Note: A single mode fibre can carry 160 channels of 10 Gbits/s, or 1.6Tbits/s.
The reason it is not widely accepted (outside of big telecommunicatons suppliers) is the huge cost. For example, each channel requires a seperate laser, and may cost in the region of $20000 each. Multiply that by the maximum number of channels and you are looking at about $3.2 million just for the lasers alone. Once you add the detectors and high speed electronics to process all this DWDM is VERY expensive.
Only long distance telco links and Microsoft can justify the expense:
Single power distribution would decrease power conversion efficency.
The main PSU is a switching power supply, which is an efficient and cheap method of dropping from 120V/240V down to 12V. A large voltage drop and high current, the alternative is a BIG transformer weighing several kgs...
Switching regulators are available, which can step down from from say 24V to 5V etc. Unfortunately, they are cost prohibitive and still will not provide the clean power required. Also, the lower the required voltage drop the less efficient they are.
Linear regulators have the opposite efficiency curve, they get more efficient the lower the voltage drop required. Of course, they dispate ALL of the dropped voltage as heat. Pdispated = Iout * (Vin - Vout). If Vin is close to Vout then they are near 100% efficient.
In a PC, the PSU provides voltages close to those that are required, and linear regulators are placed close to the components requiring the power. The linear regulators provide clean power but dispate a little as waste.
For the battery/adapter powered product I am currently working on the cost difference between a switching regulator and a linear regulator is about 10 times.
In summary, PCs are powered the way they are because it is the best cost/ performance ratio. A single power-rail will be more expensive and less efficient.
It would make sense if server power supplies could accept DC rather than AC. This would save the round tripping of AC (mains) to DC (battery) to AC (psu) to DC (computer). Telecomms has standardised on 48V, and it is possible to buy 48V DC PSUs. I bet you can get 48V dc UPS units too...
How much does a cable cost anyway?
Removing the second cable saves money, and raises the potential for a second sale. However, making the cable type optional is likely to increase costs.
If the box could contain two versions, then you double the number of product variations. Increased stock holding would be an expensive requirement in that case.
If the cable was included in the cost, but packaged separately, then you have the cost of having to match the cable to the customer order. Most of the savings would be eroded.
The lesser used cable omitted, saving money in the general case, is probably the best solution other than shipping two.
I feel Apple should include that second cable anyway, as they are leading brand in this market. I would think it gives a better impression to include what the customer may require.
It would be next to impossible, inflexible and expensive. Why not just take existing hardware and use a potting compound?
Many of those capacitors are there to reduce switching noise and provide stable power.
E.g. (stable power). The capacitor bank around the CPU is required because of the inductance of the PSU wires. Inductors oppose changes in current, so when the CPU suddenly requires more current it will get it from the capacitors.
Those capacitors have been getting LARGER over time, because the step changes in current required by a processor have been increasing.
Mind you, you could eliminate the problem by constantly drawing your maximum current (and wattage) - and wasting the excess as heat.
It would be impossible to fit those capacitors on silicon, they are just too large. You can place TINY value capacitors on silicon, but you need lots of die space (= expense).
--
Resisters could be placed on silicon, but the tolerence will be something like +/- 50%. You could laser trim to get better tolerences, but at GREAT expense.
--
The smaller chips are REUSABLE design elements, produced in bulk quantities. I doubt they could be integrated for less cost than placing them on the PCB.
--
Contary to what you might think, a lot of effort has gone in to that PCB design. And a lot of effort in to the VLSI chips on that board. If it could be done better, cheaper then it would.
If you use electrolysis on water, then you get a highly explosive gas mixture of Oxygen and Hydrogen.
To reduce buoyancy you need to either eject the gas. If you use a valve, it could clog and also be a possible ignition source. If you burn it, you it would have to be at a controlled rate, or you sink quickly.
This announcement has implications for Hewlett Packard, as they spun off everything but the PC and server business. As the real innovation of the old company is now in Agilent, I can't see long term prospects for HP being favourable.
There is two varients of the word.
In British English your version is correct, however like many words the US chose something different. They also pronounce the word differently.
See Wikipedia:
http://en.wikipedia.org/wiki/Aluminum
Explosive, incendiary or poisonous (fake) currency would have a greater affect than curious bomblets. Everyone would pick them up with great harm to themselves, general morale and the economy.
The 200LX is far superior for graphing, due to speed and a high resolution display. These two things are the main weakness of my 48GX.
Taking seconds to over a minute to plot graphs is not fun, especially when you get a blocky 141 x 64 (been a while, but it's about that) pixel graph.
However, that blocky LCD provides pretty good contrast - perhaps why HP choose a lower resolution display.
Moller said that today's economics give each M400 a theoretical price tag of around half a million dollars, but in volume production it could drop to $300,000 and in really large volumes to below $50,000.
If you can afford one of these the MPG isn't going to be an issue.
Slashdot Mirror
There is of course the wikipedia article:
m ultiplexing
http://en.wikipedia.org/wiki/Wavelength_division_
Note: A single mode fibre can carry 160 channels of 10 Gbits/s, or 1.6Tbits/s.
The reason it is not widely accepted (outside of big telecommunicatons suppliers) is the huge cost. For example, each channel requires a seperate laser, and may cost in the region of $20000 each. Multiply that by the maximum number of channels and you are looking at about $3.2 million just for the lasers alone. Once you add the detectors and high speed electronics to process all this DWDM is VERY expensive.
Only long distance telco links and Microsoft can justify the expense:
http://www.lucent.com/press/0698/980603.nsa.html
Jonathan
Single power distribution would decrease power conversion efficency.
The main PSU is a switching power supply, which is an efficient and cheap
method of dropping from 120V/240V down to 12V. A large voltage drop and
high current, the alternative is a BIG transformer weighing several kgs...
Switching regulators are available, which can step down from from say 24V
to 5V etc. Unfortunately, they are cost prohibitive and still will not
provide the clean power required. Also, the lower the required voltage
drop the less efficient they are.
Linear regulators have the opposite efficiency curve, they get more
efficient the lower the voltage drop required. Of course, they dispate ALL
of the dropped voltage as heat. Pdispated = Iout * (Vin - Vout). If Vin
is close to Vout then they are near 100% efficient.
In a PC, the PSU provides voltages close to those that are required, and
linear regulators are placed close to the components requiring the power.
The linear regulators provide clean power but dispate a little as waste.
For the battery/adapter powered product I am currently working on the
cost difference between a switching regulator and a linear regulator is
about 10 times.
In summary, PCs are powered the way they are because it is the best cost/
performance ratio. A single power-rail will be more expensive and less
efficient.
It would make sense if server power supplies could accept DC rather than
AC. This would save the round tripping of AC (mains) to DC (battery) to
AC (psu) to DC (computer). Telecomms has standardised on 48V, and it is
possible to buy 48V DC PSUs. I bet you can get 48V dc UPS units too...
Jonathan
What do you think you use as bait?
This unfortunately means you end up benchmarking loop overhead.
You could either enable loop unrolling ONLY, or keep optimisations disabled and unroll by hand.
On my 1.4 Mhz Athlon, your code runs in 17 secs (no opts) and in 0 secs (level 3 opts) :P
Unrolling by hand, 20x, gives 5 secs (no opts).
How much does a cable cost anyway? Removing the second cable saves money, and raises the potential for a second sale. However, making the cable type optional is likely to increase costs. If the box could contain two versions, then you double the number of product variations. Increased stock holding would be an expensive requirement in that case. If the cable was included in the cost, but packaged separately, then you have the cost of having to match the cable to the customer order. Most of the savings would be eroded. The lesser used cable omitted, saving money in the general case, is probably the best solution other than shipping two. I feel Apple should include that second cable anyway, as they are leading brand in this market. I would think it gives a better impression to include what the customer may require.
It would be next to impossible, inflexible and expensive. Why not just take existing hardware and use a potting compound?
Many of those capacitors are there to reduce switching noise and provide stable power.
E.g. (stable power). The capacitor bank around the CPU is required because of the inductance of the PSU wires. Inductors oppose changes in current, so when the CPU suddenly requires more current it will get it from the capacitors.
Those capacitors have been getting LARGER over time, because the step changes in current required by a processor have been increasing.
Mind you, you could eliminate the problem by constantly drawing your maximum current (and wattage) - and wasting the excess as heat.
It would be impossible to fit those capacitors on silicon, they are just too large. You can place TINY value capacitors on silicon, but you need lots of die space (= expense).
--
Resisters could be placed on silicon, but the tolerence will be something like +/- 50%. You could laser trim to get better tolerences, but at GREAT expense.
--
The smaller chips are REUSABLE design elements, produced in bulk quantities. I doubt they could be integrated for less cost than placing them on the PCB.
--
Contary to what you might think, a lot of effort has gone in to that PCB design. And a lot of effort in to the VLSI chips on that board. If it could be done better, cheaper then it would.
Oxyfuel cutting is unsuitable for cast metals, unless you add flux. The cut quality will be reduced, although you won't cut at all without.
A new finish would be required, as just to start cutting requires 900oC. Straight cuts are challenging. Definately practice.
My own experience is limited to about 30 hours of practice, on mild steel.
Oxyfuel cutting - process and fuel gases
If you use electrolysis on water, then you get a highly explosive gas mixture of Oxygen and Hydrogen.
To reduce buoyancy you need to either eject the gas. If you use a valve, it could clog and also be a possible ignition source. If you burn it, you it would have to be at a controlled rate, or you sink quickly.
This wouldn't be slashdot without the animated ASCII version. Written in duct tape (perl), Unix only (due to clear).
.. 10) { ;
#perl
use Time::HiRes 'usleep';
$_="Joke\n O \n -|-\n / \\\n";
for $i (0
system(clear);
print;
usleep(100000);
$i==9 ? s/^.*/ *Whoosh*/ : s/^/ /
}
This announcement has implications for Hewlett Packard, as they spun off everything but the PC and server business. As the real innovation of the old company is now in Agilent, I can't see long term prospects for HP being favourable.
[Note to HP: Invent, not rebadge]
Is the old (evil) data collecting overlords the NYT?
Strange how you have to give your details to read an article on data collection. And where is the traditional slashdot warning for the NYT?
In their PDF Press Kit there is a mention of a patent:
PCT/FR 04/02082
Is this useful to anyone?
Found their website:
3dsolar.com
But good luck in finding technical details there
No, current devices are QVGA but this one is the same size but full VGA resolution (640x480).
According to the story submitter:
Hence according to i4u the resolution must be greater than QVGA otherwise it wouldn't be newsworthy.
Mozilla ain't showing the character set for most of the press release (for me), but in there is "640H×RGB×480V".
I suggest you make an appointment with the doctor.
There is two varients of the word. In British English your version is correct, however like many words the US chose something different. They also pronounce the word differently. See Wikipedia: http://en.wikipedia.org/wiki/Aluminum
Science has known how to do this for a long time, they call it "jumping".
For more information:
Jump Training Techniques
Of course it is. Install Reef::Coral from CPAN and execute the following perl script:
#!/usr/bin/perl
use Reef::Coral;
Reef::Coral->save("all");
Explosive, incendiary or poisonous (fake) currency would have a greater affect than curious bomblets. Everyone would pick them up with great harm to themselves, general morale and the economy.
IANAT (I Am Not A Terrorist)
The 200LX is far superior for graphing, due to speed and a high resolution display. These two things are the main weakness of my 48GX. Taking seconds to over a minute to plot graphs is not fun, especially when you get a blocky 141 x 64 (been a while, but it's about that) pixel graph. However, that blocky LCD provides pretty good contrast - perhaps why HP choose a lower resolution display.
If you can afford one of these the MPG isn't going to be an issue.
For myself, it usually means I've forgotton to switch from letter -> A4 before printing. PC stands for Paper Cassette.
And the manufacturer will still describe it as "almost silent".
It is real money. Quid is a common slang term for 1 pound sterling (GBP). Currently for 2 quid you'd get 3.57 US dollars.
Computing wouldn't be the same without multiple definitions of the same TLAs (Two Letter Acronym).