Except that guerilla warfare doesn't really work anymore (against US forces, anyway). Night vision and IR change the equation dramatically by taking away the cover of darkness and eliminating any cover that the guerillas might have in their retreat (Apache helicopters are covered in IR optics and could make short work of retreating troops).
Buying the chips might be cost-effective if all you need is a basic ARM, but licensing an ARM core to incorporate into a custom design requires a $0.50 royalty on every every chip sold. This is fine for low-volume devices, but when you plan on selling 10 million chips, its cheaper to fork over the million $ for a MIPS design and sell the chips royalty free. Until ARM changes their licensing agreement they will be more or less locked out of the high-volume custom chip market.
"The decision to cut employees from a company is always not a very palatable issue for those who get to make the decision. I'm sure many get sleepless nights before and after making one. I've read somewhere this even ranks high among the causes of heart attacks in executives who need to make these decisions."
Absolutely man. My company had a very small layoff (less than 5%), and most everybody who was laid off was, though it isn't nice to say, deadwood - people we shouldn't have hired in the first place and weren't makin a positive contribution. Even still, all of our directors and chief officers were incredibly depressed about what they did. The director of my group came around, nearly in tears, to tell us what had happened. It wasn't an easy thing for him to do.
What worries me is how people on this board assume that it is so easy to cut headcount. It worries me because if they ever find themselves in that position, they may become as emotionally detached from the situation as they picture the current corporate masters being. It isn't an easy decision, and it shouldn't be. Sometimes it is the right thing to do, but even the most soulless CEO is aware of the fact that he is causing major disruption to people.
However, I haven't looked at the 25x that closely (basically, just what was in this article). The big bottle neck for alot of DSP tasks (most notably, filtering), is that you need a ton of on-chip, non swappable memory, because an adaptive FIR requires being able to read and write lots of coefficients very quickly.
If they have enough on-chip memory to support large coefficient tables and enough on-chip code space to run the code at full speed, they could indeed be very useful in this regard. If not, perhaps a licensed version giving these capabilities could become widely used.
In my job, a lot of what I do is researching new ideas that we could put into our software/firmware at some future time, propose as a standard, get patents, etc. This is enormously fun, and has no real time constraints.
However, what keeps us from losing customers is tweaking our current releases to meet their demands. This is somewhat less fun, and has definite time constraints.
Without my manager doing his job and helping me out setting priorities, I would be doing much more of the former, and a lot less of the latter. And the end result may be a loss of business.
Tim
Re:It's been said before...
on
More WTC News
·
· Score: 1
For plainclothes officers to act as a deterent, their presence has to be publicized, but their identity unknown. In other words, let the terrorist knows that there is an armed air-marshall on the plane, but they cannot distinguish the air-marshall from the 100+ other passengers, and therefore do not know who to attack first.
Re:MSNBC Article on BinLaden and CIA
on
More On Tragedy
·
· Score: 1
Exactly.
At the time, it was The Right Thing(tm).
At the time, *anybody* was a preferable ruler to the USSR. The muhajedin were fighting the Soviets already. They were getting slaughtered by Soviet Hind helicopters. We trained them and gave them anti-aircraft missiles.
Now, perhas you could claim that we should have helped them rebuild the country after the fighting was over, but for the most part we were shocked by the brutal civil war that followed the Soviet withdrawal. With no sides to take (good ones, anyway) the US sat it out and ended up getting the Taliban. That's just the way it goes.
Re:Change the rules, be realistic about conflict
on
More On Tragedy
·
· Score: 1
Iraq hardly rearmed itself. There is an order of magnitude more tanks strewn about the desert and airplanes permanently grounded on wrecked runways than are currently functioning in the Iraqui military. There were more tanks detroyed on the road of death than there were in Kursk.
Serbia, on the other hand was pretty adept at hiding their forces. We did successfully immobilize them (which is good, a tank that can't move is akin to an unloaded gun: a good blunt object but not a serious weapon), but high-altitude bombing proved rather incapable of actually identifying and destroying them (which is bad, we should have gotten down and dirty with apaches).
"And besides what right did we have to attack Iraq? Because we wanted to controle the price Iraq could charge for THEIR oil. If you think it was to save the poor Kuwaitis, where were we when the Hutu were commiting genocide against the Tutsi in Rowanda? At it's height the Tutsi were slaughtered at the greatest rate in history and we didn't lift a finger"
Repeat after me:
In the gulf war we could do something, in Rwanda we could do nothing.
Whatever the motivations for the GUlf War, blatantly taking over a neighboring country is way out of bounds. The US had friendly bases nearby, and the Saudis were very willing to help. We could do something here and we did.
Rwanda would have made Vietnam look like a teaparty. It is heavily forested, the US had no nearby bases, and "lifting a finger" would have literally meant genocide on a mass scale. Nothing short of anihilating both sides would have solved that situation. There were no good-guys, there was nobody it was even morally palatable to side with in the interests of peace. There were very good reasons we stayed out of Rwanda.
Tim
Re:What can be done about terrorism?
on
More On Tragedy
·
· Score: 1
"How in the world does military aid to Israel prevent the acquisition of a nuclear arsenal by Iraq? "
During the early 80's, I believe, Iraq was building a nuclear reactor. Israel promptly destroyed it. Iraq has yet to try again (as far we know).
And, from somewhere within me, a rather small (though somewhat built) computer geek, I knocked his ass to the ground and ran like hell (fortunately, I lived about 100 ft away). I must say, the adrenaline rush, and the actions I had taken left me shaking for 30+ hours during which I couldn't sleep. It was an absolutely stupid thing to do and I couldn't believe I did it, considering the potential consequences.
If I were in a situation like those on the hojacked plane, where I would have had more than an instant to think about it, however, I could almost assure you I wouldn't have done the same thing.
Hehe. At work we recently had some serious XT/coupling problems after moving to.13um. Fun stuff to debug. Nothing beats having certain opcodes only work after a nop and lacking a good explanation as to why (why, dear god, why?:)
Anyway, OT, you sound like a chip-designer (I'm a firmware guy, but end up helping our LSI guys when we get new chips), so I was wondering if you had any tools to recommend for finding these vicious bastard couplings? We have one, but it blows, and with a 10-million gate chip its difficult for the layout engineers to catch everything the tool missed.
Q: How well does your average military fighter jet fly without the software functioning 100%?
A: Like a freakin' rock.
When software needs to be written properly, it usually is.
Space Shuttles, traffic-signals, air traffic control, and anti-lock brakes are all examples of things that completely (or nearly so) dependent on computers functioning properly to prevent people from dying and as a result, the systems cost more, but they work damn good.
Dude, I don't have a degree either, but I got hired at a modem-chip company almost exclusively because I did send them code. Not that they asked for it, but I decided that I'd improve my chances by sending in some code.
Along with my resume, I sent a 320x240 space-invaders program I wrote in straight 80x86 assembly. It wasn't a large, professional style project or anything (only about 750 lines of code, not counting art) but it demonstrated that I knew my shit. I program DSPs and small embedded devices now, and I have my insanely optimized little game to thank for it:)
Anyway, the point is, even if companies don't ask to see a portfolio, show some initiative and send one anyway: it couldn't hurt. And besides, its not likely that an HR person could properly evaluate it so it improves your chances of an actual engineer or manager seeing your resume.
Great explanation! Much more clear than the one I was going to give:)
Vector units are extraordinarily fast at certain tasks. I work with a custom DSP that uses a vector processor to do FIR filtering, and the amount of processing it does is mind blowing. We clock it at somewhere between 80-120 MHz (depending on application), and at the top end of that range it gets nearly a billion ops per second.
Now, this does come with some drawbacks. First of all, it requires a tremendous amount of silicon to do properly, making development extremely expensive. Not to mention, that with all that logic running simultaneously, power consumption can become an issue as well. Secondly, it is a royal ain in the ass to program (or write a compiler for). When you have 8 operations per instruction word, making efficient use of that processing power involves writing some ugly, ugly code.
Except that for CD drives, it may work pretty well.
CD drives have to be very precisely balanced. If you were to attempt to modify one, its likely that the next thing you'll do is head to the store to buy a new drive because you completely fux0red the old one.
For an experiment, put a wad of tape on the outside of your CD drive: It'll sound like a jet airplane taking off. (I did this once to try to dampen the noise of my 48x drive, and it had the completely opposite effect. After my ears stopped bleeding, I figured out why that wasn't such a good idea:)
Ok, I'll come clean here: I write DSP firmware for DSL modems. If you want, you can send me an email at thetimdog@hotmail.com and I'll confirm it from my work email (no, I will not give that out on messageboards).
FEC is, in the strictest sense, mandatory, but it is easy enough to just set the number of redundancy bytes to zero, in which case FEC takes up zero bandwidth. Also, FEC without some interleaving is damn near pointless due to the way impulse noise works and the way DMT is modulated, so when running fast path, there are generaly zero redundancy bytes and no interleaving.
If you are getting any significant latency to the nexth hop when using fast-path DMT, that is the fault of the provider, not the DSL impelmentation.
The interleave latency you are refering to is a byproduct of the Reed-Solomon Forward Error Correction (FEC) used on DSL lines.
However, it certainly isn't provisioned with 50 ms of latency for data lines. Video lines, maybe, but there is little benefit to be seen above 32 ms, and no-one really suports anything above 16 ms (according to the G.DMT standard, anything above 16 ms is optional, so it is widely ignored).
By turning off the FEC (as is doen in most of North America), there is very, very little line latency. 1 ms, maybe. More likely something along the lines of 100 us.
PPPoE shouldn't affect the latency in any kind of noticeable manner, 250 us tops (ie, it'll never do more than require a single extra data symbol (250 us symbol period) to be sent for a given packet), but it will eat into bandwidth somewhat consistently.
As far as VoIP, both the FEC and FEC-less channels are available simultaneously on an ADSL line, so voice can be sent over the FEC-less one (for lower latency) and data over the FEC channel (for better reach/rate).
The difference between a dial-up and DSL is that the DSL physical line is always connected, only the protocol needs to "dial-up" (horrible term for it in this case, all it does is re-initiate the network protocol). With a 56K, a re-dial actually requires, you know, a physical dial-up.
Tim
Re:Is better TV definition needed ?
on
The Joys of HDTV
·
· Score: 1
"Such a nice TV, just a pain to get up a flight of stairs."
Hehe, you too?
I couldn't move for a week after carrying in my 32 inch wega (with a W, btw) myself. When I move out of my apartment, I'm definitely calling in some reinforcements:)
Heh, you play a whole different kind of game than I do.
In my view, the key to deathmatch is weapon control. I know I'm not much fun to play against (according to my fraternity brothers from college), but I can absolutely dominate a game by never letting a weapon sit on the ground. I do not (!) camp, I tend to run around like a madman grabbing all the power ups laying around and tend to play a strict denial game.
This allows me to make up for my complete lack of skills by being utterly abusive. I never involve myself in a fair fire-fight. I maintain control of the powerful weapons, and when I kill somebody, I know that they are suddenly at their weakest so I make a run to a spawn point in hopes of waxing them again. Failing that, I head to the weapon I think they are heading for and try to pick them off while they blindly dash for the RL.
Its a completely different gamestyle, and I think the combination of skills and strategy could make for a fearsome (perhaps, professional level) player. You should really adjust your strategy, and if you are as tactically good as you say you are, well fuck, who needs a day job:)
I'll agree with joel on this one: in most cases, a complete rewrite is a bad idea. You abandon everything that has been done, working, and tested.
However, like all other SE rules, there are exceptions. An 8051 embedded device that I worked on had the core state-machine written in C, while all the modules were written in assembly. At that point, we had something like 34 bytes free and a customer made a massive feature request, so I was left removing all of that bloated C and implementing "the world's most unreadable FSM" (tm) just so we could jam the extra feature into memory. It was about as complete as rewrites get.
Slashdot Mirror
It only worked for ghandi because the British were inherently rational and peaceful.
Would non-violence have worked as well if the British were willing to kill 10,000 people a day until the population surrendered?
Somehow, I don't think so.
Tim
Except that guerilla warfare doesn't really work anymore (against US forces, anyway). Night vision and IR change the equation dramatically by taking away the cover of darkness and eliminating any cover that the guerillas might have in their retreat (Apache helicopters are covered in IR optics and could make short work of retreating troops).
Tim
The biggest problem with the ARM core is cost.
Buying the chips might be cost-effective if all you need is a basic ARM, but licensing an ARM core to incorporate into a custom design requires a $0.50 royalty on every every chip sold. This is fine for low-volume devices, but when you plan on selling 10 million chips, its cheaper to fork over the million $ for a MIPS design and sell the chips royalty free. Until ARM changes their licensing agreement they will be more or less locked out of the high-volume custom chip market.
Tim
"The decision to cut employees from a company is always not a very palatable issue for those who get to make the decision. I'm sure many get sleepless nights before and after making one. I've read somewhere this even ranks high among the causes of heart attacks in executives who need to make these decisions."
Absolutely man. My company had a very small layoff (less than 5%), and most everybody who was laid off was, though it isn't nice to say, deadwood - people we shouldn't have hired in the first place and weren't makin a positive contribution. Even still, all of our directors and chief officers were incredibly depressed about what they did. The director of my group came around, nearly in tears, to tell us what had happened. It wasn't an easy thing for him to do.
What worries me is how people on this board assume that it is so easy to cut headcount. It worries me because if they ever find themselves in that position, they may become as emotionally detached from the situation as they picture the current corporate masters being. It isn't an easy decision, and it shouldn't be. Sometimes it is the right thing to do, but even the most soulless CEO is aware of the fact that he is causing major disruption to people.
Tim
As a DSP programmer, I'm inclined to agree.
However, I haven't looked at the 25x that closely (basically, just what was in this article). The big bottle neck for alot of DSP tasks (most notably, filtering), is that you need a ton of on-chip, non swappable memory, because an adaptive FIR requires being able to read and write lots of coefficients very quickly.
If they have enough on-chip memory to support large coefficient tables and enough on-chip code space to run the code at full speed, they could indeed be very useful in this regard. If not, perhaps a licensed version giving these capabilities could become widely used.
Tim
Its not ncessarily flame bait.
In my job, a lot of what I do is researching new ideas that we could put into our software/firmware at some future time, propose as a standard, get patents, etc. This is enormously fun, and has no real time constraints.
However, what keeps us from losing customers is tweaking our current releases to meet their demands. This is somewhat less fun, and has definite time constraints.
Without my manager doing his job and helping me out setting priorities, I would be doing much more of the former, and a lot less of the latter. And the end result may be a loss of business.
Tim
For plainclothes officers to act as a deterent, their presence has to be publicized, but their identity unknown. In other words, let the terrorist knows that there is an armed air-marshall on the plane, but they cannot distinguish the air-marshall from the 100+ other passengers, and therefore do not know who to attack first.
Exactly.
At the time, it was The Right Thing(tm).
At the time, *anybody* was a preferable ruler to the USSR. The muhajedin were fighting the Soviets already. They were getting slaughtered by Soviet Hind helicopters. We trained them and gave them anti-aircraft missiles.
Now, perhas you could claim that we should have helped them rebuild the country after the fighting was over, but for the most part we were shocked by the brutal civil war that followed the Soviet withdrawal. With no sides to take (good ones, anyway) the US sat it out and ended up getting the Taliban. That's just the way it goes.
Iraq hardly rearmed itself. There is an order of magnitude more tanks strewn about the desert and airplanes permanently grounded on wrecked runways than are currently functioning in the Iraqui military. There were more tanks detroyed on the road of death than there were in Kursk.
Serbia, on the other hand was pretty adept at hiding their forces. We did successfully immobilize them (which is good, a tank that can't move is akin to an unloaded gun: a good blunt object but not a serious weapon), but high-altitude bombing proved rather incapable of actually identifying and destroying them (which is bad, we should have gotten down and dirty with apaches).
Tim
"And besides what right did we have to attack Iraq? Because we wanted to controle the price Iraq could charge for THEIR oil. If you think it was to save the poor Kuwaitis, where were we when the Hutu were commiting genocide against the Tutsi in Rowanda? At it's height the Tutsi were slaughtered at the greatest rate in history and we didn't lift a finger"
Repeat after me:
In the gulf war we could do something, in Rwanda we could do nothing.
Whatever the motivations for the GUlf War, blatantly taking over a neighboring country is way out of bounds. The US had friendly bases nearby, and the Saudis were very willing to help. We could do something here and we did.
Rwanda would have made Vietnam look like a teaparty. It is heavily forested, the US had no nearby bases, and "lifting a finger" would have literally meant genocide on a mass scale. Nothing short of anihilating both sides would have solved that situation. There were no good-guys, there was nobody it was even morally palatable to side with in the interests of peace. There were very good reasons we stayed out of Rwanda.
Tim
"How in the world does military aid to Israel prevent the acquisition of a nuclear arsenal by Iraq? "
During the early 80's, I believe, Iraq was building a nuclear reactor. Israel promptly destroyed it. Iraq has yet to try again (as far we know).
Tim
I, also, have been assaulted at knifepoint.
And, from somewhere within me, a rather small (though somewhat built) computer geek, I knocked his ass to the ground and ran like hell (fortunately, I lived about 100 ft away). I must say, the adrenaline rush, and the actions I had taken left me shaking for 30+ hours during which I couldn't sleep. It was an absolutely stupid thing to do and I couldn't believe I did it, considering the potential consequences.
If I were in a situation like those on the hojacked plane, where I would have had more than an instant to think about it, however, I could almost assure you I wouldn't have done the same thing.
Tim
Oh so its the quality of coffee they drink. Wow, I never knew bad coffee could...
Oh, I get it...
Hey! What exactly are you trying to say here?
:)
Whee!
.13um. Fun stuff to debug. Nothing beats having certain opcodes only work after a nop and lacking a good explanation as to why (why, dear god, why? :)
Hehe. At work we recently had some serious XT/coupling problems after moving to
Anyway, OT, you sound like a chip-designer (I'm a firmware guy, but end up helping our LSI guys when we get new chips), so I was wondering if you had any tools to recommend for finding these vicious bastard couplings? We have one, but it blows, and with a 10-million gate chip its difficult for the layout engineers to catch everything the tool missed.
Tim
Q: How well does your average military fighter jet fly without the software functioning 100%?
A: Like a freakin' rock.
When software needs to be written properly, it usually is.
Space Shuttles, traffic-signals, air traffic control, and anti-lock brakes are all examples of things that completely (or nearly so) dependent on computers functioning properly to prevent people from dying and as a result, the systems cost more, but they work damn good.
Tim
Dude, I don't have a degree either, but I got hired at a modem-chip company almost exclusively because I did send them code. Not that they asked for it, but I decided that I'd improve my chances by sending in some code.
:)
Along with my resume, I sent a 320x240 space-invaders program I wrote in straight 80x86 assembly. It wasn't a large, professional style project or anything (only about 750 lines of code, not counting art) but it demonstrated that I knew my shit. I program DSPs and small embedded devices now, and I have my insanely optimized little game to thank for it
Anyway, the point is, even if companies don't ask to see a portfolio, show some initiative and send one anyway: it couldn't hurt. And besides, its not likely that an HR person could properly evaluate it so it improves your chances of an actual engineer or manager seeing your resume.
Tim
Great explanation! Much more clear than the one I was going to give :)
Vector units are extraordinarily fast at certain tasks. I work with a custom DSP that uses a vector processor to do FIR filtering, and the amount of processing it does is mind blowing. We clock it at somewhere between 80-120 MHz (depending on application), and at the top end of that range it gets nearly a billion ops per second.
Now, this does come with some drawbacks. First of all, it requires a tremendous amount of silicon to do properly, making development extremely expensive. Not to mention, that with all that logic running simultaneously, power consumption can become an issue as well. Secondly, it is a royal ain in the ass to program (or write a compiler for). When you have 8 operations per instruction word, making efficient use of that processing power involves writing some ugly, ugly code.
Tim
Except that for CD drives, it may work pretty well.
:)
CD drives have to be very precisely balanced. If you were to attempt to modify one, its likely that the next thing you'll do is head to the store to buy a new drive because you completely fux0red the old one.
For an experiment, put a wad of tape on the outside of your CD drive: It'll sound like a jet airplane taking off. (I did this once to try to dampen the noise of my 48x drive, and it had the completely opposite effect. After my ears stopped bleeding, I figured out why that wasn't such a good idea
Tim
Ok, I'll come clean here: I write DSP firmware for DSL modems. If you want, you can send me an email at thetimdog@hotmail.com and I'll confirm it from my work email (no, I will not give that out on messageboards).
FEC is, in the strictest sense, mandatory, but it is easy enough to just set the number of redundancy bytes to zero, in which case FEC takes up zero bandwidth. Also, FEC without some interleaving is damn near pointless due to the way impulse noise works and the way DMT is modulated, so when running fast path, there are generaly zero redundancy bytes and no interleaving.
If you are getting any significant latency to the nexth hop when using fast-path DMT, that is the fault of the provider, not the DSL impelmentation.
Tim
The interleave latency you are refering to is a byproduct of the Reed-Solomon Forward Error Correction (FEC) used on DSL lines.
However, it certainly isn't provisioned with 50 ms of latency for data lines. Video lines, maybe, but there is little benefit to be seen above 32 ms, and no-one really suports anything above 16 ms (according to the G.DMT standard, anything above 16 ms is optional, so it is widely ignored).
By turning off the FEC (as is doen in most of North America), there is very, very little line latency. 1 ms, maybe. More likely something along the lines of 100 us.
PPPoE shouldn't affect the latency in any kind of noticeable manner, 250 us tops (ie, it'll never do more than require a single extra data symbol (250 us symbol period) to be sent for a given packet), but it will eat into bandwidth somewhat consistently.
As far as VoIP, both the FEC and FEC-less channels are available simultaneously on an ADSL line, so voice can be sent over the FEC-less one (for lower latency) and data over the FEC channel (for better reach/rate).
Tim
The difference between a dial-up and DSL is that the DSL physical line is always connected, only the protocol needs to "dial-up" (horrible term for it in this case, all it does is re-initiate the network protocol). With a 56K, a re-dial actually requires, you know, a physical dial-up.
Tim
"Such a nice TV, just a pain to get up a flight of stairs."
:)
Hehe, you too?
I couldn't move for a week after carrying in my 32 inch wega (with a W, btw) myself. When I move out of my apartment, I'm definitely calling in some reinforcements
Tim
Heh, you play a whole different kind of game than I do.
:)
In my view, the key to deathmatch is weapon control. I know I'm not much fun to play against (according to my fraternity brothers from college), but I can absolutely dominate a game by never letting a weapon sit on the ground. I do not (!) camp, I tend to run around like a madman grabbing all the power ups laying around and tend to play a strict denial game.
This allows me to make up for my complete lack of skills by being utterly abusive. I never involve myself in a fair fire-fight. I maintain control of the powerful weapons, and when I kill somebody, I know that they are suddenly at their weakest so I make a run to a spawn point in hopes of waxing them again. Failing that, I head to the weapon I think they are heading for and try to pick them off while they blindly dash for the RL.
Its a completely different gamestyle, and I think the combination of skills and strategy could make for a fearsome (perhaps, professional level) player. You should really adjust your strategy, and if you are as tactically good as you say you are, well fuck, who needs a day job
Tim
I'll agree with joel on this one: in most cases, a complete rewrite is a bad idea. You abandon everything that has been done, working, and tested.
However, like all other SE rules, there are exceptions. An 8051 embedded device that I worked on had the core state-machine written in C, while all the modules were written in assembly. At that point, we had something like 34 bytes free and a customer made a massive feature request, so I was left removing all of that bloated C and implementing "the world's most unreadable FSM" (tm) just so we could jam the extra feature into memory. It was about as complete as rewrites get.
Tim
I beliebe that the US military can use GPS on high-speed aircraft, but no one else can (unless they break the encryption, of course).
:)
When GPS was deployed, Uncle Sam was somewhat against the idea us giving the Soviets a free ICBM guidance system