To Save Energy Or Not, That Is The Question.

“By all indications, by the end of this year the California Energy Commission will adopt energy efficiency guidelines for computers, becoming the first state in the nation to do so.
 
The agency estimates it will add about $18 to price of a computer but promises it will save customers and businesses much more in energy savings.”
 
See California poised to set regulations on computers
I’ve long proposed using smaller/cheaper/thinner desktop PCs but I’m not so sure this can properly be legislated. Users and uses of PCs and servers are just too diverse. It just makes no sense to require <X watts. Some folks actually crunch large volumes of data or use immense local storage. Heck, I need double the watts for a display simply because my eyes are old and never worked that well in the beginning…

What I think may not matter. California drove Sierra Bullets out of the state because the cost of doing business became too large to be competitive. Later they worked to ban all lead from bullets… California has regulated automobiles and even lawn-mowers nearly to death. My late “complies with California” rototiller always ran at top RPM because some legislation required little or no idling. That machine broke twice in two years… An engine and transmission needs to warm up some before spacings and viscosities reach the right levels to take on full load.

Computers may need to idle to keep files available for the network or because the user just can’t wait for rebooting. I just don’t see how regulations could be practical for PCs. You could try setting the maximum age, for instance, but inadvertently increase junk piles/waste of material instead of energy. Repurposing older computers for lighter loads is a good thing but it saves no energy except the energy needed for manufacture. I can see many ways this plan could really go off the rails. What if it just displaces certain jobs out of state?

Unless the state is going to run all IT, this just can’t work. Whatever throttle, limit or setting California requires will either have some means of circumvention or be counter-productive. My late roto-tiller had zero adjustments the user could make to control the engine except starting and stopping. Whatever they do for PCs is likely to wreck someone’s IT. Suppose they require thin clients, the most efficient PCs around. That would shut down all desktops that need local data or serious audio-visual throughput. I’d bet Hollywood would outsource a lot of jobs if they did that.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

28 Responses to To Save Energy Or Not, That Is The Question.

  1. oiaohm says:

    Robert Pogson the highest torque and highest speed Top Fuel dragsters are ring less and oil less using gas lubrication.

    A lot of model engines as used in RC cars, boats, planes, are nearly all ring less.
    dougman there are some ceramic engines for model helicopters that are ring less and oil less using gas lubrication.

    Gas lubrication engine design scales from small to huge. Across all torque ranges. The universal bug has been a fairly high idle requirement to maintain gas lubrication in tested designs is about 25 percent of top RPM speed. The big thing is a ringless using gas lubrication can go from zero to max rotation speed fairly much as fast as the burning fuel can do and do no damage to itself. Ringed engine doing exactly the same thing can bring out some very bad issues.

    Robert Pogson you are right a gas lubricated engine is not for direct drive low speed high torque usages with most of the current designs some of the newer designs are starting to show promise for this but durability testing on those is not complete. If you want low speed high torque out gas lubricated engine design that is well tested its use a gearbox to gear it down.

    Between gas lubricated and oil lubricated there is almost nothing in common if you compare two machines design to do the same task using the both engine types. Both engines types can do each others job to a point but they have to be built totally differently. Oil lubricated cannot do massively high RPMs it just does not work and making gas lubricated tested designs work at low RPMs causes some serous gear box issues reason why there is a lot of work to design new piston designs for gas lubrication to make it work at lower RPMs to make gearbox for low RPM work simpler. The piston work with gas lubrication is starting to pay off different people are starting to find ways to make it work. Remember gas lubrication only starts in 1988 its still a fairly young tech and are still working out how to make it perform the best.

    Efficiencies with gas lubricated vs oil lubricated turns out the only reason why gas lubricated loses is high mandated idle that may be fixed by new piston designs. Both engines running at same speed putting out the same amount of torgue both use either same amount of fuel or the oil lubricated slightly more fuel. Oil lubricated mandates rings and there have be many designs of rings over the years and 5 common ones have lived out today and gas lubricated is not old enough to have gone though this level of refinement yet.

    CA only brings in the no engine idling regulation in 2008 after the gas lubricated engines had almost completed 20 years of testing and absolutely showing no major problems. So idea of a no idle engine was valid at that time. No idle engine being oil based lubrication just does not work yet the regulation did not forbid this combination so it causes more problems than it worth. CA regulation objective was to speed up gas lubrication engine development it has not work.

    Now if someone regulates against computers idling the rule better mandate cpu/gpu/ram.. hold state without power or it going to be equal mess as CA no idle regulation.

    The difference at times between good regulation and bad regulation of a market is one point.

    Most people who though brick bats at the CA rules against no idle have no clue what its objectives were.

    Here is something bad. Large percentage of black smoke/soot out of trucks and the like is not the burning fuel it self it but what happens when a small amount oil burns with the fuel disrupting the burning process. Best way to solve this is get oil out of the burning areas of the engine. Ultra clean burning engines turbine or piston are all gas lubricated only thing possibly cleaner is fuel cell. Most likely time for oil to mix with fuel due to sealing issues is when the engine is idling. Strange right turns out gas lubrication theory does work inside a oil lubricated engine when it gets up to higher rpms so preventing the oil fuel mixing. So this is basically bugger to fix oil lubricated engine design issues will in time mandate adding gas lubricated features at that point why stay oil lubricated.

    In fact, I think the gain from ring less pistons lower friction would be lost in engine efficiency. As the seal width has little effect on moving friction, the higher the pressure the greater the friction, a piston ring that has circumferential oil grooves in the surface will have lower friction and better gas sealing capabilities, gas at high pressure will leek through a .0002” clearance which is less than the oil film thickness required for friction reduction that a piston must have to guide it in a bore.
    dougman this is the make people make over and over again looking at ring less pistons.

    Key to understand how a well designed ring less piston using gas lubrication works is understanding.
    https://en.wikipedia.org/wiki/Tesla_valve
    The Tesla valve. A tesla valve effectiveness increases as pressure increases. Basically the exact opposite way to piston rings and oil. Some proto type ring less pistons have worked with 5 mm open clearances all around the piston. So when the piston is running up and down there is a 5 mm gap between it and the piston wall and at enough speed of motion maintaining a seal equal and exceed to as if a normal piston with ring was filling out the complete bore because its gas sealing gas. It was one of the demo items made showing how current theories had it wrong. Apply all the theories of ring and oil based piston design to a 5 mm gap piston there is no way it can work but it does. Of course apply the theories of gas lubricated design and it says yes that going to work perfectly.

    Yes normal rings you worry about small area pressure leaks because this can light the oil up and do other bad things. Gas lubricated there is no oil to light up so a little leak is not panic and the fact the faster they are running the better they seal. With injection leaking in the low pressure parts of the cycle is normally not a problem with gas lubricated and that is where they most leak. Increasing pressure the more the seal improves.

    The sad reality is at higher rpms piston rings give worst sealing than using ringless designs. Piston rings give you fiction and extra wear and don’t perform as good at high rpms.

    Dougman what you said is why gas lucubrating designs were laugh at first until someone when and spent the time doing up the full theory and people started in fact testing out how they performed only to find out that ring and oil designs have very limited advantages. Issue here we don’t think of gas as sealing material. Gas lucubrating is don’t fight the gas pressure use the gas pressure to seal and does it matter if it leaks a little when the pressure is low(what is mostly no in injection based engines).

  2. dougman says:

    “not cultivating my clay”

    Better get working with that tractor! Another month and you’ll have snow flurries.

    For me, I am looking at a BCS 853 with a power-harrow attachment. To build my beds with next year.

  3. dougman wrote, “A lot of model engines as used in RC cars, boats, planes, are nearly all ring less.”

    Those are instances of high RPM, small size and very little torque, not cultivating my clay. In such situations, the friction of piston-rings may indeed be significant but apart from whipper-snippers I don’t see any use for such things here.

  4. dougman says:

    “There are no such engines here. Why discuss some theoretical concept”

    A lot of model engines as used in RC cars, boats, planes, are nearly all ring less.

    In fact, I think the gain from ring less pistons lower friction would be lost in engine efficiency. As the seal width has little effect on moving friction, the higher the pressure the greater the friction, a piston ring that has circumferential oil grooves in the surface will have lower friction and better gas sealing capabilities, gas at high pressure will leek through a .0002” clearance which is less than the oil film thickness required for friction reduction that a piston must have to guide it in a bore.

  5. oiaohm says:

    On what planet are you living? There are no such engines here. Why discuss some theoretical concept when every one of my neighbours is using the same old tech that has proven itself very well over many decades? I’ve had such machines last over 20 years. What’s the track record of your concept?
    One is not my concept. Two is not really new tech. Three is not theoretical any more.
    http://tribology.asmedigitalcollection.asme.org/article.aspx?articleid=1459383
    Theory was 1988. First produced engines 1989 and all of those still going. This is a complete different lubrication and sealing system based around gas for a combustion engine. All the engines of this type are ring less and oil less.

    Of course there is a reasons why you don’t see this class of engine using gas lubrication. Steal bore with steal pistons that are cheap to make will not cut it in a gas based system. Ceramic or Ceramic coated bores/pistons are basic requirements of gas lubrication engines to be strong enough not to burn out.

    https://books.google.com.au/books?id=iJ6P4Q6UCPIC&pg=PA61#v=onepage&q&f=false
    2009 book “Modern Diesel Technology: Diesel Engines” in fact mentions their existence.

    So Robert what planet are you living on the stone age. Engine technology has advanced massively passed being only ring based engine designs. Ring based engine design remain popular due to being made from cheaper materials and has a broader range of RPM of operation. Ring less perform better the higher RPM not a good idea to attempt to run them at low RPMs. So you run ring less flat out or stop them completely.

    Ring less advantages.
    No oil requirements.
    Able to take straight to max load no warm up.
    Ring less disadvantages.
    Performs best at top speed. Lubrication and sealing issues can appear if attempting to run at lower speeds than max.

    Ring less engines are truly no idle engines the design does not allow for an idle. Yes attempting to make a ring less engine idle like a ringed engine is how to break it. For an emergency generator ring less can be a super good match(less things to worry about)

    Yes Robert us who have worked around all different types of engines have seen ring less versions from time to time.

    Ring less engines have decades of tested functionality behind them now.

    Both ringed and ring less engines have there places. A machine you are wanting instant power and higher end speeds ring less is great. A machine you are after means to use lower speeds and will put up with warming up ringed is better.

    Yes something people are not commonly aware of that there are two camps of engines one with piston rings and one without piston rings.

    http://www.motoring.com.au/aussie-invention-eliminates-piston-rings-40773/
    This new design is about allowing a ring less to-do the RPM ranges where ringed have been good.

    Robert just because you have not seen something does not mean it does not exist.

  6. oiaohm wrote, “Cannot happen in ring less engines because ring less engines are oil less engines in the crankcase and cylinders.”

    On what planet are you living? There are no such engines here. Why discuss some theoretical concept when every one of my neighbours is using the same old tech that has proven itself very well over many decades? I’ve had such machines last over 20 years. What’s the track record of your concept?

  7. oiaohm says:

    We’re talking about small Briggs & Stratton internal combustion engines
    If you read the design those are designed to have a idle it in fact says so even in there manuals.

    Ring less pistons are based on Tesla valve.
    BTW, on the intake stroke, oil from the crankcase will reach the cylinder making smoke and really gumming up the works.
    Cannot happen in ring less engines because ring less engines are oil less engines in the crankcase and cylinders. The gas exhaust gas is the lubrication instead of oil. Try operating starting a generator on Antarctica that if it does not start you will die. −57 °C (−70.6 °F) is normal temperature on Antarctica not even classed as that cold.

    Without piston rings you’ll likely get considerable piston to cylinder contact with little lubrication for several minutes.
    The Tesla valve based piston designs you will get no cylinder wall contact after the first stroke every single time no matter the temperature of the engine. One of the first marking of a oil less and idle less engines designs is no piston rings as they are totally not required because the cylinder rides on a cushion of exhaust gas. So a completely different way of lubricating the engine. The exhaust gas method instantly lubricates the engine on first stroke.

    Robert serous-ally if engine uses lubricant oil in the cylinders and crank case it need to warm up if engine is oil less design and use exhaust gas for lubricant they are the true no idle engines. So the regulation demand no idle but did not demand using no idle engine designs that include no oil was the mistake. For a long time it was thought the little bit of leaking past exhaust gas in Tesla valve based pistons would hurt efficiency turns out it does not. Little bit of leak loss force in the Telsa valve is less than the force of friction even oiled piston rings generate. Yes reduced inputs since the engine has no oil and increased performance that is exactly what the numbers on the ring less engines say. Cost is the pistons require a stronger surface all over nothing that ceramic coating does no achieve. Ringless pistons don’t have the same tolerance requirements either so more possible to make out of pure ceramic.

    That’s surprising. I would expect about 20% reduction. I guess chips are sufficiently complex that there are extremely non-linear things happening, like overlapping gates drawing spikes of power unnecessarily. That would be what oiaohm calls overclocking.
    Yes the is silicon over-clocking/over loading. To be correct 10% change in clock speed is not it. There is a critical point with all silicon. A 1hz change in something clocked in the 2Ghz+ range can be doubling of power usage if you cross the critical point were more area of the silicon goes into overloaded. Now if you have not crossed the critical point power will be fairly linear to speed. The difference between being on the right side of the watt per area to being on the wrong side is quite minor.

    What causes the massive spike is what is called leakage. The resistive and conductive material in the silicon chip basically starts acting as a heating element once it hits a particular point of watt vs area losing more and more electricity into heat and resistance. So this is not that much to deal with complexity of design but basic property of silicon. Its like recently TSMC found that your silicon power rails should be larger than everything else most of the reason why most old silicon chips designs run as hot as they do is the power rails are always exceeding the watt vs area ratio that silicon can safely handle so turning power into heat. Odroid-C2 CPU has the mistake but area effected should be minor enough if it was correct it operating would at-least half if not a quarter. Fixing this fault will cost time remaking the plate that the Odroid-C2 CPU parts are made with.

    This is the reason to mandate watt vs area. Bad silicon designs take more power than they should because someone screwed up when the production plates were made and did not make power rails and other thing the correct width for the load they would be expected to take. Add in over-clocking bad silicon designs just makes matters even worse because more and more area of the chip just becomes a heat producing resistor because there is not enough conductive width to move the power around. Watt vs area is basically a physical constant value in silicon chips for where it goes from working effectively to performing like crap. Anywhere in a silicon chips design that is wrong to the watt vs area ratio rule will result in chip eating more power than it should. Yes just power rails wrong is over double the correct power usage.

    Watt vs area wrong in silicon does effect how much power the chip even uses in sleep.

    The old Atom and VIA CPUs with fans are idling at similar temperatures but often exceed 60C.
    Power rails in all those are completely wrong along with other parts of their silicon designs. Robert that heat is not coming from no where it a clear sign of a problem. Its stupid on PCB circuit board connections have different widths on the amount of power expect to be moved but inside silicon chip people tried to get away with exactly the same width for everything as that is simpler to compute placement for performance. Basically lets forget power effectiveness and lifespan as long as we get performance has happened. Of course in most cases same performance can be achieved with correct track widths inside the silicon yet now with out the heat production and without the shorter life caused by that.

    So yes complexity comes into it mostly that when silicon design have been made not enough processing has been done to work out what width each track inside a silicon chip should be and once you do it does make calculating length to keep different interconnects the same length harder. Power effective and Performance effective at the same time in silicon costs a lot more time in the designing stage. Key point at design stage so restricting allowed watt to area on silicon should do absolutely nothing to the top performance we can achieve only difference to stay inside watt vs area will require a little more silicon at times to achieve it not to overload anywhere. Correct watt vs area should see 30+ year life spans out of silicon in 24/7 usage.

    Silicon is having a short life at times because it basically working like a slow blow fuse due to watt vs area being exceed.

    The energy consumption of a hard drive is mostly the cost of spinning at such and such RPM in a viscous medium, air or helium or whatever… It has nothing to do with resolution.
    This is wrong. In a idle system most of the energy consumption of a hard drive is the controller board in other words the silicon. Have you not noticed most hard drives stop spinning when they have nothing to read/write for a while. Solid state hard-drives it is purely the silicon.
    ST6000VN0001 for example it sleeping mode is 0.6watt. Running mode is 9watt and idle mode is 7watt. When a computer system is idling it works out to something warped about 90 percent of the time drive sleeping/standby about 1 percent of the time drive running and about 9 percent of the time drive sitting in idle mode so lets multiply that out. 54watt sleep/standby in silicon with some reads that are data coming from drive cache leaving drive. 63watts in idle mode and 9 watts in active work. Surprising right it can work out in some drives in idling computer usage that 60 percent of the power usage is silicon sea-gates are fairly effective controllers made using samsung production tech but not all brands use quality. In idling systems the drive sleep/stand by usage is quite a large slice of overall power usage.

    I will change from 3 hard drives being active to 7, perhaps an increase of 70W.
    45nm on Beast’s CPU to 28nm on AMD A1120 in fact covers 4 more active drives in fact the number is 5. Please stop attempting to guess your way out of this. Specifications of hard-drives spinning discs cap power draw at 10 watts these days. Solid states specifications cap them at 6 watts. Yes some of the reason to get rid of some older hard-drives they are not power effective. In fact any harddrive taking more than 10 watts exceeds SAS and SATA standards for power allowed per device.

    Now lets presume most of the drives are standby/sleeping that is common in large drive arrays this is ~0.6 watt for spinning media to ~0.7 for soild state. Take of 2 drives as active so subtract 20 watts out of the 50 that leaves 30 watts so 42 drives sleep/standby and 2 active in 50 watts. Ever 10watts extra power usage is equal to a full hard drive. Do notice the solid states higher standby usage.

    I would have to go to 7nm to make up the difference.
    So you don’t have a clue. From 28nm to 14nm is 1/4 power requirement quite a steep drop in fact mostly because 14nm designs fix up the power rail mistake.

    A consumer can increase cooling with water or a bigger heatsink/cooler.
    Ever head the term silicone degradation adding cooling does not stop this. The fact cpu need coolers is the sign of the problem. Watt vs area limit exists on silicon because its basically a slow blow fuse. Exactly like a fuse if you stay under it limits it lasts insane amount of time. Yes if you put cooling on a fuse it takes longer to blow. Yes silicon is behaving exactly like a fuse. If you were having to run a fan in the fuse box of your house so the fuses did not blow you would think something was massively wrong but having to run a fan on a cpu raises no questions. They are both forms of fuses yet we don’t apply the same rules.

    As ram said is not a few watts different. Its min of a double more.

    Requiring CPUs with such and such resolution while allowing huge monitors, tons of RAM and multiple huge hard drives is just not relevant. It is the total system consumption that matters not some particular element.
    Take this in this time. I said silicon rule of watt vs area and nm. So this is all silicon chips inside computer ram, harddrive cpus the lot. Of course power regulators in power supply and places would be allowed large nm but have to stick to exactly the same watt vs area rule so they are not going to burn out like a fuse. If a design needs a fuse use a fuse not use a bit of silicon todo the job.

    Remember your old 45nm system would only be minor-ally over driven. Inside a computer 50watts of power saving is quite massive particularly when you remember hard drive specification states harddrives cannot use more than 10 watts each. Yes the one AMD video card eating up 300watts is equal to 30+ harddrives going flat stack it truly + because no harddrive is built to use all the 10 watt limit.

    Really once you start thinking of power usage inside a computer in terms of hard drives it kinda comes clear how bad some areas are. Like the most effective skylake process from intel only consumes 15watts at full load worst is about 90 odd. 1.5-9 harddrives worth of power does not sound too bad. Ram sticks between 2-5watts each lot of people have at least 2 ram sticks so about a 0.5-1 hard-drives worth of power there.

    GPU is fairly much your worst nightmare 90 to 400 watts depending on construction. You want to bring big power savings to computers regulate GPUs. Identically performing gpus one made right and one made wrong can over halve the power usage of a system doing exactly the same thing.

  8. ram wrote, ” I’ve found downclocking just a modest percentage (say around 10%) can cut power consumption to half and extend the life of the device to decades.”

    That’s surprising. I would expect about 20% reduction. I guess chips are sufficiently complex that there are extremely non-linear things happening, like overlapping gates drawing spikes of power unnecessarily. That would be what oiaohm calls overclocking.

    I just checked. With my Odroid-C2 idling there are 156 processes running, load average is about 0.1 and CPU is used just ~1%. Playing a full-screen 720p video gets to 50% and one can interact with the system very crisply at that load. That’s nearly perfect for a desktop client. You can’t beat the price/performance. Sensors show a temperature of 52C in the tiny blue plastic case. It gets to 60C under heavy load. I love it. Silicon should last indefinitely at such temperatures. The old Atom and VIA CPUs with fans are idling at similar temperatures but often exceed 60C.

  9. ram says:

    “…A consumer can adjust the clockspeed of a CPU either up or down on some motherboards…”

    Yes, they can! I’ve found downclocking just a modest percentage (say around 10%) can cut power consumption to half and extend the life of the device to decades. The performance difference is hardly measurable since other factors such as memory speed and the speed of other chips (usually with a fixed clock rate) tends to dominate the overall performance equation anyway.

  10. oiaohm referred to an article:“While conventional piston rings form an effective seal, the friction they generate absorbs some of the kinetic energy the engine is trying to create and turns it into unwanted heat energy.”

    You do realize that the friction of oily piston-rings is tiny compared to basic thermodynamic efficiency, heavily loaded crankcase/rods and exhaust/intake pressures? Replacing piston-rings with clearances might save some on costs of manufacture but very little on waste. BTW, on the intake stroke, oil from the crankcase will reach the cylinder making smoke and really gumming up the works. Piston-rings are not only sealing gases but also lubricating oil while providing a bearing surface and centring the piston. Expect shorter life of engines and lower efficiency and more pollution… Yes, we must have that… Consider starting such an engine here in Canada at -40C. How thick is that oil? Without piston rings you’ll likely get considerable piston to cylinder contact with little lubrication for several minutes. Doing so at maximum RPM for very long is a recipe for disaster. Expect galling and short life. Piston-rings have many functions. Eliminating them to get slightly lower friction is throwing the baby out with the bath-water.

  11. oiaohm rambled on… “There is a place for regulation. Watt vs area regulation on silicon reduced the risk of consumers being on the end of lemon products due to overheating and other nasty things. So that one is 100 percent for consumer.”

    Government has no business regulating nm or areas of chips. Consumers largely don’t care about either of these issues. A consumer can adjust the clockspeed of a CPU either up or down on some motherboards. A consumer can increase cooling with water or a bigger heatsink/cooler. Requiring such and such a nm or area of chip has no effect on the desired reduction in energy-consumption. It’s just silly irrational behaviour, government being seen to be doing something. Allowing luxury cars on the road while banning CPUs with a few watts more consumption is idiotic. Requiring CPUs with such and such resolution while allowing huge monitors, tons of RAM and multiple huge hard drives is just not relevant. It is the total system consumption that matters not some particular element.

    oiaohm trying to be really rude and ignorant, wrote, “How many hard drives are the grand inquisitors going to allow?
    This is you just being stupid. Nothing I have said contains any about physical size of system. Why that is not enforceable with people building their own systems so by that al-own not worth considering.”

    Do the maths, nitwit! If I regulate a CPU with 14nm resolution and a few mm2 you can have a CPU that runs on a small number of watts but if you then allow 47 hard drives internal and via USB and Ethernet from a NAS, there is ZERO net energy savings to be had. The tiny saving on the CPU might compensate for the first few hard drives. Then the power bill just keeps rising. The energy consumption of a hard drive is mostly the cost of spinning at such and such RPM in a viscous medium, air or helium or whatever… It has nothing to do with resolution.

    I was recently talking to a young man who used to work with 3 big monitors. Recently he changed to one huge monitor with 4 desktops displayed simultaneously. That change may have affected his consumption of energy more than any change of CPU. What is the point of regulating CPU resolution if it’s irrelevant to the problem of conserving energy in the whole system?

    I am about to drastically reduce the energy-consumption of IT in my home. I will be going from 45nm on Beast’s CPU to 28nm on AMD A1120. That will likely save only 50W. Meanwhile, I will change from 3 hard drives being active to 7, perhaps an increase of 70W. I would have to go to 7nm to make up the difference. The parts aren’t even available that will run this system at 7nm. Oh, and I’m increasing RAM from 6GB to 22GB. How does regulating the CPU’s resolution compensate for that? Governments should do what we can’t individually or in small groups. Setting standards is a valuable such service but speed-limits on highways make sense. Regulating IC-resolution does not. Most likely governments will require energy-savings in IT to come from shutting down idling machines, blanking screens and such, not fabrication of chips. As much as I’d like to see CA ban Intel’s wastage of energy that’s not their role. If they want to do something useful, they should make sure there is competition in the markets so the superior or most efficient products would thrive so no regulation as envisioned by oiaohm would be required at all.

    CA is a very wired state. If new regulations were retroactive and existing hardware had to be scrapped, it’s likely that the average PC would need to be scrapped/recycled promptly. That would burden the Earth with more electronic waste, waste of energy making and shipping new machines, causing much more damage than continuing current energy consumption patterns. CA has ~30million residents with likely 50million PCs. Scrapping half of those as being inefficient would be a nightmarish cost creating thousands of tons of toxic scrap, costing $billions and banning imports would prevent most IT from working so CA would be sent back to the Stone Age for zero benefit.

    If CA were really serious and rational about limiting energy consumption of IT, they would ban all thick clients and allow only the most efficient servers (transactions/s/watt) and ban all other uses of IT. That would be simply insane in disruption of IT but it would likely shut down many millions of PCs and replace them with 2W thin clients like Odroid-C2. That would save many times the energy of simply requiring more efficient CPUs and be simpler to legislate. The CPU is just a tiny part of the problem. Drastic changes in the design of whole systems of IT is much more efficient than replacing any number of CPUs.

    Aside: Odroid-C2 has proven itself capable of running for TLW as a thick client. I had to change to 720P to do that but it’s snappy, cheap and only uses a few watts. It looks like I will buy several more. The hardware is solid even if it’s several steps behind in Moore’s Law and it’s cheap, ~$100 CDN delivered. It’s not government’s role to pick such a winner but I can and I will save serious energy over the lifetime of the new system. I showed it to a guy who uses state-of-the-art equipment all day at work and he was amazed. It’s not the best performance but it’s certainly suitable for most of the client-usage in my home. We surfed, ran local applications, played video from local files and YouTube. It’s all solid and reliable and neither Intel nor CPU-resolution had much to do with it. We’re changing to ARM and enjoying a similar performance folks enjoy in their smartphones on a much more powerful PC-ish environment. The only remaining problem? The things are so tiny we have to be careful not to crush, bend, fold, staple or mutilate… New tech brings new problems but I can live with them even if the nm of the CPU is behind a bit.

  12. oiaohm wrote, “An engine and transmission needs to warm up some before spacings and viscosities reach the right levels to take on full load.”
    So this is not true either Robert. It depends on the design of the engine.”

    We’re talking about small Briggs & Stratton internal combustion engines. They need oil. The manuals all state a brief warm up is desired. Putting such an engine on a roto-tiller at high RPM subjected to heavy loads is just stupid. It has nothing to do with the design of the engine but the properties of the lubricant, some form of oil. Typically these things run on SAE 30 or 10W-30 oil. The stuff tends to more viscous at starting and this increases frictional losses. Further the circulation of oil that is required to keep hot spots and highly stressed parts from wearing prematurely is inhibited somewhat. Also metal parts expand when heated so the clearances between parts won’t be optimal until the engine warms. Many engines recommend a few minutes of idling before use. Diesels may take 15 minutes because of larger masses and higher efficiencies. I don’t care that some exotic new designs are out there that may use magnetic bearings and high clearances like turbine engines. These don’t exist in my market. Current designs have proven simple, cheap and reliable. There is high resistance to changing that. Even the crap-tiller that raced at high RPM from starting stated that it should not be heavily loaded for a few minutes. It took only a year to destroy each transmission. The engine is fine though. It was the transmission that failed under load being an inefficient worm-gear. Not good in heavy clay. Must gear down…

  13. oiaohm says:

    By the way the issue with CA regulation against engine idling is they did not at the same time regulate using other solutions.
    http://www.motoring.com.au/aussie-invention-eliminates-piston-rings-40773/
    This is one of the newest solutions.

    An engine and transmission needs to warm up some before spacings and viscosities reach the right levels to take on full load.
    So this is not true either Robert. It depends on the design of the engine.
    Robert there are engine design that don’t need oil and don’t need to warm up to a particular level before taking full load.

    You would say CA with no idle engine was 1 key regulation short and the result is instead of companies retooling to make engines that in fact work perfectly in no idle configurations they have taken engines that require a idle and attempted to use them no idle that of course does not work well. This is exactly the same problem we have with companies using the wrong nm silicon.

  14. oiaohm says:

    dougman here is the stupid thing prevent people in CA from buying lower grade silicon will save more in CO2 emissions than building additional nuke factories.

    Some of the worst silicon chips are in LCD screens lot are still made at 2002 standards. Yes companies taken the point of view it works we don’t have to fix it. Result in improvements in nm mixed with improvements in production CO2 cost in making a LCD screen using current tech vs out of date tech same size is 1 to 10 and then it has lower running costs. Remember the production of screen electronics using out of date tech end up using more power than the screen will use in it complete operational life.

    Why would companies stay with out date silicon production. It costs a lot of time to update a fabrication plant. It costs time to redo the production plates. Both cost a lot of human time not exactly that hard on the environment but hard on bottom line.

    Scary part is most of the silicon production plants around the world power source is coal based power plants. Since they are not in CA territory the only way to control output of those is really is control what is acceptable imports.

    Also don’t joke about CA sinking into the sea.
    http://pacinst.org/app/uploads/2014/04/sea-level-rise.pdf
    Read section in that PDF Hazardous Materials Sites .

    If the state of CA only partly sinks into the sea we have one huge mother of a mess on our hands. Part of the reason why CA is not interesting at this stage in building more Nuke power plants is you have to be sure you can handle the waste material safely. Due to rising sea water risks they have a huge amount of existing hazardous material that needs to be relocated without adding more Nuke power plants and producing more.

    CA is in quite a horible location really.

  15. dougman says:

    Won’t build additional nuke plants, instead lets regulate computers! The entire State could slide into the Pacific and nothing would be lost.

  16. oiaohm says:

    http://wccftech.com/foundries-tsmc-companies-shift-300mm-wafers/
    This is another fun fact about silicon production. You want to reduce environmental cost mandate that the silicon chips come off bigger wafers. Power cost does not change when making a 200mm 300mm or 450mm. Yes companies using 300mm over 200mm per chip has cut the production power used by 40%. Yes 200mm is 2002 style silicon designs. So groups are using out date production that is absolutely no good for the environment. Of course this is harder to enforce.

    Once you start understanding the numbers you start waking up there is a lot of areas were regulation makes sense the silicon that makes up all the parts around a computer and other things is one of those areas.

  17. oiaohm says:

    Robert Pogson if the regulation I suggested on nm of production had been in place AMD would have bitten their pride sooner and done the deal sooner with TMSC and Samsung production they have done at the start of this year instead of wasting quite a few years sticking on out date uncompetitive nms. Result we would have had AMD cpu and gpu existing as something competitive.

    One of the things making the intel only world at the moment is the fact AMD with GlobalFoundries got nm behind and instead of doing what was required to catch up they chose the path of over drive the silicon.

    Allowing parties to put out second grade tech does not do market any good either.

    The unintended consequence of allowing companies to make over driven silicon is allowing companies to damage there own core reputation so doing even more damage to competition.

    There is a place for regulation. Watt vs area regulation on silicon reduced the risk of consumers being on the end of lemon products due to overheating and other nasty things. So that one is 100 percent for consumer.

    nm regulation would fairly much stop companies for putting brining uncompetitive into the market in the first place so reconsider there production processes. TSMC and Samsung and Intel foundries are always inside months of each other at each new nm level. There was absolutely no reason other than pride that AMD got behind with nm there were foundries that would have taken their work until their existing foundry got up to speed.

    Robert Pogson like it or not the unintended consequence of leaving AMD and others to their own devices has been Intel and Nvidia losing competitors or having weak competitors. So yes argue against what I am suggesting all you like. But I serous-ally suggest you look way more closely at what has happened. My two lots of regulation are about making sure companies do what ever is required to put up competitive products.

    nm regulation could for sure apply to harddrive controllers and ram no problems.

    Do you believe there’s less air pollution if everyone drives a Cadillac “meeting the standards”?
    To put a car on a road it has to meet a min level of standards. watt vs area in silicon is basically the metric that the thing will not explode. nm would be environmental controls.

    Of course there is what is enforceable vs ideal. Both what I have suggested are 100 percent enforceable they are not 100 percent ideal.

    How many hard drives are the grand inquisitors going to allow?
    This is you just being stupid. Nothing I have said contains any about physical size of system. Why that is not enforceable with people building their own systems so by that al-own not worth considering.

    nm control for types of silicon when it comes to cares its like emission control regulation on cars. If a car is a 4 or a 24 cylinder its still facing the same set of regulation.
    It’s the luxury car versus compact car all over again.
    So absolutely nothing I am suggesting is anything like this.

    There are regulations that make sense. Letting markets be too far deregulated does no one any good because the result is always more defective products in the market than there should be. Of course there is such thing as over regulated nothing I am suggesting is in that camp. The nm limit on silicon would also fall into anti-dumping laws. Someone has a old out of date silicon factory they are not willing to up date or outsource to someone with current plant they then produce a stack of lower grade parts hurting the environment in the process then basically dumps that in your market. This is exactly what AMD and others recently got allowed todo.

    Nothing I am suggesting is a bad idea. Its more go look at what has been done and work out the regulation to prevent companies doing actions that make themselves uncompetitive and harmful. The worse thing you can have happen with silicon is a batches of product made that just has to be crushed because no one is buying it because its poor quality yes this has been happening to amd gpu and cpus recently.

  18. oiaohm wrote, “Watts limit on Area on silicon sound simple but the reality it is that simple.
     
    Identical Idling chip at 22nm and 14nm doing exactly the same thing the 22nm is using 2 to 3 times as much in operation power if both run inside safe operating limit.”

    So, you’re going to have CA regulating the resolutions of chips? What about hard drives and RAM? How many hard drives are the grand inquisitors going to allow? How many gigabytes? gigabit v 100 mbits? Number of monitors? It’s the luxury car versus compact car all over again. Do you believe there’s less air pollution if everyone drives a Cadillac “meeting the standards”? What about VW TDI? How complex will IT become with CA getting involved in another layer around all IT-decisions? The last business out of CA should remember to turn off the lights.

    One unintended consequence of oiaohm’s vision of regulation could be an Intel-only regime of IT in CA, substantially raising costs to users. Wasn’t it bad enough when Intel paid OEMs not to install AMD’s stuff? Now we could have CA legislating monopoly in IT. This stinks.

  19. oiaohm says:

    Watts limit on Area on silicon sound simple but the reality it is that simple.

    Identical Idling chip at 22nm and 14nm doing exactly the same thing the 22nm is using 2 to 3 times as much in operation power if both run inside safe operating limit. The 22nm costs 2x as much power when it was manufactured compare to the 14nm chip exactly the same. But they can both have the same watt per area as long as neither chip is over driven.

    Consider Intel. They have a gazillion CPUs. Do they even reveal the area of every core/chip?
    They do because there is no point keeping it secret. You can measure the silicon area of a chip by ultra-sounding it. This test is absolutely non destructive because chips in factories are ultra-sounded to make sure they look somewhere right in construction.

    Power under full load intel and all other silicon makers has to release this for board designers. Specifications for all silicon chips include a min use power and a max use power. Again not trade secret because someone can put a meter on a chip and run it though tests and get this information anyhow. When you make a circuit board width of power rails has to match up to power usage of chips or else board will break at times.

    Please note “Silicon Die area vs wattage.” this is a metric you would apply to all silicon chips in a system. Ram, CPU…. Fun part is the area vs wattage metric applies all the way up to power mosfets. So everything silicon can have exactly the same limit applied.

    Watts under load directly relate to watts vs area plus nm.

    operating voltage and frequency, stuff that can be varied by the user in some cases.
    Some of these settings are should you allow them. Some of them you should not. Like some chips .01 of a voltage up equals the chip doubling waste heat produced and tripling the amp usage basically sending the watts usage sky-rocketing . Why that small voltage change stepped outside the safe watts vs area value.

    The reality you take area of silicon and calculate the worst possible power usage if all the silicon was being used effectively. If a chip is worst than that it has to be power ineffective. The nightmare as well about going outside max possible efficient wattage is you fairly much half the life span of the hardware while using at least three times more power.

    Of course putting a watt vs silicon area limit does not prevent the intel like stunt of powering down 3 cores to over clock 1.

    Banning idling cpus would also require mandating that computers hold state without power. The scary part about idling with a cpu most of the power usage there caused by the ram type requiring refreshing.

    Consider a super-computer. Does area/watt matter when it comes to energy consumption?
    The answer is not no it yes but I am leaving out one factor nm. Bad area/watt chips equal more heat produced equalling more cooling required when the supercomputer is under load. Super computers most are not over-clocked and operating inside the area/watt rule of the chips. If they need more processing they add on more hardware because its more power effective to expand the size of the super computer than it is to over drive anything. So area/watt rule would have basically no major effect on data centres or super computers. Area/watt rule would also have no major effect on general business computers.

    Gamers and pro users who over drive hardware would be in the firing line of an area/watt rule. area/watt rule pure targets over driven bits of silicon that are going to have shorter operational live and are just going to be wasting power like it going out to style. Area/watt rule would cover all nm of silicon production without issue.

    Area/watt limit is about the most vendor neutral rule you could bring in.

    Trying to get power down by enforceable rules Area/watt and nm of production. This is information no silicon maker keeps secret also information you can extract without disassembling the chip. So truly enforceable.

    This is fairly much just setting the rules of the playing field. This would not stop amd or intel or arm or who ever competing for the most power effective design. It just stops people with uncompetitive designs resorting to over driving silicon to hide the fact they are behind.

  20. oiaohm wrote, “Silicon Die area vs wattage.”

    That’s way too simple. e.g. RISC machines and highest resolution would be mandated. That’s just not going to happen as promptly as legislation can be generated. e.g. wattage at what load? Maxed-out? That’s quite unrealistic as most PCs are idling. If the object is to save energy smaller PCs idling are way better than the latest tech running flat out yet they may have same area/watt of TDP. The wattage varies with load while the area doesn’t. There are cases where cores are shut down by software so even the area may vary effectively. IT is just too diverse to be described by or regulated by a concept as simple as area/watt. Consider a super-computer. Does area/watt matter when it comes to energy consumption? Nope. Just watts under load.

    Consider Intel. They have a gazillion CPUs. Do they even reveal the area of every core/chip? Probably some of those are trade-secrets. Power under load? Nope. That’s highly dependent on peripherals and RAM as well as operating voltage and frequency, stuff that can be varied by the user in some cases. Will CA regulate what can be controlled by users, just like my stupid roto-tiller from Sears?

  21. oiaohm says:

    Robert Pogson funny enough performance per watt limitation is very simple to define. Silicon Die area vs wattage. When you are over driving silicon you use more watts per cubic mm of silicon than you really should Result of over driving silicon is producing more and more heat that is not productive action.. Attempt to use a lower grade of nm production find self over the wattage limit as well.

    suppose all idling CPUs are banned.
    Mind you this could be banned and it would in fact make sense and force chip makers to invest in tech that currently exists..

    Then Beast would have to wake up from hibernation at the start of every process, a huge loss in human productivity.
    That is presuming cpu and ram maker remain using the current cheep garbage tech they do.
    IBM experimented with a Magnetoresistive random-access memory (MRAM) version of a powerchip. That chip could power down and hold it last state perfectly.

    MRAM is faster than our current dram and its totally non-volatile. Of course no one wants to have to pay patents over to IBM.

    So its perfectly possible to built a current day computer with no idle that while there is no task to perform just stops completely dead when a interrupt appears powers up and goes on as normal as if it has not been stopped. So completely no idle only powered when it has something to process. There was a original Magnetic-core memory computer that functioned exactly like this is not even a new idea. Yes old computer does not need a ups just unplug it from one power point plug it into the next power point and it goes along as if nothing happened. No batteries no capacitors state just held by magnetic field.

    This kinda gives you an idea how bad our current tech is.

    An example: Suppose I’m making movies and all heavy-hitting workstations are banned. What’s to prevent me from installing 100 whimpy PCs in place of some hog to get equivalent total performance while increasing energy consumptions 30 times?
    Issue we have now is this exactly what person making movies can do at times. So instead of paying for Nvidia or Intel gpu that currently at a decent NM use AMD ones that are not power effective. Banning on nm vs watt would not ban all heavy-hitting workstations. Also you have the numbers backwards.

    100 whimpy PC replacing one heavy-hitting workstation using over driven silicon chips the 100 whimpy PC would be using 1/3 of the power even allowing for 20 percent power loses in the duplicated hardware. This is the big bad problem with anything with over driven silicon it makes no logical sense to-do it other than saving floor space.

    Its the old overclocking stupidity problem. Yes you can make X cpu/gpu go faster by over clocking/over driving it at the price the thing ends up wasting power hand over fist. So you gain like 2 percent in performance and double power usage. You could have increased performance by doubling the computer and using the same power. Out lawing over-driving silicon makes sense. In fact out lawing cpus from being able idle so they have use MRAM or equal makes lots of sense kills off most people need for a UPS units that would help the environment a lot. Lead batteries in UPS are quite a on going cost.

    In fact passing a law that all computers have to be able to hold state with power disconnected and restored without using batteries or capacitors would be a great thing. Its not like computers like this cannot be made.

    Our current day computers are no where near as power effective as they can be.

  22. AdmFubar wrote, “sounds like someone is trying to push users to their “cloud” infrastructure by law…”.

    That makes as much sense as anything but clearly some high-performance tasks can’t run using clouds unless they are local and on the LAN. CA would have to legislate minimum ISP throughput as gigabit/s… TFA mentions $18 per PC as “cost” but raising ISP prices tenfold would make that a drop in the bucket. What’s the cost of various departments or whole businesses leaving the state? What’s the cost of legislating higher rates of refresh for IT? At the least it’s a huge disruption. Would the beleaguered IT-manufacturers produce stuff certified by CA or just quit selling there? The whole world benefits from standardization. CA is planning fragmentation. CA is already stressed in budget and drought. What’s the hit on GDP due to intruding this way into IT? There is a price for energy. Why can’t the market decide, perhaps with a campaign of education/rating, instead of going the discredited “central control” route? I can tell you if my province, Manitoba, required me to switch to Odroid-C2-only performance with Beast outlawed, I’d likely cheat or move. Odroid-C2 can do most of what we do but hour-long builds of Linux and less than fullscreen video are unacceptable. There needs to be diversity in IT, not discrimination based on 1-D thinking. To be workable, such legislation would have to be utterly without benefit while imposing huge costs.

  23. oiaohm wrote, “Study finds 70% of the energy a typical laptop will consume during its life span is used in manufacturing the computer”

    A performance/watt legislation would have to legally define performance. That would discriminate between AMD/Intel, and I/O v CPU-intensiveness. There is no way a 1 or even 2-D set of numbers could properly define performance limits in various usage patterns: reading text/idling, gaming, 3-D simulation, weather forecasting, maths, multimedia creation, playback, running 10000 different applications…

    Consider my poor Beast. At any point in time Beast might be idling with nothing but DHCP answering, loaded heavily with two or more browsing clients, maxed out in I/O making backups or maxed out in CPU compiling Linux. There’s no fair way Beast’s performance per watt could be defined, let alone measured. e.g. suppose all idling CPUs are banned. Then Beast would have to wake up from hibernation at the start of every process, a huge loss in human productivity. What’s the legally/socially acceptable limit on that hit? How many impatient millions would be impelled to cheat or leave CA in response? Would there have to be a “war on …” to fight human nature? How many such wars have actually made the world a better place?

    Consider EPA-like regulations of automobiles. We have the nonsense that 12-cylinder and one-cylinder vehicles are both legal on the basis of air-pollution. It’s utter nonsense to accept a 4L/100km compact car and a 13L/100km luxury car as being at the legal limits. That clearly is a broken system. EPA, for instance sets the legal limit on the whole fleet of production which clearly discriminates between large and small producers rather than between good and bad pollution-models. That’s the kind of nonsense I see coming. An example: Suppose I’m making movies and all heavy-hitting workstations are banned. What’s to prevent me from installing 100 whimpy PCs in place of some hog to get equivalent total performance while increasing energy consumptions 30 times? Will there be a ban on installations like urban bans on multiple pets? What if I create 100 “independent” organizational units to get around that, or I just move out of state by a mile? How will the environment be the least bit better? No. This is just a move to make it harder to use or to provide IT in CA.

  24. oiaohm says:

    Robert Pogson please note the numbers in that report are based on a 2002 device. Things have moved quite a lot.

    Scary part is a 2002 device has 4 to 8 times as much silicon in it as a current day device. Most power consuming part of making a computer is making the silicon.

    https://en.wikipedia.org/wiki/List_of_Intel_Core_i7_microprocessors

    The issue people have missed is ram and cpu have had a reducing die size what is reducing silicon. 4 to 8 reduction in silicon in new computer results is a old computer using more power over it life than it will cost to crush the old computer and make a new one. Over driven silicon like some AMD CPU and GPU burn 10 to 12 times more power than what is sane for the silicon size. Yes a GPU sucking down 300+watts of power is clear sign of o my wrong. The reality is size of silicon in the AMD gpu caught doing 300+watt equals burning through the power cost of production of it in 2 years. Driven to the point the silicon is not generating massive excess heat based on chip size and construction should have been pulling between 5 and 10 watts.

    Good old hardware and good new hardware is about 70% percent production 30% usage in usage power. The catch is the amount of power to make new hardware is way less than the amount of power to make the old hardware add in that the old hardware performance per watt is not great that is how you find your self with a old machine that is better to recycle it than keep on using it. Yes new hardware is running way more power effective as well so the new hardware has cut down its power usage so the 70/30 ratio has not moved.

    Bad old and new computer hardware can be 10% production energy cost and 90% in use energy cost. Lot you would find in this case 10% production and 90% in use are in the high end GPU and CPU class where they did not care about having to put on huge cooling to take away the massive bits of wasted heat being produced.

    People forget as early as the 386 and 486 computers we had 450watt power supplies and these are not 90% effective power-supplies.

    Basically old machines need to be sorted. Old computer with a large old gamers/high end pro video card and cpu in it you are not saving the environment with that video card or cpu in a lot of cases.

    Your general slugger all round office PC you can be saving the environment keeping those going for a long time.

    It may only take ~$40 per annum to run a PC so cutting energy consumption in half may not make sense for a PC costing $hundreds.
    This really depends on the machine. General office and power effective computer put under $40 per annum on power bill. Old Gamer and Old high end pro machines put 100-200 dollars on your bill. Comes quite clear what ones are not environmentally sound. When you end up over 10 years having a larger power bill than the machine was worth new the machine and since old gammer and old high end was over price they environmentally sound in the first place.

    The fact amd and others at times have made cards that we don’t care about power usage we just need to push the tech to limit so we benchmark as good as our competition there need to be some regulation. A performance per watt mandate would do it that anything more than 10% worse than the competition is not a sellable production.

  25. oiaohm wrote, “The difference between newer silicon and old silicon could mean from a power point of view you would be better off to crush and recycle the materials of older computers than leave them running.”

    It does take a lot of energy to mine, ship, refine and manufacture. It may only take ~$40 per annum to run a PC so cutting energy consumption in half may not make sense for a PC costing $hundreds. You can’t have it both ways, running the thing 10 years to save money in the long run and scrapping it before end of life to benefit from Moore’s Law.

  26. AdmFubar says:

    sounds like someone is trying to push users to their “cloud” infrastructure by law…
    always gotta look at the where the money is really gonna flow..

  27. oiaohm says:

    dougman there is a old saying that applies.
    The road to Hell is paved with good intentions
    Robert Pogson at this point performance per watt limit could be a good idea.

    http://www.phoronix.com/scan.php?page=article&item=9800gtx-gtx-1080&num=7

    Notice here the new GeForce GTX 1080 from nvidia consumes the less power than GeForce GTX 680 only slight more than GeForce GTX 980 and everything else newer than a GTX680 that is not a GTX1080 uses more power. Then you remember if you take this to power to watt GeForce GTX 1080 kicks all the old cards ass.

    Repurposing older computers for lighter loads is a good thing but it saves no energy except the energy needed for manufacture.
    Problem we have here is we cannot be sure this is a saving. The difference between newer silicon and old silicon could mean from a power point of view you would be better off to crush and recycle the materials of older computers than leave them running.

    The energy to manufacture does reduce to produce the same performance as nm of silicon decreases because the number of chips per wafer goes up. There is line where the power usage of existing machine exceeds the cost to make new from recycling that machine. So some older computers should be going straight in the crushers heck some of the AMD new chips should be going straight in the crushers they are not good enough.

  28. dougman says:

    California is filled with nothing but idiots at the governmental level. I was twice offered positions for the west coast. I turned both down for good reason.

Leave a Reply