A modern CPU may require all voltages to be supplied externally, or may generate everything from a single 3.3V or 5V supply, depending on the application.
This is old information. Nowadays, DC-to-DC converters on the motherboard take 12V power (f.e. from that 4-pin connector that was added back when P4s came out) and convert it to the lower voltages for the CPU and RAM. Increasingly few things, such as expansion cards, USB, and drives use the 3.3V and 5V power (take a look at what wattages are supported for each voltage on modern power supplies), and what Google did is ask for a motherboard that dispenses with the need for them completely - just add a few more DC-to-DC converters to the motherboard if those voltages are still needed.
And since it's long-term, I'm willing to wager it won't be lead-acid, but NiMH. So no real maintenance issues.
Google is definitely using lead-acid batteries - the same 12V sealed lead-acid batteries that would go in a small UPS, keycard system, fire alarm system, etc.
As for maintenance, most lead-acid batteries last about 4 years, and both Google and Facebook are probably retiring their old hardware faster than that.
As for containment on failure, I'd say that's a well-understood solved problem.
Most of those boxes have a "conflicting green" detector circuit that automatically puts the signal in "safe mode" when it detects two conflicting green lights.
That's what's supposed to happen. I have anecdotal non-proof from several years ago that that may not be universal, though. In Renton, WA, just west of the Sunset Blvd intersection, Bronson Way crosses a couple railroad tracks (which recently were only used for the Spirit of Washington dinner train and have now been torn up completely in Bellevue). When a train came, the gates would lower and all four sides would see flashing red lights. After the train passed and the gates lifted, both Bronson and Sunset traffic attempted to move immediately, then slammed on the brakes to avoid hitting each other. This happened not once but three out of the three times I was able to witness it. Now, this being Renton, it's possible that the other drivers were just retarded, but the most likely explanation is that everyone was seeing a green light. From my viewpoint I couldn't see the traffic lights pointing the other direction, so I can't be sure.
I think EVs need to be more strictly regulated in their mileage claims.
I would be in favor of mileage claims based on the ratio of distance traveled vs. energy used (MJ/mile, miles/MJ, MJ/km, km/MJ, etc), along with a specified expected electric-only range (for plug-in hybrids).
But with all this talk of the entire planet harvesting wind I don't think I've seen so much as a single study on what taking the large chunks of energy out of the wind will do to our planet.
You misspelled "insignificantly small chunks". And we've already taken out not-quite-as-insignificantly small chunks by building billions of houses.
It would really suck if we just traded one "uh oh" for another.
Every method of energy production has an environmental impact. That is a red herring. A useful discussion will center around which set of environmental impacts can be most easily tolerated.
So unless you are prepared to lie flat in a coffin shaped car, your pretty much stuck with the crossection of a Honda as the minimum useful car. Thus there's no way to beat that power demand by more than a small percentage let alone a factor of even 2.
Mike Turner modified his Honda Civic to reduce its Cd from 0.34 to 0.17 for less than $400 in parts. That is exactly a factor of 2.
User is left alone in a sound-proof room and given a set of tasks to perform.
I've participated in this type of testing. While it is very valuable, there are a couple problems with relying heavily on it:
The "tasks" subconsciously get described in the testing team's terminology, not the users'. This can provide the user an advantage that they wouldn't have in real life.
The methodology considers a users' ability to stumble through an application but does not consider support staff's ability to walk the user through the application via written procedures or over the phone. Yes, the former should be a priority, but not to the exclusion of the latter.
I hope they put a capacitor in there with a bridge rectifier instead of just ignoring half of the 50/60 Hz cycle.
A capacitor would negatively affect the bulb's power factor unless carefully sized and paired with a matched inductor. Just using the bridge rectifier would allow the LED to use both halves. LEDs tend to blow up when you get them too hot, not when you feed them too much current for very short periods, so there isn't any particular need to smooth out the current.
Seriously, build batteries? That's your bright idea?
There's a couple other ways (that we've already deployed and that we know work). With solar thermal, you can store the heat energy in large vats of molten salt, using that energy to spin turbines for a few days. With hydroelectric, you can pump the water back uphill. As for batteries, don't write them off entirely. Large Sodium-Sulfur batteries have been deployed on Japan's grid to store their wind energy.
You cannot reach space using just a scramjet anyway. You need rockets. There is no air in space.
Obviously. But a rocket that only has to add that last third or fourth of orbital delta-v can be designed a lot more conservatively and carry a lot more payload (which means a lower cost per kg to orbit). This scramjet is maneuverable - it can stay in air to breathe for as long as it takes to accelerate, then it can let the upper rocket stage finish the job.
The reason is simple - the reason a rocket costs so darn much has nothing to do with fuel. It has to do with complexity - it's very expensive to make something as complicated as a rocket work under all the stresses of a launch.
All of that complexity in turn stems from the insanely high mass ratios that chemical rockets require to achieve orbital velocity (93.5% of the Shuttle's weight is fuel/tank/boosters). That, in turn, means the rocket itself must be made as lightly as possible; you cannot merely overengineer things like you would a car (or to a lesser extent, a plane) and hope that you'll still get off the ground with them. And if your calculations are wrong, the part fails and perhaps takes the rocket with it.
Since a scramjet gets its oxygen from the air, you can potentially carry up a lot more mass. Some of this could be payload, some of it could be a heavier-duty structure with a better safety margin.
Also, most of the complexity of a scramjet is in the design of the shape. This test vehicle has only one moving part, the fuel pump.
I want this too, but most of the time I don't need it. It'd be just as good for me to carry a separate GPS unit for those occasions (or other self-locating device - some cell phones have this via cell tower triangulation for E911) which can log where I've been when, and correlate the image timestamp with the location after the fact (either on a PC or by connecting the camera and GPS via USB/Wifi/Bluetooth).
With 15 1.5TB drives in a RAID6, I wouldn't bother. It's just too likely that one or more drives will not merely fail but return garbage data, which RAID3/4/5/6 will propagate across the array on rebuild.
If you do this, don't focus on building systems redundant. Focus on building redundant systems. Make your software tolerate losing any one array or server. If you lose an array, don't rebuild; start again from scratch.
Also, it's quite clear that these servers are not particularly speed-oriented. They have 15 drives serviced by a single PCI SATA controller (through SATA port multipliers). PCI's maximum bandwidth is 133MB/sec (33MHz, 4 bytes per cycle). Today, two drives could soak that with sequential reads and writes. That's fine for this company's users (who are backing up to these servers over their slow internet link) but if you're thinking of putting one in at your office to back up your 20TB fileserver, you better think long and hard about how much time you'll have to do a complete restore from it before you're filing unemployment claims. You could speed things up quite a bit by putting in two PCIe 24-port RAID controllers, dispensing with the port multipliers, and using RAID10, but that does add a few thousand to the price and cut the capacity quite a bit.
Sorry, I just read about the hangar requirements (among other things, a max wingspan of 44 feet). This obviously disqualifies Voyager and many other motorgliders. Probably the winner (as scored by their 1/((1/mph) + (2/Passenger-MPGe)) formula) will look like a cross between the two types of planes.
I'm going to guess, however, that if it actually IS done, that it will be some variation of a Rutan Long-EZ, since they are widely known/touted as "the plane" for high-efficiency experimental-class aircraft.
Wikipedia says that a Long-EZ will do 1600 miles on 52 gallons of fuel. That's 61.5 passenger miles per gallon. It also typically cruises at 184mph - parasitic drag will be 3.39 times less at 100mph, but induced drag will be 3.39 times greater. I am unable to find a chart of both for a Long-EZ (here is a generic one), but 100mph probably isn't that far off from the minimum drag speed.
I suspect it'll be some variation of a motorglider - probably one that seats at least two. They have much higher aspect ratio wings, much lower sink rates, and would probably have much lower drag at 100mph.
The Voyager around-the-world aircraft (another Rutan creation) did only 41 passenger miles per gallon (averaged across the entire flight), but they were hauling 9000 lbs of fuel towards the start (53 passengers worth). I suspect it could win the challenge right now - but it'd make a lot more sense to build a different one than to unhook it from the Smithsonian ceiling.
I've tried an X25-M on a few servers with LSI SAS controllers (as used by PERC 6i, though I don't think I've used that exact chip) and been disappointed to encounter IO hangs
I've seen the same with X25-Ms using the onboard RAID on Intel S5000 server boards (which use an LSI SAS controller). Putting in a 3ware 9650 fixes it (There are probably lots of other controllers that would do a good job too).
Hopefully Intel will do an SAS version at some point; they could compete against 15kRPM drives rather well, I think.
Once you get it working on a decent controller, you'll forget all about 15krpm drives (and SAS) entirely.
I don't recall being able to apply 1.23 eV across a cell.
You don't. For example, with ordinary pure water electrolysis, you apply voltage (at least 1.23 volts - any less than that and no reactions will occur) to the water, and 1.23 eV gets used up with each and every molecule you split (lots more can potentially get used up as waste heat, and the goal of efficient electrolysis is to minimize this).
Slashdot Mirror
This is old information. Nowadays, DC-to-DC converters on the motherboard take 12V power (f.e. from that 4-pin connector that was added back when P4s came out) and convert it to the lower voltages for the CPU and RAM. Increasingly few things, such as expansion cards, USB, and drives use the 3.3V and 5V power (take a look at what wattages are supported for each voltage on modern power supplies), and what Google did is ask for a motherboard that dispenses with the need for them completely - just add a few more DC-to-DC converters to the motherboard if those voltages are still needed.
Google is definitely using lead-acid batteries - the same 12V sealed lead-acid batteries that would go in a small UPS, keycard system, fire alarm system, etc.
As for maintenance, most lead-acid batteries last about 4 years, and both Google and Facebook are probably retiring their old hardware faster than that.
As for containment on failure, I'd say that's a well-understood solved problem.
That's what's supposed to happen. I have anecdotal non-proof from several years ago that that may not be universal, though. In Renton, WA, just west of the Sunset Blvd intersection, Bronson Way crosses a couple railroad tracks (which recently were only used for the Spirit of Washington dinner train and have now been torn up completely in Bellevue). When a train came, the gates would lower and all four sides would see flashing red lights. After the train passed and the gates lifted, both Bronson and Sunset traffic attempted to move immediately, then slammed on the brakes to avoid hitting each other. This happened not once but three out of the three times I was able to witness it. Now, this being Renton, it's possible that the other drivers were just retarded, but the most likely explanation is that everyone was seeing a green light. From my viewpoint I couldn't see the traffic lights pointing the other direction, so I can't be sure.
We better start cooling our surroundings to absolute zero immediately!
I would be in favor of mileage claims based on the ratio of distance traveled vs. energy used (MJ/mile, miles/MJ, MJ/km, km/MJ, etc), along with a specified expected electric-only range (for plug-in hybrids).
You misspelled "insignificantly small chunks". And we've already taken out not-quite-as-insignificantly small chunks by building billions of houses.
Every method of energy production has an environmental impact. That is a red herring. A useful discussion will center around which set of environmental impacts can be most easily tolerated.
Mike Turner modified his Honda Civic to reduce its Cd from 0.34 to 0.17 for less than $400 in parts. That is exactly a factor of 2.
I've participated in this type of testing. While it is very valuable, there are a couple problems with relying heavily on it:
True. However, if you want to walk someone through the new interface over the phone, it SUCKS.
A capacitor would negatively affect the bulb's power factor unless carefully sized and paired with a matched inductor. Just using the bridge rectifier would allow the LED to use both halves. LEDs tend to blow up when you get them too hot, not when you feed them too much current for very short periods, so there isn't any particular need to smooth out the current.
There's a couple other ways (that we've already deployed and that we know work). With solar thermal, you can store the heat energy in large vats of molten salt, using that energy to spin turbines for a few days. With hydroelectric, you can pump the water back uphill. As for batteries, don't write them off entirely. Large Sodium-Sulfur batteries have been deployed on Japan's grid to store their wind energy.
Obviously. But a rocket that only has to add that last third or fourth of orbital delta-v can be designed a lot more conservatively and carry a lot more payload (which means a lower cost per kg to orbit). This scramjet is maneuverable - it can stay in air to breathe for as long as it takes to accelerate, then it can let the upper rocket stage finish the job.
All of that complexity in turn stems from the insanely high mass ratios that chemical rockets require to achieve orbital velocity (93.5% of the Shuttle's weight is fuel/tank/boosters). That, in turn, means the rocket itself must be made as lightly as possible; you cannot merely overengineer things like you would a car (or to a lesser extent, a plane) and hope that you'll still get off the ground with them. And if your calculations are wrong, the part fails and perhaps takes the rocket with it.
Since a scramjet gets its oxygen from the air, you can potentially carry up a lot more mass. Some of this could be payload, some of it could be a heavier-duty structure with a better safety margin.
Also, most of the complexity of a scramjet is in the design of the shape. This test vehicle has only one moving part, the fuel pump.
I want this too, but most of the time I don't need it. It'd be just as good for me to carry a separate GPS unit for those occasions (or other self-locating device - some cell phones have this via cell tower triangulation for E911) which can log where I've been when, and correlate the image timestamp with the location after the fact (either on a PC or by connecting the camera and GPS via USB/Wifi/Bluetooth).
With 15 1.5TB drives in a RAID6, I wouldn't bother. It's just too likely that one or more drives will not merely fail but return garbage data, which RAID3/4/5/6 will propagate across the array on rebuild.
If you do this, don't focus on building systems redundant. Focus on building redundant systems. Make your software tolerate losing any one array or server. If you lose an array, don't rebuild; start again from scratch.
Also, it's quite clear that these servers are not particularly speed-oriented. They have 15 drives serviced by a single PCI SATA controller (through SATA port multipliers). PCI's maximum bandwidth is 133MB/sec (33MHz, 4 bytes per cycle). Today, two drives could soak that with sequential reads and writes. That's fine for this company's users (who are backing up to these servers over their slow internet link) but if you're thinking of putting one in at your office to back up your 20TB fileserver, you better think long and hard about how much time you'll have to do a complete restore from it before you're filing unemployment claims. You could speed things up quite a bit by putting in two PCIe 24-port RAID controllers, dispensing with the port multipliers, and using RAID10, but that does add a few thousand to the price and cut the capacity quite a bit.
And just like that, Europe's diesel infrastructure vanished in a puff of logic.
Wouldn't Bilski render all of these patents invalid and therefore make this all moot?
My favorite was "Rat Shack".
Sorry, I just read about the hangar requirements (among other things, a max wingspan of 44 feet). This obviously disqualifies Voyager and many other motorgliders. Probably the winner (as scored by their 1/((1/mph) + (2/Passenger-MPGe)) formula) will look like a cross between the two types of planes.
Wikipedia says that a Long-EZ will do 1600 miles on 52 gallons of fuel. That's 61.5 passenger miles per gallon. It also typically cruises at 184mph - parasitic drag will be 3.39 times less at 100mph, but induced drag will be 3.39 times greater. I am unable to find a chart of both for a Long-EZ (here is a generic one), but 100mph probably isn't that far off from the minimum drag speed.
I suspect it'll be some variation of a motorglider - probably one that seats at least two. They have much higher aspect ratio wings, much lower sink rates, and would probably have much lower drag at 100mph.
The Voyager around-the-world aircraft (another Rutan creation) did only 41 passenger miles per gallon (averaged across the entire flight), but they were hauling 9000 lbs of fuel towards the start (53 passengers worth). I suspect it could win the challenge right now - but it'd make a lot more sense to build a different one than to unhook it from the Smithsonian ceiling.
The eye can respond to single photons, but the brain filters them out below about a hundred per second.
I've seen the same with X25-Ms using the onboard RAID on Intel S5000 server boards (which use an LSI SAS controller). Putting in a 3ware 9650 fixes it (There are probably lots of other controllers that would do a good job too).
Once you get it working on a decent controller, you'll forget all about 15krpm drives (and SAS) entirely.
Just like my point-and-shoot camera doesn't care whether something is 100 feet away or several miles away when I manually set it to infinite focus, the Hubble Space Telescope doesn't care whether something is a light second or several billion light years away. It has imaged every planet in the solar system except Mercury (including Earth), has imaged the moon, and once indirectly imaged the sun.
Without that last qualification things get a little hairier.
You don't. For example, with ordinary pure water electrolysis, you apply voltage (at least 1.23 volts - any less than that and no reactions will occur) to the water, and 1.23 eV gets used up with each and every molecule you split (lots more can potentially get used up as waste heat, and the goal of efficient electrolysis is to minimize this).