Intel Labs unveils PC power plans Engineers at Intel Labs are toiling away at a "holistic approach" to energy savings that could result in lower-power processors, shrunken power bricks and batteries and worldwide power bills shrinking by billions of dollars. At a gathering of reporters Thursday in San Francisco, the director of circuits and system research at …
Every time I do something to save energy the power company raises their rates negating my savings, last year alone they raised their rates FOUR times...the best I could do is hold my own but even now that is a loosing proposition.
I was going to get a laptop to reduce power consumption from 28,000 KWH per year to 9000 KWH per year, but figuring in rate increases it would take me 26 years to beak even on the purchase price of $434.00 for the Laptop.
What the world need is more of what Intel have planned and stiffer control of Power Utility Companies.
http://www.theregister.co.uk/Design/graphics/icons/comment/dead_vulture_32.png
I agree with this premise... to a point. ARM does inject competition at the lower power CPU and that does have it's place. However, it is just plain physics that you have to use more power to get more power. I know it sounds strange, but that's the way it is in this universe.
High-power (i.e. FASTER) processing will require more juice at the brick. Some advances will probably take this to what seems incredible levels of less than 60 Watts per CPU. However, I don't see high-power CPU's getting much below this without a breakthrough in manufacturing and processing.
The advances made so far are nothing less than spectacular if you go back a mere 5 years.
And this changes *nothing* about that.
BTW. How many of those "power spikes" happen at bootup? IIRC power spikes cause include HDD startup and smoothing cap charging for the DRAMs. Unless your going to keep that super cap (which is not *just* a name but refers to a specific technology type) charged after shut down.
Historically the "Shrink the devices will save energy" line is complete BS. Narrow wires are shorter but also *thinner* but not taller, so resistance goes up. Multiply this by the "Oh look, we've got so much more space free let's stick a big cache in there" and you get a power saving of zero.
Previous posters also noted that that this tech was actually demonstrated on an ARM processor, not anything resembling an Intel Pentium.
It will probably help make Intel some more cash. Now your processor can be as throttled as your internet connection.
Thumbs up for ingenuity. But not a good solution.
"It depends on how agressively you want to correct errors, so there's a range," Wang told The Reg. He claimed the overhead to be between three and five per cent: "It depends on how happy you want to be."
I imagine most programmers would like the CPU error rate to be zero. Then again, most end-users would like the programmer error rate to be zero, too, and they make do with a rather higher figure. So...are we on the brink of a new programming paradigm: Error-Oriented Programming?
That high end PC power supplys have 'super capacitors' built into each power cable. Its quite neat and is designed to help keep the power clean for all the devices and rather more importantly to provide sudden peek supplys or brown outs (the little ones you dont notice but your hardware might).
http://www.antec.com/Believe_it/product.php?id=MjEyOQ==
^ Of course they have called theres PowerCache, and it has little to do with trying to save the world by using less power... infact if your buying a 1200W Psu like this, I suspect you are a PowerCompany Share holder.
-cen
What the heck? That doesn't make sense at all.
If you're machine is consuming 17.5W, your brick will on be consuming about 22 to 25 Watts (depending on conversion efficiency). You can verify this with one of those kill-a-watt gadgets.
But even better, it can be refuted with math. Electric power is amps of current through the load multiplied by the voltage across the load. P=IV.
These bricks are voltage regulated, usually around 19.5V for the bigger/hungrier laptops, at around 8 AMPS max. So if you're laptop is only consuming 17.5 Watts, it's still consuming 17.5W from the brick, but the brick is only delivering .9Amps.
Not to mention, most laptop bricks are oversized not for spikes in demand from the motherboard, but so that they can run the motherboard at maximum throughput while simultaneously charging your possibly depleted battery in a reasonable amount of time. Most of these motherboards will already run on much less juice than the rated power output of the power pack, but if you only supplied that much, you couldn't charge battery.
Finally, I hope he was not implying that the processor would have a 5% execution error rate on the running code, but that it was having to retry things at a 5% rate, or increase its power consumption to maintain no greater than 5%. I can't imagine how a programmer could begin to predict how analog noise (some of the most random information available) could effect your code flow.
So, if you want to charge your machine, and run, say, video encoding, it will start to slow to a crawl.
This sounds like a godawful user experience.
Like the other guy said, your power brick/PSU does not consume max power all the time, this would make no sense, it would have to provide extra load over and above the load the PC is currently drawing, and then piss it away in the form of extra heat in the PSU. This simply does not happen. Efficiencies vary through the range of load (inefficient at very low and very high loads, but efficient inbetween, typically), and choosing the right size PSU is beneficial, but undersizing the PSU is utter folly.
You can get a power meter at any given time, plug it in between your machine and the wall, and watch the load change based on any arbritrary condition (put more CPU load on, put more GPU load on) and watch it drop back down after you stop drawing that load.
This "super cap" idea is utterly retarded, sorry to say.
I hate these sorts of slides showing massive savings using bad ass-umptions!!! What marketing/sales/politician droid got a hold of these slides? (grumble, grumble, rant, rant)
NO! There are not 150 million PC's that will automatically give you 6$Bn of savings if ProxZzzy is deployed. A fare percentage of these PC's already have power management operating in some manner and will have less savings, plus many have long running processes that ProxZzzy will not be suited... you know, like screen savers, SETI at home etc....
Yes ProxZzzy is more efficient and will be easier to deploy, but it will not save the world $6bn!!
Grrrr
Ever wonder what you're going to do with a six-core desktop? I know I do, seeing as most of the desktops in our business are used by users doing little more than email, Word and browsing (well, web-interfaces to nasty stuff like SAP, but not really heavy use) - hardly requires the dual-cores most of our users already have. Yet the purchasing cycle grinds on, and soon we will replace the current crop of perfectly adequate desktops with new, even more powerful desktops. So, my idea would be to use all those cores more effectively.
In particular, I'd like the OS to run three copies of the same thread in parallel, compare the three results for an error, check if the result is likely to cause a problem and only go to the next cycle if it is all looking good. Not new stuff, it's how NonStop does it in high-end computing. But I will probably have lots of CPU cycles going spare in my new desktop, and the idea of a Windoze that doesn't ever blue-screen or apps that don't crash sounds good. Only problem is that would require more power as more cores would be kept active, and everyone seems fixated on reducing power consumption.
This post has been deleted by its author
Electrically efficient desktop PCs can be made now, but are not. The most clear evidence for this is to look at the UK site for energy saving electrical goods:
http://www.sust-it.net/energy_saving.php?id=20
The top six machines costing about £7 per year to run are all Apple Mac Minis which are proper office PCs with a Intel Core duo processor and dedicated graphics. In competition, other major suppliers offer Atom based machines which still cost more to run than the Mac mini. It costs over £12 to run the nearest rival PC based on on a decent processor, the Fujitsu Siemens ESPRIMO Q5030. There is no rocket science in the Apple approach, simply the use of processors that Intel designates for laptops, a 2.5" disk, an efficient power supply and good mechanical design.
Manufacturers are keen to to publicise their green credentials, but they need to be doing more to give consumers the choice of more efficient PCs. Consumers need to stop buying PCs solely on the basis of the most GHz and cores for the money.
"Manufacturers are keen to to publicise their green credentials, but they need to be doing more to give consumers the choice of more efficient PCs. Consumers need to stop buying PCs solely on the basis of the most GHz and cores for the money."
Should yes. However when they start asking people it's never *that* big a priority against bigger RAM, better graphics, faster HDD etc.
Plus it might mean them having to do some actual design work, instead of just putting in an order to some Taiwanese mother board maker.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
