better MIPS:W.

Probably Jobs just wanted the scale economies and vendor diversity, and the Wall Street lemmings, that come with Intel CPUs. But why did he say it was performance:efficiency when he'd look wrong after a short while? Was it just a better excuse than admitting he'd been wrong to stay off Intel for so many years? Or maybe Intel just made him some kind of deal we don't know about?

Re:Why Did Apple Switch?

[imgod2u]

Power7 is also 130W+. IBM couldn't deliver a power-efficient mobile chip that was on par with Intel's offerings both in performance and power consumption, so Apple cut it. You can only false-advertise so much before people realize what BS backdrop commercials are.

Lynnfields TDP at ~65W and most of the time less, yet offers performance on low thread-counts (1-4) similar to higher end server processors. That's saying something.

Re:Why Did Apple Switch?

[jeremyp]

Probably Jobs just wanted the scale economies and vendor diversity,

I think he wanted a processor he could put in his laptops that didn't suck.

The G5 was too power hungry to put in a laptop and the G4 clocked at 1GHz when "equivalent" Intel and AMD laptop processors were achieving about 2GHz (if my memory serves me correct).

i5/os?

[iSrzMan]

"Power7 servers will run AIX and Linux" ... and presumably i5/OS?

Re:i5/os?

[ralphdaugherty]

yes, i5/OS as well immediately available, although now that we're past Power 5 and 6, IBM is calling it i...something or other totally inane.

I think I'll just stick with i5/OS.

  rd

Re:This is Bad News

[swamp boy]

Mod parent up.

These systems are extremely impressive. The hypervisor functionality by itself is amazing. Too bad IBM doesn't know how to market these systems very well.

Re:This is Bad News

[jacobsm]

Even though they aren't the same chip, IBM uses the technology and most of the manufacturing processes for the Power series CPU's for its very lucrative mainframe line. The current Power 6 technology is used for IBM's z10 processors and the next generation z11's will be powered by the new Power 7 based CPU's.

This is money well spent by IBM.

Re:IBM makes CPUs?

[TerranFury]

Joke? Anyway, since forever; just not for the consumer market since Apple jumped ship for Intel a few years back. They make very fast, very expensive server chips.

Re:This is Bad News

[asciiRider]

A couple of reasons -

Dynamic LPAR, add 5 fiber adapters and 20G of RAM without bringing the OS down. Maybe add 4 virtual processors and only entitle the system to 1, allowing it to donate unused cycles back to the sharepool.

When I want to apply a service pack I don't get some silly Python error. It just works. Oh, also I can apply the service pack to a copy of the disk while the system is up and just reboot when I please. (alt_disk_copy.)

Feel like adding a virtual ethernet adapter to the system mentioned above? No Problem? While it's up? No problem!

If you want to move a completely virtualized system to a different machine, while it's running, no problem!

  AIX LVM simply blows everything else away.

The briefing center in Texas has great breakfast burritos.

Anyway, you get the idea. You don't simply buy a processor, you buy a enterprise Unix OS that can do everything. Linux is almost there...

Power consumption

[ThrowAwaySociety]

What's your point? All that matters is price/performance. If processor A can get the job done more cheaply with twice the processors, why would I care?

Because the cost of powering and heating those extra CPUs is not trivial for large-scale deployments.

Re:This is Bad News

[foldingstock]

Or, learn a damned thing about those systems and only work one year for the rest of your life. :) Beats three months every year.

Re:IBM makes CPUs?

[Jenming]

IBM makes consumer processors. You will find them in the XBOX 360, Wii and PS 3.

Re:More accurately on OS support...

[fatherjoecode]

Thanks for mentioning the IBM i. We, the beleaguered computer nerds who toil writing code for that O/S, have always felt a little left out. I call upon the /. editors to please correct the blurb up top.

Windows7?

[SnarfQuest]

Is Power7 designed for Windows7?

Please, don't let it be!

Re:This is Bad News

[Rexdude]

You'd buy a Power7 because it comes with 63 other Power7 friends in a single box and runs an operating system specifically designed for the ridiculous number of cores and capable of handling even the most data intensive legacy applications.

And as I mentioned in my earlier comment in this story, you can still get Linux running on this hardware if that suits your needs better.

Re:IBM makes CPUs?

[TerranFury]

IBM makes consumer processors. You will find them in the XBOX 360, Wii and PS 3.

Ack! Yes! How'd I forget about that?!