SPCR

Brian Won's article Ars System Guide special: it's easy being green focuses on reducing power and ignores the fact that 75% of a computer's environmental impact occurs before it ever gets into the hands of the end user.

Those who really want to make a difference will simply not buy new, keeping older gear running as long as possible or buying used gear that's lower power if possible. If you have to buy new, get the most minimalist system you can -- both size and power-wise.

Simple fact: The environmental cost of manufacturing far outweighs minor improvements in power efficiency for desktop computers. IE, if power efficiency reduces electricity consumption during use by 30%, the total environmental cost of the computer will drop by just 7.5%. (30% of 25% is 7.5%) This does nothing to change the upfront environmental cost of manufacturing.

Being green with computers is not easy at all.

See page 2 of Life Cycle Analysis and Eco PC Review for more details and references.

That's all good 'n' stuff, but you still need to be able to play the latest games.

On the other hand, the CPU I bought was used on eBay, and I'm reusing several components from my prior system (the old bits were passed down to family members - my nephew now has my old Opteron 175 and ASRock 939DUAL-SATA2 and my parents have my 6600GT and Scythe Ninja). The only 'new' bits in my system are the i975Xa-YDG the 8800GT and the RAM.

I bet if you polled users on SPCR, you'd find a lot of them indulge in what I call the 'rolling upgrade'. This is where random bits and pieces are carried over from prior configurations as you surrender to the various flavors of Moore's Law. By this logic, my current system is an upgrade (via multiple degrees of separation) from my original 1993-vintage 386DX-25, as there has always been at least one (if not more) component that was carried forward.

Not exactly what you were thinking, I'm sure, but it's a start.

-D

I had the same thought when I first read the article, but then I thought, "what's the real intent of the article?" I think that if you look at it from the point-of-view that the new system is being built PERIOD, why not do what you can to reduce power consumption and still get the performance you need? I don't think anyone in their right mind (key part, there) would go out and buy a brand-new computer (or it's equivalent components) to be more efficient. It's about as self-defeating as driving 20min across town to save $0.05 in gas prices.

My opinions is Brian Won is parroting the cultural identity of "green".

He writes, "The marketing buzz says that a green PC will result in reduced power bills, smaller servers, and lower carbon footprint for the environment." which makes the intent of his article clear: green = less electricity required.
Culturally, green means less of anything that is costly. It incorporates a dual identity of money and environmental impact.
What exactly is reduced depends on the context of where green is used.

Brian Won writes that green "[...] might also allow one less data center to be built, which means one less building to staff and all of the other associated costs". He changes the identity of green to saving money and leaves it at that for the rest of the article.

As MikeC pointed out, Brian Won is ambivalent to the measly impact that a new, 50-100 watt less consuming computer has in promoting a "lower carbon footprint." He's in error in assuming that keeping a "power-hungry Netburst-era Pentium 4" will result in a higher carbon footprint than purchasing a modern computer according to the advice in his article.

Here's a few wacky thoughts:

• - What if motherboards had daughter cards hosting the CPU / RAM / Northbridge so that you could reuse the motherboard?
  - What if video cards had removable GPUs?
  - What if Intel and AMD agreed to not change their socket architecture for 5 years and all revisions of their processors in that time period would be backwards compatible?
  - What if Microsoft/Apple/SUN agreed to ship an OS that required no hardware upgrades to function at a similar level to their prior OSes?

Unfortunately, I don't think any of these are terribly realistic. On the other hand, there is the model of most corporate datacenters - servers are amortized on a 3-year lifecycle. You buy a server in 2007 and it is considered EOL in 2010 when its warranty expires. However, during that three year envelope, you design your architecture (web site, storage, whatever) with that hardware in mind.

-D

For the most part I agree with MikeC's take on the article. Like so much of what seen lately, seems to equate green = low energy consumption, and little consideration of efficiency, lifetime cost, etc. (The article made a few references to these things - but seemed very confused about it.)
"Green" is just another marketing slogan to sell more stuff.

However, this comment I found surprising:

MikeC wrote: If you have to buy new, get the most minimalist system you can -- both size and power-wise.

and I wondered about rationale/evidence?
Is this covered in the book mentioned on the ecopcreview website?

Two things surprise me:
* While small size (at similar density) implies less material, it often translates in computers into proprietary, limited upgradeability, harder to fix, etc.
(Also, since this is SPCR, I thought we were the ones into big fans, big heatsinks, and lots of space for airflow.
i.e. is small size really that important over, for instance, standardized parts?
e.g. I have read opinion pieces about greenness of laptops - but haven't seen evidence to support it (aside from the green = takes less energy to run). Even if they are a poor example (because of excess integration, etc.) What is the cost per pound of PC (e.g.)?

* Why shoot for minimalist in new parts?
What about shooting for the most efficient parts that will have the longest likely lifespan? Don't want bleeding edge, but for applications that require computer power, get something efficient and fairly fast so that:
1) The job will get done quicker, so you can turn the thing off.
2) It will be longer time before you have to get another one.

I keep seeing things that confuse power - low energy systems vs low speed systems. If people get one of these low speed, low energy use boxes, and it does what they need for a long time, great. But I wonder how many of them will get these systems and then get new ones sooner because the performance doesn't keep up.
(In a few years, a fat client may be repurposed as a thin client, but if you start with a thin client, haven't you lost a few years of lifespan vs. starting fat?)

Can you suggest any references on this?

(I guess this is more of academic than practical interest for me, since my newest computer is an Athalon 2000+ which I got second hand.
With all the great used systems out there, my main new purchases anymore are quiet bits.)

Thanks

He's in error in assuming that keeping a "power-hungry Netburst-era Pentium 4" will result in a higher carbon footprint than purchasing a modern computer according to the advice in his article.

is it our fault that China (where all these bits are made) uses planet-destroying coal instead of renewable energy? So I'm supposed to be punished with a crappy computer for the sins of China's energy policy?

MikeC wrote:... ignores the fact that 75% of a computer's environmental impact occurs before it ever gets into the hands of the end user.

Simple fact: The environmental cost of manufacturing far outweighs minor improvements in power efficiency for desktop computers. IE, if power efficiency reduces electricity consumption during use by 30%, the total environmental cost of the computer will drop by just 7.5%. (30% of 25% is 7.5%) This does nothing to change the upfront environmental cost of manufacturing.

Mike, do you know what assumptions go into those figures? Surely it makes a difference if you consider a few scenarios:

1) My parents, whose PC is on for only a few hours a day and is mostly idle even when it is on (maybe 50W * 2hrs = 100W-hrs per day)
2) A folder, whose PC is on 24/7 in a loaded state the entire time (100W * 24 = 2400W-hrs per day)

So a folder could easily use 24 times the energy of my parents. Surely at some point the energy consumption will take over from the stated "75% of impact occurs in manufacturing". Depending on the assumptions used in deriving that 75%, seems to me it may be more beneficial for certain types of users to replace old, inefficient gear with newer efficient parts. If we knew more about those assumptions, we could make the most environmental choice for our usage patterns.

Whenever these kinds of numbers are tossed about, typical PC usage is assumed. Typical PC usage -- and that includes me -- is that the system, when on, idles 90% or more of the time. It's certainly true for me -- and I certainly consider myself a power user. Even check during games -- you'll be surprised, if you can keep a log of AC power draw -- just how variable the power demand is.

Typical PC lifespan is set at 4~5 years, it depends who you ask. Some would even say 2 but that's not realistic for most folks (ie -- non-geekheads like those in this forum.)

Also, environmental costs don't equate just to electricity/energy consumption. There are many different types of raw materials "consumed" and disposed as waste during high tech mfg, and many different toxic wastes, not just CO2. The raw materials to final product weight ratio is something like 30 to 1, irrc. (Think about a CPU. It takes 1.7kg of raw materials to make one, never mind water and energy.)

If you're running something like folding, for sure, your energy consumption will be higher, and maybe, just maybe, if you replaced your PC with something that consumes just half the energy, in the long run, you might offset the environmental cost of manufacturing your new PC in say 5 years.

In the meanwhile, your old PC has to be disposed of...

Then finally, I have to ask what good you think you're really doing with folding or other types of distributed computing projects. Just how valuable is "your contribution" and who does it really benefit? I asked myself that some years ago, did a lot of research, and in the end, stopped all of them. The benefits seemed obscure and unassured, and if/when they were going to happen, in the end, would tend to enrich corporations whose shares are mostly owned by the rich and already privileged.

very nicely put, Mike.

mcoleg wrote:very nicely put, Mike.

Seconded, thanks for the response. I am interested in green stuff, but probably not enough (yet) to track down the paper and analyze their methods.

Interesting thoughts on folding. I am not a folder, but I do run SETI@home, which is probably even more of a pipe-dream in actually helping anyone. I may have to re-examine this.

Hello,

To play devil's advocate here: if most of the energy input goes into building the computer, then it could be argued that NOT using it for something like Folding@Home would be a waste. Sure, the results of F@H are unknowable -- but if they do come up with a treatment for Alzheimer's or cancer; that would be priceless. Particularly, running Folding@Home during the winter/heating season would seem to have a pretty low overall cost.

NeilBlanchard wrote:Particularly, running Folding@Home during the winter/heating season would seem to have a pretty low overall cost.

Yep, since you'll be paying to heat your place, anyway. Might as well get some work done while doing it!

Considering I'm the person who wrote the article-- I do agree, the easiest way to be green, as the article clearly states:
http://arstechnica.com/guides/buyer/gui ... reen.ars/4

The environmental impact from manufacturing all of these new components is probably greater than reusing a bunch of two-year-old ones, but who's making sense anyway?

As a System Guide article, the goal of it wasn't to encourage NOT building a computer, but to encourage and spec out a more efficient box that used RoHS friendly, hopefully more recyclable components. Naturally, with PCBs and other items-- that's not the most practical. I absolutely agree, it doesn't make sense to build a new box just to save electricity or so you can send the old case to the metal recycler...

Also, environmental costs don't equate just to electricity/energy consumption. There are many different types of raw materials "consumed" and disposed as waste during high tech mfg, and many different toxic wastes, not just CO2. T

Also very true. Much like buying a new car vs. keeping an older beater around, environmentally speaking, you're generally better off keeping the beater around given equivalent fuel economy or even somewhat better fuel economy in the new car.

... ignores the fact that 75% of a computer's environmental impact occurs before it ever gets into the hands of the end user.

Again, it wasn't ignored-- but it was given the short shrift, and that was entirely deliberate. Calculating the impact on that would have been much more difficult-- and again, not the point of the guide. The System Guides on Ars are to help our readers spec out and build new computers, not to help them NOT build a new computer.

"what's the real intent of the article?" I think that if you look at it from the point-of-view that the new system is being built PERIOD

That was the management's intent, at least. I've argued the difficulties in doing a a green or a low-noise guide on Ars several times due to the difficulties in scope that such a guide represents, and short of what would easily result in a many, many page proper life cycle impact report... I personally prefer to leave such things to EcoPCReview.