Slashdot Mirror
Intel To Make A Greener Microprocessor
crem_d_genes writes "According to the San Jose Mercury News, Intel is planning microprocessors that have a reduced amount of lead in them (reportedly 95% lower). It's about time a company started this - good job - and let's hope other tech companies take the hint. While many places in the US have banned the disposal of computer parts, there have been unintended consequences of the eco-friendly laws. Many 'recycled' computers currently get shipped overseas where parts eventually make their ways into the hands of workers who usually 'burn' the parts to get rid of plastic and recover small amounts of valuable metals. In the process they are exposed to the toxic compounds that are released. In other cases, lead makes its way into drinking water."
Why does x86 processors consume so much power?
All don't. It's a marchitecture thing, Intel wanted high frequencies no matter what. As a result we have processors which do less work per clock cycle, huge pipelines and high power consumption.
All x86 processors don't have this issue. Via's C3 is miles away from Intel's Pentium 4. AMD is also somewhat better than Intel, and Intel's own Banias (Pentium M) is also rather low power.
The problem is, Intel's been brainwashing the public that YOU WANT A COMPUTER WITH MANY MANY GIGAHERTZ for so long now that the're more or less stuck with high power consumption until they have time to create a whole new architecture.
.: Max Romantschuk
Actually, intel is moving away from measuring chip speed by GHZ. Wired just had this article about it.
Basically, Intel is a couple years behind AMD who is now using numbers like 2300+ to describe chip speed.
I suspect this has more to do with complying with the law than a desire to become more green. EU law requires Pb use to be reduced imminently. If they really wanted to become more green then power consumption would be a more profitable place to start. As would reducing the evil chemicals (and power) used in manufacture.
Seriously, look at the bigger monitor tubes (especially in the EU); they have a radio-dosage sticker certifying the level of beta radiation emitted, usually at the preset acceleration voltage.
Jon.
All the lead is in the monitor glass.
The amount of lead in a base unit is limited to solder and tiny amounts within the ICs.
I suspect this is related to the EU directive.
I work for a European Semiconductor company, and have some involvement in our drive to be lead free, so i know a little about this.
Lead is used in the lead frame of the chip, as the coating to make it solderable, and also in some BGA packages as the balls. Pb is not used in the actual chip manufacture.
There are alternatives to Pb, but normally they require higher temperatures for soldering, which have an impact on the package thermal characteristics and material, which in turn may have some influence on the performance of the chip itself, so these changes have to be handled carefully.
At the moment the US does not have a deadline for phasing out Pb (I think they refused to sign up?) but the EU does, so if Intel wants to sell chips in the EU, or Japan, they have to provide Pb free alternatives.
One person mentioned that this is a small percentage compared to the rest of the Pb in a PC - which comes from the solder mainly, but what you should remember is that the EU directive applies to ALL Pb products, and therefore all circuit boards will be Pb free too.
It's only in the US that you might get a Pb free chip, with no reduction in the ammount of Pb in the rest of the machine.
This is a lot of work for a lot of people. It's not a small change, and all companies have to do this, not just Intel.
The Register has an article with more info.
A flip-chip package currently contains 0.4 grams of lead. A tiny amount compared to that in the solder in a motherboard, let alone a monitor.
The main problem relates to the higher temperatures needed to melt lead free solder. These higher temperatures can stress components and are particuarly worrying in products that have to last 20 years.
IT purchasing is notoriously independent of standards, and it is not just clock speeds. We see jokes such as 500W PC speakers (supplied with a 20VA transformer) and the ubiquitous use of 'X' (4X AGP, 56X CDR).
Standards exist but, apparently, buying a PC is more of an emotial experience than a scientific one!
And if you thought that was boring you obviously havn't read my Journal ;-)
All major solder manufacturers allready have lead free products in place, check out thier websites for exact formulations.
BTW, a lot of chip manufacturers have allready done thier lead free packaging. Intels move is late in the day, which is ironic because they are making hi end high cost chips were gold is often used for bonding and plating rather than the solder used to tin the pins of lowwer cost chips.
The higher costs (which are actually quite small compared to the overall price of finished goods) has meant that blanket legislation is inevitable to change habits in this highly competitive market.
The higher reflow temperatures are an issue as well, but most components can allready withstand these temperatures, especially as higher temperature solders are somtimes used as a production technique.
Believe it or not the legislators did actually check that it was feasible before making the laws!
And if you thought that was boring you obviously havn't read my Journal ;-)
8080 and 8085 are source-compatible with 8086 but as far as i know not binary compatible. the same with 8008.
Conservatism: The fear that somewhere, somehow, someone you think is your inferior is being treated as your equal.
AMD has been working on going lead free for a while (apparently since 2001).
My server
Ever since the more watts = more power advertising race started, people started looking at just watts, not how it was measuered. In the beginning, true RMS watts per channel was the standard. It included at what amount of distortion of a sine wave was permitted such as 1%, 0.1%, 0.005%, etc. into a specified resistive load such as 4 or 8 ohms.
Some smart advertiser found if they take all the channels of a 2 or 4 channel amplifier, ignore low distortion (square wave clipped output is ok) list the power delivered at the peak voltage, not RMS and specify an overload condition of a couple ohms and high supply voltage, then a 12 watt per channel amplifier could be advertised as a 250 watt peak total power amplifier. So would you buy the 25 watt/channel amplifier or the 250 watt amplifier? The truth is the 25 watt amplifier is twice the power (real power) of the 250 watt amplifier. I used to demonstrate this with dummy loads, sine wave generator, and scope in my old shop. The 250 watt amplifier in the demonstration did not have an inverter power supply. It clipped at what ever the supply voltage was. The 25 watt amplifier did have an inverter and clipped at close to 20 volts peak to peak. Customers understood the demonstration and would then seek buying advice. I then pointed out the 25 watt amplifier was fused at 12 amps and the 250 watt amplifier was fused at 8 amps. How can a 250 watt amplifier put out it's rated power when the fuse blows at (12volts X 8 amps = 96 watts) Who wants massive clipping?
Today's chips suffer the same fate. High Megahertz does not fix all the bottlenecks that appear. Parallel instructions do more per clock. This is not measured in Megahertz. Pipelining takes more clocks to get an instruction done. This is not measured in Megahertz. Predictive branching; same thing. In terms of real processing power, the amount of power used per transister per cycle is way down. In the days of the PC and PC XT, the chips did not use fans or heatsinks. There were only a few hundred K transisters clocked at 4.77 - 8 Megahertz. Power dissipation was somewhere in the 5 watt range. Now you are talking several million transistors clocked at several thousands of Megahertz (Gigahertz) or in simple terms, almost 1,000 times faster using only 10-20 times the power and the power is divided by many times the number of transisters. The effeciency in real terms of transistor performance has gone way up. Using the current technology, if you wanted a PC (8088 CPU) it could be clocked at 1 GHZ at under a watt of power. It could be clocked at 4.77 Magahertz at just a few miliwatts. Both would be fanless heatsink-less solutions. Other than display, and drives, a solar power PC XT is possible if you can survive with no USB, Windows, GUI, etc. They don't make them as there is no market big enough to tool up to produce them. Even a cell phone now uses more CPU power than an old XT. (at much less power consumption)
The truth shall set you free!