Filed under: Analysis / Opinion, Apple
Debunking ZDnet on Intel and power consumption
Earlier this week, ZDnet blogger Paul Murphy posted a full-bore critique of Apple's move from PPC to Intel processors from the perspective of power consumption. By his calculations, the shift to the new processor architecture added hundreds of gigawatts to the energy cost of the Mac population, and thousands of tons of carbon to the atmosphere every year. His tongue-clucking extends to Apple's most visible environmental activist, board member Al Gore, who he says "not only voted for the MacTel switch, but actively campaigned on Intel's behalf prior to the vote," thereby adding pollution to the air while "hurt[ing] America's economic diversity" by cutting IBM out of the Mac processor market.If your response to this is "Wha?!? Everyone knows that the Intel switch was about LESS power consumption per cycle," well, apparently, everyone but Mr. Murphy. Over at Roughly Drafted, there's a precise and scathing debunking of these bogus statistics and correspondingly off-the-wall conclusions. The core points: the numbers for Intel power consumption are off track for the actual Mac configurations; the PowerPC low chip power figures are for the embedded-system versions (not the G4 and G5 that Apple used); CPU power consumption doesn't contribute nearly as much as, say, CRT power usage (which Apple replaced with low-power flat displays); and, the PowerPC platform is doing fine without Apple as a customer, thanks very much. To sum up, every new Intel Mac uses less power than the older Mac (or, dare we say it, vintage PC) it replaces, and Murphy's carbon calculations are full of hot air.
![TUAW [Cafepress]](http://www.blogsmithmedia.com/www.tuaw.com/media/tuaw-cafepress-promo.png)
Reader Comments (Page 1 of 1)
mongul said 11:32PM on 5-13-2007
Paul Murphy is at it since the very beginning, he's entitled to his opinion etc but he's so biased it hurts (mainly his credibility!). Just ignore him.
The PowerPC architecture is a just fine, though.
Reply
Wheels said 11:38PM on 5-13-2007
I hope those who bash CRT power consumption aren't doing so with plasma TV screens hanging on their walls, especially since you can run two CRT TVs with the wattage of one plasma TV with wattage to spare. Actually, a well designed CRT monitor or TV beats comparable LCD screens in power consumption.
The real reasons why Apple phased out CRTs are because of the size and weight of a CRT vs. LCD. This reduces their carbon foot print because you can ship more monitors using the same amount of fuels. Also, a CRT uses metals and other pollutants that are not easily disposed of. So, really, this green move has very little to do with power consumption on the consumer level.
Reply
Todd said 12:06AM on 5-14-2007
"Everyone knows that the Intel switch was about LESS power consumption per cycle" No. The comparison Jobs used was "performance per watt", i.e. the Intel chips use more power, but their performance is enough to justify it. To get the same performance, you would have to use multiple PowerPC chips which would add up to more power.
The PowerPC chips did use less power than the Intel replacements (well, the G4s at least.. the G5s were IBM chips with different design parameters). My PowerBook G4 used less power than my MacBook Pro. The CPU on my old G4 Cube ran at about 10W. That lower power usage design requirement was one reason the PowerPC lagged behind the Intel offerings in horsepower.
Let's compare two systems I run today:
Mac Mini G4 1.42Ghz. Power Consumption: 20W, at the high end of G4 power. ( http://www.gridtoday.com/04/0301/102774.html )
Mac Mini 1.66GHz Core Duo: Power Consumption: 31W ( http://users.erols.com/chare/elec.htm )
Yes, I am quite happy with x86 Macs, the performance is very good. But, the power usage has gone up. The PowerPC chips are excellent low power options. They now have good dual core versions, and lower power G5s that Apple never took advantage of.
Reply
Jeffrey Bergier said 12:25AM on 5-14-2007
it also is worth mentioning that the new intel chips do use more power than the g4. this the need for the increased battery and powersupply size for the macbook pro. however, they are much faster than the g4's. however, the power consumption is something easily measured, while preformance is something that is quite difficult to measure. so who really knows if the intel ones are more efficient. i think all-in-all it was a bad move for apple because now it is impossible for a mac to be faster than a PC. IMPOSSIBLE. before it was most likely not true, but at least possible. now a mac will always be as fast or slower. apple will always take a few months after a new chip is released to get it into their computers. it also means far more often updates from apple, which always upset current owners.
Reply
Tom said 1:36AM on 5-14-2007
That's pretty ridiculous to criticize Apple for switching to Intel, especially without criticizing other PC manufacturers for NOT switching to PowerPC.
Reply
mongul said 1:32AM on 5-14-2007
Paul Murphy: "Core Duo architecture processors - ranging in power demand from a claimed 31 watts for the low end of the laptop line to well over 180 watts at the high end of the desktop line. […] A bit less than half of Apple's Macintosh sales are desktops, the rest laptops."
Apple is using mobile chips in every Mac (Merom but also the older Yonah in the Mac mini), except for the Mac Pro.
Core 2 Duo: design thermal power up to 35W at 2.4 GHz.
Source:
download.intel.com/design/mobile/datashts/31674501.pdf
Xeon 5300 (8-cores Mac Pro): design thermal power up to 120W at 2.66 GHz, the 3.0 GHz part is not listed.
Source:
download.intel.com/products/processor/xeon/dc53kprodbrief.pdf
Xeon 5100 (other Mac Pro): design thermal power up to 80W at 3.00 GHz.
Source:
download.intel.com/products/processor/xeon/dc51kprodbrief.pdf
According to Freescale's marketing blurb "The MPC8641D processor integrates two e600 cores, each scaling up to 1.5 GHz […] -all at just 15 watts of power dissipation." But the MPC8641D processor is not using 15W at 1.5 GHz. Freescale is still in the same ~30W thermal envelope. Pretty good but not unheard of.
MPC8641D - dual core - 90nm
Typical (max is higher): 23.9W at 1.33 GHz, 32.1W at 1.5 GHz.
Source:
www.freescale.com/files/32bit/doc/fact_sheet/MPC8641DDLCRFS.pdf?fpsp=1
Apple used the 7447A in its notebooks before the switch, the 7448 is a 90nm die shrink:
7448 - single core
Typical: 25.6W, Maximum: 29.8W at 1.7 GHz.
Source:
www.freescale.com/files/32bit/doc/data_sheet/MPC7448ECS02AD.pdf?fpsp=1
7447A - single core
Typical: 21.0W, Maximum: 30.0W at 1.42 GHz.
Source:
www.freescale.com/files/32bit/doc/data_sheet/MPC7447AEC.pdf?fpsp=1
A quick look at PA-Semi:
PA6T - dual core - 64-bit - 65nm:
Typical: 13W, Maximum: 25W at 2.0 GHz.
Source:
www.pasemi.com/downloads/PA_Semi_ISSCC_2007.pdf
PWRficient PA6T-1682M projected power dissipation:
Typical: 17W, Maximum: 25W at 2.0 GHz.
Source:
www.pasemi.com/downloads/PA_Semi_HotChips_18_Presentation.pdf
Reply
michel said 2:27AM on 5-14-2007
@ Tom : you have to understand. Apple should save the world, you see ?
Save the macintosh, Save the world !
Reply
JeffDM said 8:02AM on 5-14-2007
I really can't comment on the notebooks, though I remember that long time Mac users were saying that their PowerBooks ran hotter than their MBPs.
Wheels; I'm not buying the assertion that a CRT is more power efficient than an LCD, or that the power to run a plasma can run two CRTs unless _very_ differently sized screens are compared. My 21" CRT runs at about 175W, and the 30" Apple display is rated at 150W. I haven't run a Kill-A-Watt yet, but one 30" has about twice the screen area so you'd have to compare one 30" to two 21" CRTs.
Reply
digitalintrigue said 11:30AM on 5-14-2007
"bogus statistics and correspondingly off-the-wall conclusions." --->this must refer to 'An Inconvenient Truth.'
Reply
Wheels said 11:53AM on 5-14-2007
JeffDM, check this out.
http://reviews.cnet.com/4520-6475_7-6400401-2.html
And for those of you who don't want to click above, here's the raw synopsis:
Power consumption compared
TVs:
Average plasma: 328 watts
Average rear-projection: 208 watts
Average LCD: 193 watts
Average CRT: 146 watts
I too was kind of shocked that CRTs edged out LCDs (by 23% though, that's significant), but I know the plasma vs. CRT wattage numbers to be accurate. I have a friend who had worked for a power company and he was shocked when he checked the power consumption of plasma TVs, since he was interested in buying one. He didn't buy one based on their power hog status.
Reply
Michael Rose said 12:36PM on 5-14-2007
Wheels, check the rest of that Cnet article... they compare on a per-area basis, which is relevant even though plasmas and LCDs tend to be larger than the CRTs they replace:
----
* Microdisplay rear projector: 0.14 watt per square inch
* LCD: 0.29 watt per square inch
* Plasma: 0.34 watt per square inch
* CRT: 0.34 watt per square inch
If power efficiency is all you're after, the clear choice is rear-projection technology. Of the four, plasma screens are generally the most power hungry, but that's more because of their size--on a square-inch basis, they are roughly equivalent to a large CRT set. Flat-panel LCDs often have a good brightness-to-consumption ratio, but they're not exactly consistent. Some LCDs are as low as 0.11 watt-per-square-inch, but some go as high as 0.37 watt. Of course, there's always exceptions, the most noteworthy being Sharp's 65-inch LCD that pulled down an amazing 583 watts in use and 76 watts even when it was "off." Luckily, more and more new TVs are coming with a power-saver mode, which we've found can drastically cut power consumption.
Reply
Michael Rose said 12:39PM on 5-14-2007
I left out my actual point, which is: as Roughly Drafted says, the CPU is not the most power-hungry part of the system, the display (usually) is. Challenging Apple's carbon profile based on the Intel switch is like blaming Shell for not putting solar panels on the gas pumps. It's not the most significant bit.
#4 -- Jeffrey, the relative performance story for the Mac vs. Windows isn't what's important about the Intel switch. It's the virtualization/Boot Camp "permission to switch" that makes the difference in Apple's CPU sales. Windows users now have confidence that they can still run that one application they need while getting the best of both worlds.
Reply
Stephen Lang said 2:12PM on 5-14-2007
The ZDNet author seems to be making his statements based on press statements about CPU's that promise a lot but have proven little in the real world. There are a lot of smaller companies that offer seemingly-exotic solutions that end up not passing muster in the real world (at least for mainstream applications such as notebook computers.)
And even though the MacBook uses more power than say the last gen iBook, it is MUCH more powerful. The main basis for Apple's move to Intel was power per watt for the notebooks, as IBM was failing to provide any sort of workable notebook G5.
Next thing you know, he's going to argue that Apple shoulve have used Transmeta CPU's...
Reply