Each channel of 43 Mb/s (38 Mb/s usable) is shared among subscribers using TDMA or S-CDMA. The newer DOCSIS 3.0 standard allows channel bonding so that any one particular subscriber could use up to 4 channels simultaneously but that is not going to matter unless more channels are available.
Fluorescent tubes are still more efficient then high power LEDs so even with the converter losses and lower efficiency of space constrained cold cathode tubes, it will be still be a while before improved LED technology catches up. Where LEDs really pay off is in smaller form factors where the disadvantage of high voltage wiring and the space needed for the converter can not be accepted. The real gain will happen if any of the emissive technology displays can become viable since they do not have to discard a major portion of their light with filters.
I have noticed LED back lights becoming available in desktop LCD displays where power is not as significant an issue.
A highway hundreds of lanes wide. Most with pitfalls for potholes. Privately operated bridges and overpasses. No highway patrol. A couple of rent-a-cops on bicycles with broken whistles. 500 member vigilante posses with nuclear weapons. A minimum of 237 on ramps at every intersection.
No signs. Wanna get to Ensenada? Holler out the window at a passing truck to ask directions.
Ad hoc traffic laws. Some lanes would vote to make use by a single-occupant-vehicle a capital offense on Monday through Friday between 7:00 and 9:00. Other lanes would just shoot you without a trial for talking on a car phone.
AOL would be a giant diesel-smoking bus with hundreds of ebola victims on board throwing dead wombats and rotten cabbage at the other cars, most of which have been assembled at home from kits. Some are built around 2.5 horsepower lawn mower engines with a top speed of nine miles an hour. Others burn nitroglycerin and idle at 120.
No license plates. World War II bomber nose art instead. Terrifying paintings of huge teeth or vampire eagles. Bumper mounted machine guns. Flip somebody the finger on this highway and get a white phosphorus grenade up your tailpipe. Flatbed trucks cruise around with anti-aircraft missile batteries to shoot down the traffic helicopter. Little kids on tricycles with squirt guns filled with hydrochloric acid switch lanes without warning.
No off ramps. None.
Author (maybe, it's hard to track down sources on the Net): Jim Wiedman
I have three of the older AMS cases which are quiet enough but the original fans did need to be replaced with higher quality ones. I am not sure about the newer enclosures but the old full sized ones conveniently have a standard 120 volt AC socket on the back chained with the IEC input. They are also designed to stack which some of the slimmer ones discourage by being oddly shaped.
One of the reasons external power bricks are so common is that they make UL certification easier.
With the advent of eSATA and port multiplication there should be some better options in the near future. What might be nice is a medium sized eSATA enclosure which uses a standard ATX power supply.
American Media Systems (http://www.american-media.com) produces 3.5 inch and 5.25 inch external USB and Firewire enclosures that have internal AC power supplies which at least gets rid of the power brick problem. They recently revamped their web site with Flash unfortunately.
Niven's short story "Rammer" which later became the novel "A World Out of Time" featured a galactic sized black hole at the center of the milky way . Admittedly, this was 10 years after "At the Core".
There is statutory law at both the state and federal level that protects certain civil rights from private party infringement but the application of constitutional rights to common law is more nebulous. A good example are the statutory protections against discrimination by race or sex.
The law blog "The Volokh Conspiracy" has a recent discussion on a court case involving an employer, United Parcel Service, firing an employee for having a firearm stored in his car which was not on company property:
It would be better to mark an invalid vote to prevent someone from marking the ballot with a valid choice. Spoiled ballots are not reported that I know of. One could always vote for a third party as a protest (Duverger's Law and all of that) but they are not always counted.
I am not clear on the exact mechanism but in the case of XP and Vista 32, the limit is caused by only allowing the page table size to support up to 32 bits of physical addressing even though PAE mode is supported or used for security. Since the memory mapped I/O has to be placed within the processor's physical address space, RAM can not fill all 4 GBytes. Interestingly enough, Microsoft shows Windows XP previous to SP2 supporting 4GBytes of RAM in PAE mode instead of just physical address space but I have never had the occasion to test it.
As you point out in your post, for 32 bit operating systems that use PAE and large page tables, all of the physical RAM can be made available in sections of up to 4 GBytes each.
So if the article was talking about 32-bit windows, the headline could read "max out your RAM" rather than "4 GB is the sweet spot."
That is what I assumed as well. As far as I have been able to find out, Vista 32 suffers from the same physical address limitations that XP does do to using a 32 bit page table in PAE mode.
I am actually not sure how the rectification efficiencies compare to the step down efficiencies since they are normally measured together. The rectification is something that can be done at a more central location on a large scale which is not possible for the last high to low voltage step down stage because of copper loss and wiring issues.
Part of the problem is that the power factor correction that often needs to be added to the AC to DC rectification stage itself has losses associated with it but simultaneously allows for higher AC transmission efficiency by lowering the RMS current for a given power.
The existing 120/240 volt AC supply will be replaced by a simpler DC to DC converter that will not need any power factor correction. Higher efficiency do to lack of the rectifier directly translates into a lower power dissipation in the servers themselves for a given power level. The high voltage DC distribution also simplifies UPS design which no longer requires an inverter. None of this affects the point of load regulation needed for high current low voltage power supplies.
Telecommunications equipment will no doubt stay with 48 volts DC at least where it interfaces with the copper distribution network.
48 volts DC is indeed too low because of the resistive losses. The server DC distribution standard being considered is actually much higher with 380 volts DC being a major candidate. I presume they are looking at a large scale active power factor corrected boost converter with input voltages from 208 to 277 volts AC outside of the server room.
I also live near St. Louis and have faced similar problems with the phone company provided DSL. The previous homeowner here put up with 384/128 DSL for years before me living with the excuse that he was just at the edge of the service area. When I got service through a third party, I had the same problem until noticing that every incoming line suffered from a dry open and that leaving the phone off the hook yielded the maximum performance available. 2 weeks of daily complaining about static on the line fixed it. The local phone company still "accidently" disconnects my DSL service periodically just to make sure I remember to call them.
An other difference is the absence of overhead lines in Europe, a higher up front investment but no issues with the weather.
We have the most powerful indigenous species of backhoe in existence. Underground lines would last until someone decided it was again time to dig up the streets.
I'm not sure how low a modem will go when doing retraining. If it could keep dropping the speed all the way down to 300baud FSK then I can't see any reason why it wouldn't be able to find a (low) speed where the tollerances are great enough to make it work. However, I'm pretty sure that modems will give up long before they get down to the really low speeds (doesn't retraining only happen for speeds over about 14.4Kbps?).
V.32 includes retraining and fallback down to 4800 bits/s and V.22 also supports retraining at a more limited level. I have never tried forcing rates lower then that through poor channel conditions. I suspect V.22 or V.22bis at 1200 bits/s would work just as well as V.21 at 300 bits/s but of course new equipment may not implement outdated signaling standards.
In any case, I think that trying to put a modulated data signal over VoIP is an incredibly daft idea - far better to either plug directly into the network, or demodulate it locally and then just send the raw demodulated data over the network and remodulate it on the other side.
I certainly agree but the problem is interfacing with legacy equipment. Given the current installed base and availability of relatively inexpensive large scale integration, it may well be worthwhile to include demodulation and modulation at the VOIP endpoints to handle legacy standards in a more comprehensive way then T.38 or G.711 currently do.
I wonder how much the lack of routable endpoint addresses in IPv4 will hold back native IP systems from replacing legacy FAX, security, and other devices that currently depend on the POTS. Internet security issues could of course be an even larger problem.
There shouldn't be any far-end crosstalk and echo since that should've been removed by the hybred. With POTS at both ends you basically have 2 places where echo can be introduced
How long is the FIR filter used for canceling echo and crosstalk then? I have always figured it was long enough to account for effects at the far end of the POTS connection (implying an analog to digital to analog connection) and that the shorter ones used for low latency modems limited their performance on longer latency calls. As you pointed out, connections that are digital on one side would be able to use a much shorter filter but I never specifically tested for this change. Did the 56K standards do something like this yielding lower latency?
I have always found hybrids to be a significant source of crosstalk and echo often contributing more then the physical transmission line itself. At least in the analog domain when used on a large scale, balancing them can be tricky although 8 bit performance should be easily attained. I would expect the use of ulaw and alaw to increase the sensitivity to this. Somewhere I have the design for an op-amp based hybrid which can be expanded into a multiport circulator that gives better then 8 bit performance into the VHF range.
1. Compression (this isn't a problem if you use G.711)
This is what I thought when I first looked into the problem but I found many cases where even when compression was disabled and high bit rates above 64 kbit/s were used, the modem to modem connection over VOIP failed. Retraining would normally continue indefinitely until the connection was dropped. This seemed to coincide with variable connection latency.
2. Packet loss - VoIP systems can usually work like GSM and fill in for missing audio packets. This helps reduce how noticable the loss is to humans, but for modems it just won't cut it. 3. Jitter - if a packet arrives too late then it's going to be handled the same way as a lost packet.
I wonder if the older protocols like the later MNP ones designed for satellite and cell phone connections would help here. I am unclear if their techniques were included in later standards.
4. PCM timing stability - even if you have no dropped packets, the stability of the PCM timing is not going to be anywhere near as good as you get on POTS. The sender and receiver's timers aren't synchronised so the receiver has to adjust its output to compensate for the data arriving marginally too quickly or slowly. This might be done by adjusting the actual PCM timing, but more frequently it will just delete a sample or insert an interpolated sample every so often.
Would this actually cause a high speed modem to fail though or just negotiate a lower speed? I never tried to test for this owing to the difficulty involved and after I found that continuously changing the network latency would clobber the connection anyway.
I don't know of too many U.S. VoIP providers that *don't* support G.711u, which is your best bet for fax/modem stuff anyway.
The problem here is that older and most current analog modems and fax machines make some simplifying assumptions about the physical link which are violated by VOIP. For instance, the latency itself even if constant may exceed the length of the FIR filter used to adjust for far end crosstalk and echo.
I had not considered that the survey would have such a flaw. Thanks for pointing it out.
Assuming it is the case that he only wants to ban automatic weapons, I still would not vote for him because of the specious thinking involved. How many crimes involve the use of automatic weapons all of which are illegal in Illinois anyway except for law enforcement? Is that the same number of crimes that involve.50 caliber rifles in California? The Illinois state police made a lot of noise about arresting Missouri CCW holders after the law change so I try not to dally in the stygian abyss between St. Louis, Missouri and Louisville, Kentucky.
I should have said that head and piston changes would preclude using low octane gas in the future but that a supercharger (or turbocharger, I meant to be inclusive) would not. My own personal experience has been with milling the head, improving the air flow, and adjusting the timing but that was before E85 was available. I actually have an E85 compatible GMC pickup truck but have never had the occasion to take advantage of that feature. Gas mileage is suppose to be 20 to 30 percent poorer and every time I have checked the miles per dollar it has been at best a break even proposition.
E85 has an octane rating of 105. Beyond any control system adjustments that would have to be made I suspect swapping out the pistons and possibly the head would be necessary to take maximum advantage of it unless you have a supercharger at the cost of not being able to use low octane fuel.
Slashdot Mirror
Each channel of 43 Mb/s (38 Mb/s usable) is shared among subscribers using TDMA or S-CDMA. The newer DOCSIS 3.0 standard allows channel bonding so that any one particular subscriber could use up to 4 channels simultaneously but that is not going to matter unless more channels are available.
Fluorescent tubes are still more efficient then high power LEDs so even with the converter losses and lower efficiency of space constrained cold cathode tubes, it will be still be a while before improved LED technology catches up. Where LEDs really pay off is in smaller form factors where the disadvantage of high voltage wiring and the space needed for the converter can not be accepted. The real gain will happen if any of the emissive technology displays can become viable since they do not have to discard a major portion of their light with filters.
I have noticed LED back lights becoming available in desktop LCD displays where power is not as significant an issue.
A highway hundreds of lanes wide. Most with pitfalls for potholes. Privately operated bridges and overpasses. No highway patrol. A couple of rent-a-cops on bicycles with broken whistles. 500 member vigilante posses with nuclear weapons. A minimum of 237 on ramps at every intersection.
No signs. Wanna get to Ensenada? Holler out the window at a passing truck to ask directions.
Ad hoc traffic laws. Some lanes would vote to make use by a single-occupant-vehicle a capital offense on Monday through Friday between 7:00 and 9:00. Other lanes would just shoot you without a trial for talking on a car phone.
AOL would be a giant diesel-smoking bus with hundreds of ebola victims on board throwing dead wombats and rotten cabbage at the other cars, most of which have been assembled at home from kits. Some are built around 2.5 horsepower lawn mower engines with a top speed of nine miles an hour. Others burn nitroglycerin and idle at 120.
No license plates. World War II bomber nose art instead. Terrifying paintings of huge teeth or vampire eagles. Bumper mounted machine guns. Flip somebody the finger on this highway and get a white phosphorus grenade up your tailpipe. Flatbed trucks cruise around with anti-aircraft missile batteries to shoot down the traffic helicopter. Little kids on tricycles with squirt guns filled with hydrochloric acid switch lanes without warning.
No off ramps. None.
Author (maybe, it's hard to track down sources on the Net): Jim Wiedman
I have three of the older AMS cases which are quiet enough but the original fans did need to be replaced with higher quality ones. I am not sure about the newer enclosures but the old full sized ones conveniently have a standard 120 volt AC socket on the back chained with the IEC input. They are also designed to stack which some of the slimmer ones discourage by being oddly shaped.
One of the reasons external power bricks are so common is that they make UL certification easier.
With the advent of eSATA and port multiplication there should be some better options in the near future. What might be nice is a medium sized eSATA enclosure which uses a standard ATX power supply.
American Media Systems (http://www.american-media.com) produces 3.5 inch and 5.25 inch external USB and Firewire enclosures that have internal AC power supplies which at least gets rid of the power brick problem. They recently revamped their web site with Flash unfortunately.
I am so used to hating her that I am taken aback by her reported statement.
Niven's short story "Rammer" which later became the novel "A World Out of Time" featured a galactic sized black hole at the center of the milky way . Admittedly, this was 10 years after "At the Core".
. . . So, what exactly do you want?
There is statutory law at both the state and federal level that protects certain civil rights from private party infringement but the application of constitutional rights to common law is more nebulous. A good example are the statutory protections against discrimination by race or sex.
The law blog "The Volokh Conspiracy" has a recent discussion on a court case involving an employer, United Parcel Service, firing an employee for having a firearm stored in his car which was not on company property:
http://volokh.com/posts/1172173543.shtml
We are not the only nation to suffer from this problem: http://en.wikipedia.org/wiki/Duverger's_law
It would be better to mark an invalid vote to prevent someone from marking the ballot with a valid choice. Spoiled ballots are not reported that I know of. One could always vote for a third party as a protest (Duverger's Law and all of that) but they are not always counted.
the 4GB is the Process Memory limit.
I am not clear on the exact mechanism but in the case of XP and Vista 32, the limit is caused by only allowing the page table size to support up to 32 bits of physical addressing even though PAE mode is supported or used for security. Since the memory mapped I/O has to be placed within the processor's physical address space, RAM can not fill all 4 GBytes. Interestingly enough, Microsoft shows Windows XP previous to SP2 supporting 4GBytes of RAM in PAE mode instead of just physical address space but I have never had the occasion to test it.
As you point out in your post, for 32 bit operating systems that use PAE and large page tables, all of the physical RAM can be made available in sections of up to 4 GBytes each.
So if the article was talking about 32-bit windows, the headline could read "max out your RAM" rather than "4 GB is the sweet spot."
That is what I assumed as well. As far as I have been able to find out, Vista 32 suffers from the same physical address limitations that XP does do to using a 32 bit page table in PAE mode.
Otisburg.
I am actually not sure how the rectification efficiencies compare to the step down efficiencies since they are normally measured together. The rectification is something that can be done at a more central location on a large scale which is not possible for the last high to low voltage step down stage because of copper loss and wiring issues.
Part of the problem is that the power factor correction that often needs to be added to the AC to DC rectification stage itself has losses associated with it but simultaneously allows for higher AC transmission efficiency by lowering the RMS current for a given power.
The existing 120/240 volt AC supply will be replaced by a simpler DC to DC converter that will not need any power factor correction. Higher efficiency do to lack of the rectifier directly translates into a lower power dissipation in the servers themselves for a given power level. The high voltage DC distribution also simplifies UPS design which no longer requires an inverter. None of this affects the point of load regulation needed for high current low voltage power supplies.
Telecommunications equipment will no doubt stay with 48 volts DC at least where it interfaces with the copper distribution network.
48 volts DC is indeed too low because of the resistive losses. The server DC distribution standard being considered is actually much higher with 380 volts DC being a major candidate. I presume they are looking at a large scale active power factor corrected boost converter with input voltages from 208 to 277 volts AC outside of the server room.
I also live near St. Louis and have faced similar problems with the phone company provided DSL. The previous homeowner here put up with 384/128 DSL for years before me living with the excuse that he was just at the edge of the service area. When I got service through a third party, I had the same problem until noticing that every incoming line suffered from a dry open and that leaving the phone off the hook yielded the maximum performance available. 2 weeks of daily complaining about static on the line fixed it. The local phone company still "accidently" disconnects my DSL service periodically just to make sure I remember to call them.
An other difference is the absence of overhead lines in Europe, a higher up front investment but no issues with the weather.
We have the most powerful indigenous species of backhoe in existence. Underground lines would last until someone decided it was again time to dig up the streets.
I'm not sure how low a modem will go when doing retraining. If it could keep dropping the speed all the way down to 300baud FSK then I can't see any reason why it wouldn't be able to find a (low) speed where the tollerances are great enough to make it work. However, I'm pretty sure that modems will give up long before they get down to the really low speeds (doesn't retraining only happen for speeds over about 14.4Kbps?).
V.32 includes retraining and fallback down to 4800 bits/s and V.22 also supports retraining at a more limited level. I have never tried forcing rates lower then that through poor channel conditions. I suspect V.22 or V.22bis at 1200 bits/s would work just as well as V.21 at 300 bits/s but of course new equipment may not implement outdated signaling standards.
In any case, I think that trying to put a modulated data signal over VoIP is an incredibly daft idea - far better to either plug directly into the network, or demodulate it locally and then just send the raw demodulated data over the network and remodulate it on the other side.
I certainly agree but the problem is interfacing with legacy equipment. Given the current installed base and availability of relatively inexpensive large scale integration, it may well be worthwhile to include demodulation and modulation at the VOIP endpoints to handle legacy standards in a more comprehensive way then T.38 or G.711 currently do.
I wonder how much the lack of routable endpoint addresses in IPv4 will hold back native IP systems from replacing legacy FAX, security, and other devices that currently depend on the POTS. Internet security issues could of course be an even larger problem.
There shouldn't be any far-end crosstalk and echo since that should've been removed by the hybred. With POTS at both ends you basically have 2 places where echo can be introduced
How long is the FIR filter used for canceling echo and crosstalk then? I have always figured it was long enough to account for effects at the far end of the POTS connection (implying an analog to digital to analog connection) and that the shorter ones used for low latency modems limited their performance on longer latency calls. As you pointed out, connections that are digital on one side would be able to use a much shorter filter but I never specifically tested for this change. Did the 56K standards do something like this yielding lower latency?
I have always found hybrids to be a significant source of crosstalk and echo often contributing more then the physical transmission line itself. At least in the analog domain when used on a large scale, balancing them can be tricky although 8 bit performance should be easily attained. I would expect the use of ulaw and alaw to increase the sensitivity to this. Somewhere I have the design for an op-amp based hybrid which can be expanded into a multiport circulator that gives better then 8 bit performance into the VHF range.
1. Compression (this isn't a problem if you use G.711)
This is what I thought when I first looked into the problem but I found many cases where even when compression was disabled and high bit rates above 64 kbit/s were used, the modem to modem connection over VOIP failed. Retraining would normally continue indefinitely until the connection was dropped. This seemed to coincide with variable connection latency.
2. Packet loss - VoIP systems can usually work like GSM and fill in for missing audio packets. This helps reduce how noticable the loss is to humans, but for modems it just won't cut it.
3. Jitter - if a packet arrives too late then it's going to be handled the same way as a lost packet.
I wonder if the older protocols like the later MNP ones designed for satellite and cell phone connections would help here. I am unclear if their techniques were included in later standards.
4. PCM timing stability - even if you have no dropped packets, the stability of the PCM timing is not going to be anywhere near as good as you get on POTS. The sender and receiver's timers aren't synchronised so the receiver has to adjust its output to compensate for the data arriving marginally too quickly or slowly. This might be done by adjusting the actual PCM timing, but more frequently it will just delete a sample or insert an interpolated sample every so often.
Would this actually cause a high speed modem to fail though or just negotiate a lower speed? I never tried to test for this owing to the difficulty involved and after I found that continuously changing the network latency would clobber the connection anyway.
I don't know of too many U.S. VoIP providers that *don't* support G.711u, which is your best bet for fax/modem stuff anyway.
The problem here is that older and most current analog modems and fax machines make some simplifying assumptions about the physical link which are violated by VOIP. For instance, the latency itself even if constant may exceed the length of the FIR filter used to adjust for far end crosstalk and echo.
I had not considered that the survey would have such a flaw. Thanks for pointing it out.
.50 caliber rifles in California? The Illinois state police made a lot of noise about arresting Missouri CCW holders after the law change so I try not to dally in the stygian abyss between St. Louis, Missouri and Louisville, Kentucky.
Assuming it is the case that he only wants to ban automatic weapons, I still would not vote for him because of the specious thinking involved. How many crimes involve the use of automatic weapons all of which are illegal in Illinois anyway except for law enforcement? Is that the same number of crimes that involve
I should have said that head and piston changes would preclude using low octane gas in the future but that a supercharger (or turbocharger, I meant to be inclusive) would not. My own personal experience has been with milling the head, improving the air flow, and adjusting the timing but that was before E85 was available. I actually have an E85 compatible GMC pickup truck but have never had the occasion to take advantage of that feature. Gas mileage is suppose to be 20 to 30 percent poorer and every time I have checked the miles per dollar it has been at best a break even proposition.
E85 has an octane rating of 105. Beyond any control system adjustments that would have to be made I suspect swapping out the pistons and possibly the head would be necessary to take maximum advantage of it unless you have a supercharger at the cost of not being able to use low octane fuel.