Slashdot Mirror
Cell Hits 45nm, PS3 Price Drop Likely to Follow
Septimus writes "At this weeks ISSCC, IBM announced that the Cell CPU used in the PlayStation 3 will soon make the transition to IBM's next-gen 45nm high-k process. 'The 45nm Cell will use about 40 percent less power than its 65nm predecessor, and its die area will be reduced by 34 percent. The greatly reduced power budget will cut down on the amount of active cooling required by the console, which in turn will make it cheaper to produce and more reliable (this means fewer warrantied returns). Also affecting Sony's per-unit cost is the reduction in overall die size. A smaller die means a smaller, cheaper package; it also means that yields will be better and that each chip will cost less overall.'"
I don't know what it is about measuring things in nanometers and terabytes that gives me such a hardon.
Thank you IBM.
PS: Please don't put Skynet online.
Virginia is for lovers. EVE is for griefers.
My only question is, will this reduce the cost?
Actually, I think that there are too many processors and cpu's. Too much internal memory. I yearn for the 8086 chip still. Or an IBM 360 mainframe. Now that was computing.
This would be a great thing if they allow PS3/Linux users to access 7 of 8 SPUs instead of only six.
Otherwise, it's nice but not that big a deal...
This fits in well with the rumors of a slim version of the PS3 in the works. See here for more details.
Curiosity was framed, Ignorance killed the cat.
A price drop would be nice (though the PS3 is now competitive), but the more interesting bit is when is the PS3 slim going to appear. All the pieces are in place for a slim. Sony have been aggressively shrinking the motherboard in the PS3, and the chip size has dropped from 90nm, to 65nm and now 45nm. All that means less power (smaller PSU) and less heat (less fans & heatsinks). There have been other announcements such as thinner blu ray reader headers. It can only be a matter of time before a slim and I think it will hit before the holidays this year. I think it will sell by the shitload too when it does appear. The question is will we see a slim 360 to compete with it? I think there must be a lot of empty space in the 360 too.
The article mentions the cost savings to Sony (maybe they'll be passed on to the consumer...two or three years from now), but the real kicker is at the bottom where IBM apparently had to maintain cycle compatibility with the old chip to make sure they don't break any games. They didn't use the die shrink to optimize or enhance any parts of the chip like you normally would. The supercomputer folks might end up losing out a bit in an effort to keep the game console folks happy.
I read the internet for the articles.
Which is another important factor in bringing the price down. Percentage-wise with more die per wafer yields may go up as well; but in the end yields will be dependent on other things such as how good IBM is with its 45nm process.
Think it will be possible to swap out the chip/board but still save my emotion engine? i cant be without my beloved emotion engine; i'll have nothing left to gloat about to my friends who don't have it now.
~> FatJackson
I believe it's because the chip is smaller therefore more fit on the same size wafer.
for a PS3 price drop, but.... when's it going to happen?
It would be really great that they are moving to a smaller process, (/me takes deep breath)
IF THEY WOULD SELL YOU THE DAMN THINGS!
Where I work, we approached them to try to buy Cell processors for our equipment: the SPUs would make dandy DSP replacements, and we really could use the closer coupling of the processors instead of having a bunch of DSPs and spending all our time schlepping data around.
IBM wouldn't sell us any modules, wouldn't let us design our own CPU board, nothing. They seem supremely uninterested in actually getting these out into the hands of anybody other than their own divisions and Sony.
HEY IBM! How about you guys release these in a MicroTCA formfactor, or as a module that can be integrated into a MicroTCA?
www.eFax.com are spammers
Is the fact they've dropped hardware PS2 emulation.
[Fuck Beta]
o0t!
Since when does going to a smaller process increase yields?
Always has.
Assume there will be 20 defects on a wafer that will render 19 large chips (out of 100) unusable. Your yield is 81%.
Same 20 defects, but affecting 20 small chips (out of 170). Now your yield is 88%, or 150 chips versus 81 chips per wafer.
The number of defect sites per wafer is generally rather constant, thus the more chips you can fit on a wafer, the better the yield.
so, the PS3's doesn't need cooling?. just leave the damn thing in. I absolutely hate power bricks. I still have an old Compaq Armada notebook from 1998. It doesn't work anymore :D but it's nice that it has the power supply inside. No need to carry power bricks around. I look at those huge Sony Vaios with 14 or 15 inch screens, and I wonder... so much space, and they can't put the power supply inside? I guess it's because the power supply is very likely to fail, so it's cheaper to replace a power supply than having to send your computer for repair if something goes wrong.
If' they're dropping cooling components due to lower heat output, I wonder if that means this picture is for real.
Moore's law is dead. Atoms aren't getting any smaller. With 5 atoms thick, when you try and go to 2.5 atoms thick, let me know and I'll get far away.
It isn't the smaller process that increases yields, it is the fact that the ensuing smaller die takes up less space on the wafer. Less wafer real-estate == less chance that a defect in the wafer will occupy the space of a particular chip.
For instance, say one 300mm wafer has 50 defects evenly distributed over it's surface, and one wafer can hold 100 chips with the old process, 200 with the new. The 10 defects result in a 50% yield with the old process, a 75% yield with the new process.
That said, yes, almost all new processes take a little while to work out the bugs. But after the bugs are worked out, you can achieve much higher yields...
SirWired
Power supplies also generate a lot of heat -- notice that those bricks tend to be warm under load, even through that insulation. Put them inside the laptop and you're adding a bunch of heat to a place that you want to be removing heat from. So you need bigger fans and it takes even more space. It's just unworkable.
I have seen the future, and it is inconvenient.
From the article: "So IBM may have suckered Sony into buying a supercomputing coprocessor disguised as a gaming chip, but it looks like Sony could get the last laugh." Why would they? IBM is selling more than ever.
The reason is that wafer size doesn't change. I don't remember what is current, 8 inch I believe (that's the largest I've seen) but regardless. So when you reduce the size of an individual chip, you get more chips per wafer. Now unless the percentage of chips that fail increases, that means you get a better yield/wafer.
Well cost is based per wafer. It doesn't cost any more to make a wafer with 1000 small chips than it does to make one with 4 big chips. In either case it is the same size wafer, same mask, same process, etc.
Now yield could go down if a company has problems with a new process. Suppose that the old process yields 10% non-working chips per wafer. You get a new process that yields 20% more chips per wafer than the old one, however now 50% of them are non-working. That would equal a lower yield, despite the more chips per wafer.
However assuming a roughly equal failure rate, shrinking the die size will increase the yield.
The article linked to says power consumption will be 40% lower than the 65nm predecessor, but I haven't been able to find a reliable figure for the 65nm power consumption. Does anyone have that or a link? All I've found online are puff-pieces saying the power is very low or may be in the future, but no actual spec.
Simply put, they reduce the cost of production, they lose less money on each one they sell. Considering the Playstation 3 is slowly gaining market share at it's current price, they have no need to drop the price right away.
~Mike (Titan_X)
Does anyone have insider or otherwise info on how much more rulesets does the 45nm t ech have, compared to the 65nm?
"The agriculture ministry is not in charge of Gundam" - Japanese ministry official.
Wouldn't the smaller process increase the defect density?
And the men who hold high places must be the ones who start
To mold a new reality... closer to the heart
I remember the biggest problem with the PS3's availability and cost was the blue laser (as it has low yields and is expensive to make). Sony was already taking a hit on cost, since a stand-alone Blu Ray player in November actually cost about the same or more.
It's nice that the cell processor is lowering in cost, but I'm not sure that it ever was a significant enough percentage of the unit cost to see a drop of more then a few tens of dollars.
Say it with me:
Will it blend?
Pancakes. Oh I blew it.
The size of a defect is of a fixed size. Usually it is a particle of dust that got in the way of the optical etching process. The distribution of such defects is even across the surface of the silicon wafer, so the distribution can be modelled mathematically.
Suppose there are 20 defects across the wafer. If your chip were the size of the entire wafer, it would be guaranteed to be defective.
Try half the size of the wafer, and there would be on average 10 defects. A quarter of the wafer, 5 defects. If you have a chip that is one hundredth the size of a single wafer, then the odds are now in your favour; on average 20/100 that you will have a defect, 80/100 that you will not.
The Cell processor is etched with eight processors anyway. If one is defective, they can ignore it, otherwise if all eight are working, then they will just deactivate one.
I wonder how long it will be before they start adding more processors to the chip.
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
Relevance of CBE beyond PS3 of course depends in large degree on its computing performance. For the applications I've looked at, I haven't been very impressed. They say it does 204GFLOPS, but approaching that requires being able to use all multiply-add instructions, which count as two operations. (Some sources say the two operations per clock cycle per SPU is due to there being two pipelines, however, only one of the pipelines handles arithmentic operations and the other is exclusively for load, store, control, and a few shift operations.) Also, it seems to take a lot of select, shift, and shuffle instructions to make efficient use of the quadword (SIMD) instructions. With Xeon and Opteron, use of the quadword instructions seems to require far fewer other additional cycles. And this is with floats, with instruction related stalls completely eliminated on CBE through careful loop unrolling and other methods. (The quadword instructions have 6 cycle latencies.) I can only get performance comparable to 2 quad-core Xeons, which doesn't seem that good considering what is advertized, and considering the 4x difference in the peak performance specs. And CBE does much worse where double precision is necessary, with 6 cycle stalls being unaviodable on every instruction. It seems overblown. Comments?
I'm not sure about current hardware revisions... but the 360 I have... not very much space to go around.
But with some hard work it appears you can thin it out a bit as demonstrated in the Xbox360 Laptop. My only question would be the failure rates on these things due to heat issues. As it's already been shown your standard store bought Xbox is affected by the excessive GPU heat causing motherboard warping and extra stress to solder points.
I'm sick of following my dreams. I'm just going to ask where they're goin' and hook up with 'em later.
This assumes that the chips are small enough that there are never 2 or more defects per wafer. It may be possible that a bigger chip that would have 2 defects is split into 2 chips each with one defect each. Thus, the yield increases in a somewhat, but not completely linear fashion.
Even those who arrange and design shrubberies are under considerable economic stress at this period in history.
My understanding is that defects are caused by imperfections in the wafer itself. The smaller process allows you to squeeze more chips onto the wafer but doesn't have any impact on the imperfections.
Now that blu-ray has won the format wars and a PS3 gives you a gaming console AND a blu-ray player in one bundle, why would Sony lower costs? Wouldn't they just e happy with the higher profits?
Good luck getting your body to replicate chips.
I doubt you'd want to leave it on. When the XBox360 went from 90nm to 65nm, power consumption dropped 50W, from 170W to 120W peak, but 120W is still a whole lot of watts. The PS3 is in the same ballpark for power consumption. The PS2 and Wii use much less. My PS2 slim uses 25-30W.
What a stupid inflammatory conclusion. A remap is just what it is - a remap. The article blames IBM for missing the "opportunity" to spend a lot of money on tweaks, i.e. on logic design and verification, and to screw up compatibility for Sony at the same time.
They gave away too much info. Now we will expect an affordable, reliable game console.
It is dangerous to be right when the government is wrong.
Last I heard, Sony was still losing a ton of money on every PS3 they sold. So even if this upgrade makes it significantly cheaper to manufacture PS3s, I don't see why that would lead to a drop in retail price.
If anything, I'd guess Sony wants to keep the PS3 at its current price, now that they've basically won the next-gen DVD skirmish. Plenty of people who want Blu-Ray players probably already see the PS3 as a good choice (just like I bought a PS2 to play DVDs back in the days of yore).
Go and play a little with this tool. Choose a fixed amount of defects and play with chip dimensions.
My 0.02 cents
Why deactivate one?
Not so long as the consoles continue to sell at the current price. Sony charges what they think people are willing to pay, no more and no less.
Nowadays, one of the hottest components in a computer (and consoles) is the GPU.
Most of the XBox 360 problems come from a very hot GPU situated under the DVD drive.
I think taht the GPU used in the PlayStation3 is a 90nm derivative from the GeForce 7800 (a quite hungry GPU) (you can see it in http://en.wikipedia.org/wiki/RSX_'Reality_Synthesizer' ). I think that this GPU should be shrink into another process before creating a "Slim & Lite" PS3...
That's great and all but let's also hear about what's happening with Nvidia's contributed RSX. When will it his 45nm and be included in the PS3? I'd reckon that a 45nm Cell and 65nm (or heaven forbid 90nm!) in the PS3 would still generate a whack of heat. I understand that Nvidia has their latest iteration of GPU's coming out at 45nm so I wouldn't think this to be a stretch. Any official word?
Because some customers would be getting a PS3 with 8 active cells, and some would be getting one with only 7. Since they paid the same price they would be pissed off. If they always deactivate one cell, they can use all of the chips they make with less than 2 bad cells and everyone gets the same product.
Anyone know what the die size of the XBox 360's IBM cpu is? Wouldn't be surprised if it's the same, of course.
"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
The redundancy of the Cell's 8 SPUs (DSP coprocessors) is the main point of the Cell's design. Defective SPUs (nearly always from dust particles in the nearly - but not quite - perfect "clean rooms" in which they're manufactured) can be tested and turned off as they roll off the assembly line. The shut down SPUs are even physically disconnected from power by hard fuses, so they don't cost any performance in operation. The perfect Cells with 8 SPUs cost the most, in high-end IBM RS/6000 workstations (and some blade servers). 7 SPUs go into PS3s. The rest of the yield, supposedly down to a single SPU (but even 0 SPUs still have a 3.2GHz PPC and superfast IO), go into HDTVs and other consumer electronics. All of the yield gets sold, instead of a fraction in older manufacturing processes.
So smaller dies don't really affect Cell yields. Smaller dies just mean smaller parts of the wafer that would get spoiled by a single defect, which is already taken care of with the redundant SPUs.
In fact, smaller dies mean multiple defects are less likely to land on a single die. Which means that more Cells would turn into low-SPU, cheaper Cells. While larger dies would concentrate multiple defects into a single dies, by landing on a single die more often, leaving more perfect Cells getting the highest prices.
45nm does mean more Cells, at any defect rate, per wafer. Which means, for the same number of defects per wafer, more dies per wafer. So there is a yield increase, but not for the same reasons as traditional ones. And of course 45nm has so many other valuable benefits, like speed, and more transistors if they keep the same die size, that the move is very valuable overall.
--
make install -not war
The PS3 is interesting because it's so much power in such a cheap box, but it's subsidized by Sony. I think Sony will be lowering prices less while reducing the subsidy more.
But where are the Cell PCs already? The PS3 is cute, but it's locked down with a Sony hypervisor, it's got no PCI or other expansion, only a single SATA connector, and a puny 512MB hardwired RAM (its Cell can rip through 512MB, peforming 64bit floating point math on it all, in under 0.0025s). Its RSX video chip is locked out from Linux, so no HW acceleration (and no addon videocard is possible).
IBM is now cranking out these chips. It lost Apple, its biggest CPU (PPC) customer, to Intel. Where's a PC built on a Cell that includes PCI-e, expandible XDR RAM, Gb-e networking, and a more open nVidia graphics card (or two)? Since the Cell is cheap due to its higher yields, a $1000 Cell PC could make a $1000 Intel PC (Mac or Windows/Linux/etc) look like a 286 with its extremely high speeds. Sony has proven it can be mass manufactured with mostly commodity parts for under $750.
Since Ubuntu already runs on Cell, a cheap Windows killer could take the Cell architecture to the top of the CPU stakes in record time from release. It would be a much easier/cheaper/faster target for porting PS3 games than Intel PCs. Apple, which supposedly dropped PPC for Intel because of heat:performance limitations, would have to look seriously at a return to PPC, especially since 45nm Cell with only a few SPUs could be a perfect fit for an iPhone successor. If not from Apple, then from someone smart enough to use Cell in the biggest market of all.
--
make install -not war
So I won't be able to buy a full 2 Wii's for the price of a PS3, only 1.8? Getting closer to reasonable...
"Teach a man to build a fire, and he's warm for a day. Set a man on fire and he's warm for the rest of his life."
And because of the new and improved processor that is smaller and runs cooler and more reliably we are announcing a 23% price increase for fall of 2008 ;)
ALL of these processors don't go to Sony, IBM has other uses.
yes, they get warm -- they don't heat up like a microprocessor. if you put a small fan over them, they almost don't warm. if it can work without a fan in an airtight case, it would be better in a more "open" space, such as the inside of a computer. there is a fan somewhere in there.
I think the general slant of the question was whether the price drops now as a result of this, or does Sony put the saving toward reducing their losses on each system sold.
Sony doesn't lose anything on any PS3 sold - I'm not sure they ever did, but they certainly do not today. This processor shrinkage is just another in many steps that Sony has already undertaken to reduce the cost of the PS3 to assemble.
People tend to discount Sony, but what they do not understand is that Sony is very, very good at ratcheting down costs constantly while still keeping quality pretty high. That is a powerful force even if they make other mistakes along the way.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
A 40% drop in heat will make a huge difference in the suitability for business application of the PS3. (Before you say that business use of a PS3 is contradictory, please consider the accommodation industry.)
The instant they can get a PS3 (or an Xbox) that does not spew heat and use fans akin to a Boeing, it will have a place in the entertainment centers in luxury accommodation suites around the world. The region free PS3 game discs will seal the deal. Surfing internet on the TV and being able to show photos straight from your memory card is also a plus.
Late last year, we tried rolling the current model of PS3s into some guest suites. In the end there was no way to accomplish this without a major retooling of the entertainment centers, costing hundreds of dollars extra per unit. In one case the excess heat generated by the PS3 caused the TV to overheat!!
The drop in power bills will also be a big plus, as guests will generally never be bothered to switch off an appliance. I had thought that the PS3s were supposed to automatically regulate the amount of processor power needed. But they seem to run as many fans even when idling at the top menu.
For business use the maintainability and operation costs are a much bigger factor than the original cost per unit. If they can actually get the heat under control, Sony will break into a huge new market of corporate clients.
Particularly if they were savage, bloodythirsty Aztec chihuahuas who I could order off the cart at a moment's notice to devour anyone saying "Aww, wook at the cute puppies".
Help poke pirates in the eyepatch, arr.
That nVidia is on 65nm, I'd expect they'd move the 7800 to that process, if they haven't already, for the PS3.
XML is like violence. If it doesn't solve the problem, use more.
That we can't have quad core Intel processors, since Intel has to manufacture chips for laptops which need lower power utilization than is currently not feasible with quad core. Oh wait, Intel manufactures different variations of the same architecture to suit different products/environments. Power chips, likewise, are used in tiny embedded appliances with low TDP and other crippling situations, but also at extremely high TDP in Unix servers, with many more capabilities.
It's obviously a bad move to presume IBM would be so stupid as to lock themselves into only ever implementing one specific instance of a design without variation if they want to deviate. I agree with the other post, seemingly pointing and laughing at IBM over this 'limitation' is going to backfire if IBM considers it a viable platform outside of gaming and releases a serious variant of the product as this advancement gets rolled out in parallel.
XML is like violence. If it doesn't solve the problem, use more.
Much to AMDs dismay, the new Intel 45nm chips are remarkably good. Only 4 watts idle and 50W when at max load and also a bit faster per clock. That idle figure is quite amazing considering 20-30W was considered good a generation or two ago.
So if Sony gets similar benefits from 4nm+high k, the new PS3 will use much less power. They managed to get it to work without active cooling as it was, so this should give them a lot of headroom to shrink the box or do something new (build in PS2 hardware again would be a good idea...).
For every expert, there is an equal and opposite expert. - Arthur C. Clarke
Defectivity (i.e. the "dust problem") is just one of the yield detractors. There are many more and they get worse and worse. For instance, there are litho problems, etching problems, CMP problems, not to mention gate leakage, and a bunch of other parametric issues. So, you can not just look at defectivity. Even if you did, with a smaller feature size, small particles that could be tolerated in an older generation will now cause yield loss.
PS: the distribution you are talking about is a poisson distribution
Cape 69$
Red underwear 15$
Cart 200$
100 purse sized dogs 15 000$
YouTube immortality Priceless!
Just wonder if there will be something like VIA's Mini ITX with CELL processor on board available somewhere in future.
/Z
Imagine building own MythTV with that.
Ultimate PVR!
Processor dependent software (for the Mac) was essentially irrelevant at the time of the switch.
Not quite, since the emulator will never be as fast as the real CPU (ergo, executing binaries that can't be re-compiled) but that was not an argument in favor of the x86 at that time.
One might argue that drivers don't have to be re-written as much, I suppose, but the reality of driver availability for the Mac has not really changed, as near as I can tell.
The switch to iNTEL seems like a great thing, but it will turn into a blind alley within five years. The x86 architecture has some serious flaws that iNTEL has been able to fake their way past by pushing the speed, but the wall stands in front of us, and the wall requires a different architecture to get through.
Von-Neumann equivalence is only useful as long as you have effectively infinite stack and infinitesimal CPU cycles.
(Our entire industry, for the last thirty years, has been behaving like the proverbial drunk looking for his wallet under the street lamp where there is light instead of down the other side of the bar where he dropped it in the dark.)
Computer memory is just fancy paper, CPUs just fancy pens with fancy erasers; the 'net is just a fancy backyard fence.
does sound ridiculous, but I'm imagining 50 chihuahuas and one clydesdale and I'm thinking, if I could figure out the harness problem, and the problem of getting them all pulling in the same direction, I'd at least like to give it a try once.
As a rough guess, fifty chihuahua-power is probably greater than one horsepower.
How many Alaskan huskies equal a clydesdale? I wonder if anyone has mentioned horsepower per husky on wikipedia.
Yeah, off-topic, except that the harness problem is somewhat analogous to the problem of writing efficient muli-(erm)threading software.
Computer memory is just fancy paper, CPUs just fancy pens with fancy erasers; the 'net is just a fancy backyard fence.
Here I am, replying to AC... Slow day indeed.
Sorry, but you are wrong (as your moderation points out).
You can buy a PS3 to do numerics and the Cell inside it is an average performer. Not bad at all for under $1000.
But, if you need to upgrade (and consider your workload is heavily parallelized and optimized for the SPUs because it already runs on your PS3's SPUs) you can buy one or more IBM QS21 blades and a suitable chassis. It's obvious these new Cells will be in these blades as soon as they become available. In the blades, the Cell is not limited as it is in the PS3, there is plenty of memory for the PPUs and you can run all your SPUs at full throttle if your data and programs demand it. And, while you are at it, you can add POWER or x86 blades to the chassis as well as Linux does not run particularly fast on the Cell PPUs and you may want a fast machine to feed the Cell node.
Sorry if you wanted current supercomputer power on the cheap. The PS3 is good enough for a lot of stuff and a lot cheaper than anything that approaches its numeric performance.
http://www.dieblinkenlights.com
Jeez. You ALWATS get a performance boost when you move to a higher resolution process, that an bugfixes and a few choice tweaks, and you wind up with a significant performance boost. I wonder what the performance boost will do?
And which lucky bastard got to syphon off the rounding errors?
They already dropped the price for the reason that their sales were getting pummled by the wii and 360 - I seriously doubt anyone outside the tech sector is going to know what a die shrink means and/or correlate that to a cheaper price.
and also, it seems to me that the PS3, at 399 is a great deal - considering it has a blue-ray player in it and rivals the strongest console systems, graphics wise. With that being said, the reason I will not buy one of their new versions is it will not play playstation 2 games like its predicessor.
(yes I know my spell check is broken and coffee hasn't kicked in yet.)
I think those 2.5 thick ones run real *hot*.
English is not my first language. Corrections and suggestions are welcome.
45 nautical miles? Wow -- that's almost as big as the Pentium Pro. What's the big deal?
Oh. SI units. My bad. Didn't expect to see them, stateside...
Paleotechnologist and connoisseur of pretty shiny things.
Would it make more sense to keep selling at a loss, even if it is $30, because then that person will buy 5-6 games in the year making up for the loss 10x.
Consider each small loss, a 1000% investment. They can print billions of BR games, at $1ea.
If you can magically sell 2x more hardware thats a win, if you can make 2x more from your factory.
Liberty freedom are no1, not dicks in suits.