The fact that people watch these over cooked bullhorns of bullshit, is a sign of how fucked our "free press is"
Yeah, I mean god forbid people be allowed to watch what *they* want to watch instead of what elitist fucks like yourself want to force them to watch...
I got into my current company after leaving college without a degree mostly by sending in a sapce-invaders game I wrote in 80x86 assembly language along with my resume. Of course, this was a device-driver and embedded systems job, so showing off some m@d 5ki11z (did I even do that right?:) ) with assembly language was particularly relevant and gave me a starting point in my interviews (and helped my resume stand out, too).
It never hurts to be a bit of a show-off during an interview. I do much different work now (DSP programming), so asking for a portfolio of actual code (as opposed to a thesis or something) would be rather difficult, but I still like it when a candidate likes to show off and demonstrates some pride in his accomplishments. I take it as a sign that he/she (who am I kidding, he:) ) will want to show that same pride in his production code and work to make it great (not perfect, perfect is the enemy of good enough) and on schedule rather than just going through the motions.
The problem with China is that the PRC doesn't allow Chinese banks (outside of Hong Kong) to hold foreign assets. Under this arrangement, improting foreign products can become painfully difficult for the population to buy imported products, but still export on the same terms.
Now, this arrangement may (and probably will, if they continue it indefinitely) lead to a consumer revolt, it doesn't effect their export prices or economy at all in the short-term (as for the long term, see Japan) as long as they keep the Yuan pegged to the US Dollar.
The problem with Indian and Chinese workers, as I see it (through my interactions with 30+ of each), is that their EE degree (in my specific case) is seen more as a ticket out than an undying passion.
This doesn't mean that there aren't very intelligent Indian or Chinese coders out there - In fact, I work with several of them. However, I also work with several that are reasonably intelligent people (they do have that MS or PhD after all) but are horrid developer because they lack the passion for for engineering.
Look at it this way - in the US or Western Europe (though to a degree less so), if you're a smart guy you can go into almost any field you want and bring home a decent salary. In India or China, your choices for bringing home mad bank are limited to getting a job in EE or CS. Being an accountant in China sucks ass, so even though running numbers may be your passion you're not going to do it if you feel you have some aptitude in programming or circuit design.
So what you end up with is not a less educated or intelligent workforce, but one that didn't go through a natural filtering system. We saw this here in the US during the.com days, but to a much lesser degree, because you could still earn a decent living being an accountant or finance guy or chemist or whatever if you had the capacity to do it. But in India and China, the pay-scales are so out of whack and have been for a while now, that there is a totally undermotivated, wrongly-educated workforce (though still certainly intelligent enough to handle the job) in place and its a crap-shoot when hiring.
Nyquist is the sample-rate guy (the Nyquist theorem).
Shannon capacity is the theoretical bit-rate you can stuff through a channel of a given physical bandwidth with a given signal to noise ratio:
C = W * Log2(1+S/N)
where C is capacity (bits per second), W is bandwidth (Hz), S is signal power, and N is noise power.
As far as Turbo Codes go, without getting too technical, its an extension of the principle that you can increase the efficiency of a channel while transmitting at a given power by attaching some redundant bits (redundant signal dimensions is probably a better way to look at it, though) to the signal. I'm not too familiar with the particulars of Turbo Coding, but it is a lot like Viterbi coding where these redundant bits are dependent on the data bits and when you detect an error (the redundant bit doesn't match the proper sequence), you back-trace through your data and find the most likely non-errored sequence and adjust your data bits accordingly.
Thank you for saving me the time of pointing out in excruciating detail that most (all, maybe) commercial modems wouldn't work if we needed 10x the Nyquist to reconstruct the waveform.
Really folks - you only need a touch more than than 2x the Nyquist. If the highest frequency we're interested in is 2 Mhz, then all we need to reconstruct a sine wave at that frequency is a sample rate of 4MHz + some infinitely small amount.
If you disagree, please take up your argument with the ghost of Claude Shannon.
It depends which kind of status meeting. If its just a meeting to fill in the PM, then yeah, 25 minutes is good. But if its more engineering oriented, taking a while isn't necessarily a bad thing.
For a project I was recently working on, we did a combo meeting - the first 15-30 minutes was spent filling in the PM on our progress, and the next 30-45 minutes was spent helping out people who were stuck on their current task. The second half wasn't mandatory, you could leave if you wanted to, but I think it was pretty useful given the kind of work we do (DSP programming on a proprietary chip) can at times resemble voodoo - sometimes other people have run across a similar situation in the past and can sum up a solution in 5 minutes where you might spend 5 weeks to come to the same answer.
Going back to Uni is also not an option (at least in my state, north-rhine westfalia, other states already did that or announced to do it in the next years), because it costs EUR 650, in addition to the ~ EUR 150-200 that it cost before, for "long-time" students
I'm sorry, this kills me - 850 Euro? $1100? And you're bitching and whining?
If you want to go back to university, take out a freaking loan and just do it! It isn't all that much - even if you went for four years, that's only 5K Euro in debt (after accumulating interest). You'll probably be paying around 120 Euro/month to pay that off in 5 years after finishing - I don't understand the big deal.
I took out about $60K in loans to go to school here in the US (and got about another $60K in gratis government assistance), and 4 years later they're about half paid off and I still net way more each month after making my $700 payments than I would have if I didn't take those loans. And I didn't even bother to graduate:)
So less whining, more action. If you want to do anything - make money, get more education, whatever - action always accomplishes more than whining. And don't worry about debt, there's more to life than having chickenshit money right now (like having a lot of money later:) oh and fun and family and friends and education and personal satisfaction and all that stuff too ).
That's not really true - Windows, for instance, runs in protected mode, where the chip intercepts memory access and hands it off to the OS for translation which then hands it back to the chip for execution. This way, the OS (or the old DOS memory managers, if you remember them) translate your address into a flat memory model. Under this system, each application gets its own little memory sandbox - In theory, I shouldn't be able to corrupt the memory of another program unless I have both cunning and malicious intent.
Now, what this means is that if I write a Win32 application-layer program that attempts to directly modify, say, the interrupt vector table, I'm not writing to the real interrupt vector table that the hardware uses when it receives physical interrupt signals, I'm writing to a version of it in my slice of memory, and what happens next is anybody's guess.
In otherwords, access to memory-mapped system resources like screen-buffers and interrupt vectors is arbitrated by the OS memory manager. What I want to do may work, it may not. I imagine Linux is more friendly in this regard, but I really don't know. DOS, on the otherhand, was anarchy - there was no memory manager (unless you intentionally ran one with your application) so all requests from the application layer were carried out without any intervention.
Screw you. I've been at this since the late 60's. Coding in assembler is fine for tinie, winie stuff, not applications which accomplish most of the world's functionality
Oh really?
How about the computer that makes your car engine work efficiently? Assembly.
The physical layer of virtually every communications device on the planet? Assembly.
Avionics systems? Quite a bit of assembly.
Higher level languages have their place - I use Matlab and C extensively - but in any system that is competing on cost (like modems) or has size/weight/power constraints (avionics, engines) assembly is the language of choice for production because it uses less silicon to get the same result.
Oh, and the guys writing the compilers had better have a good grasp of assembly as well if they plan on making decent optimizations.
When I started my current job, the compiler for one of our processing engines was only kinda-sorta done. Common instructions were supported, but most mode settings and some other things weren't. (Un)Fortunately, the compiler did let us hack in the correct opcode, leading to code like this:
------------
...stuff I can read...//I'll put a useful comment here
...stuff I can read...//and here
...stuff I can read...
0x6782;//comments?
0x9567;//we don't need no stinking comments!
0x8102;
...stuff I can read...
0x7230;//It should be completely obvious what this does, no comment here
...stuff I can read...
------------
Even as somebody who *prefers* to program on the metal, I would never want to revisit that hell.
First off, while I am not a Linux expert, I imagine that Linux would allow you more freedom to actually do stuff like mess with interrupt vectors and system resources (aasuming you are running at the correct privelege layer) whereas from my minimal exposure to Win32 driver-layer assembly, it seemed you are rather restricted in some regards.
IMHO, DOS is the perfect OS to learn assembly-language programming on - no restrictions at all (and fast reboots when you fuck up). But maybe I'm just biased because I program DSPs with no operating system whatsoever:)
Along those lines, I'm a former physics student (dropout) now doing DSP/digital-communications work at the physical layer.
I started college at an EE, switched to CS, switched to Physics, left, got a job doing testing, a couple days later (after a bit of showing off) changed into driver development, a month or two later turned into an embedded systems programmer, a year or so later moved into TC-layer development (layer 1/2 stuff), and after that sorta morphed into the DSP engineer I am now.
Life can take weird turns - I went full circle. So don't pigeonhole yourself.
I test drove an S-2000 last time I was looking for a car. Its pretty quick, its great with the top down, the speakers sucked (I hope they fixed that), but to me, the real problem was that it was insanely noisy.
The powerband is over 7000 RPM, so its got a very squeally nature to it. A high pitched whine. I couldn't stand it. Maybe if I bought one, I'd have built the internal filter for that frequency, but it could have driven me nuts, too.
If you include a couple of papers that you co-authored, we will pull them out and the engineers you interview with will have read them. And, personally, I will ask you about it.
So if its something you got a co-authorship on when helping out a crazy-genius professor in college but don't fully understand yourself, cop to that immediately when asked about it (the honesty is appreciated, and its a fairly common situation - it won't count against you, it will just realign expectations with reality). If you don't, I (and many of my colleagues) will expect you to be all-knowing god on that subject, and in some cases you may be interviewing with someone who *is* an all-knowing on the subject - in fact, if you have a co-authorship for a paper on, say, Delta-Sigma DAC techniques, you will be scheduled to interview with our resident delta-sigma guru. You may be able to play the bullshit with some of us who have a more basic understanding of foo, but the expert will catch you.
A few months ago, we received a 27 (!) page resume (single-spaced) for an entry-level DSP engineering position.
Now, admittedly, he was obviously a smart guy - with 80+ IEEE papers (listed individually, making for the bulk of that 27 pages). But there was just no way we could possibly hire a guy like that for an entry level position - for a senior researcher, fine, but we have enough of those. Seemingly, he was interested in doing actual engineering work but there was nothing there to indicate that he spent time doing anything other than churn out papers.
We normally schedule about 6 hours of interviews - this guy lasted 15 minutes before him and my boss realized they were looking for very different things:)
And here's a hint - if you have voluminous output in published journals (or the patent office - we get similar resumes stuffed with patents, though not to such an absurd degree), cite a couple that are relevant - for instance, we're a DSL company - all your papers on cell networks are nice (and semi-relevant), but if you highlighted your 3 papers on wireline communications that you co-authored and included a line about having 77 more papers published you would save everyone involved a lot of time - if we're that interested (for verification purposes or morbid curiousity), we'll go to IEEE.org or uspto.gov and search for your name. Listing that much individually just makes you seem arrogant. (on the otherhand, if you've only published a couple of things, go ahead and list them individually regardless of subject - it'll at least give us a good starting point in an interview and a chance to show-off).
Yes you can. You can define a binary system with however many bits you like. It may not look pretty in a hexadecimal representation, but there's no reason you can't do it.
In fact, I work on a processor in which one engine has 20 bit words, another has 18 bit words, and another does multiplies out to 29 bits (15 bit x 14 bit).
Don't they use unemployment claims to measure that?
No, they don't. Not the OECD numbers, anyways (which is the 6%).
There are two methods used:
1) Calling people up and asking if they are employed or looking for work.
2) Polling a realtively large sample of employers about their payroll-size. On a quarterly basis, this one isn't very good because it misses new startups, but over the long term its probably the more accurate number. Lately, this one has been 6%, and the first one has been somewhat lower.
Oh BTW, neat idea, but it's really hard to make inductors out of silicon.
I'm not an LSI guy (though I do work closely with our LSI team), but IIRC from a college course a million (ok, 4:) ) years ago, there is a way to simulate an inductor using capacitors and op-amps. I'm pretty sure that analog-LSI people do this all the time when they make our uber-cool programmable A/D-D/A chips with tunable filters and adaptive hybrids.
But from my fuzzy recollection, it seemed like really convoluted deign - you'd probably waste more power trying to simulate the inductor than you would save by having your sim-inductor.
I haven't seen the August numbers yet, but in July YahooBB! (well, SoftbankBB now) absolutely creamed NTT in new subscribers. It wasn't even close - something like 2-1 over NTT East and West combined.
Acca and E-Access, on the other hand, are in for a world of hurt precisely because they are small, and the even smaller providers are going to get hit worse. If you're really interested in the details, I'll explain further (I really shouldn't on a public BB) - email me: thetimdog@hotmail.com.
While I agree that the stuff you did sounds like cheating, you could also define it as asymmetric advantage.
Or, to put it back in the warfare setting: what's the difference between having the rifle fir without pulling the trigger and having an attack helicopter provide support? What's the difference between obscuring the sensor and damn good body armor? What's the difference between a god-mode cheat and divine intervention (ok, I'm going way out on a limb here)?
Well, let's analyze:
1) Totally defeats the training for the shooter, but gives the defender something more to think about. This should enhance the effect of the shootee's training - total suprise is the best way to go - don't even get into the enemy's field of fire.
2) Is pretty realistic. Modern body armor is pretty damn good and the MILES system of one-shot-one-kill doesn't necessarily apply. Both sides should be prepared for this and prepare accordingly.
3) Is just plain cheating. WTF was your CO thinking?
Err, DMT ADSL (which is like 99.9% of all ADSL connections), as per the G.992 standard, is most certainly a baseband signal using an OFDM modualtion with bin 0 defined at DC (each bin has a width of 4.3125 KHz). And while no actual data is transmitted below bin 6 (6*4.3125 KHz), the signal is processed in a baseband fashion (though I suppose it would be possible to demodulate and then decode, you'd have to be a real masochist to do so).
Not quite. Antenna diversity, while always a good thing, only applies to the receiver. MIMO is just what it says - more antennas at both the transmitter and the receiver.
You can check my earlier post on it. I'm not retyping all of that:)
The idea behind MIMO is that wireless signal are orthgonal in space. Think about it this way:
Lets say I have three transmit antennas, sending three signals (x,y, and z). I also have three (or more) receive antennas (1, 2, and 3).
The first thing I do is train my receivers by having the transmitters transmit a known signal one at a time, such that when transmitter x is transmitting, at receiver 1 I receive a signal (a1*x), at receiver 2 I get (a2*x), and at receiver 3 I get (a3*x). When y transmits, I get (b[1,2,3]*y) and similarly for z (c[1,2,3]*z) at my receivers. Since I know what x,y, and z are supposed to be at this point, I can calculate values for a[1,2,3], b[1,2,3], and c[1,2,3].
After the training sequence is complete, I begin transmitting on all three at once and assume a,b, and c haven't changed that much (a reasonable assumption as long you're not moving much) leaving me with:
Given that the only unknowns in this equation are x,y and z, its perfectly solvable.
Unfortunately, its not quite that simple. The signals will mutually interfere, so a long process begins where you estimate one signal and start subtracting it and then estimate and subtract the other signals based on your result and then keep repeating process until your estimation is satisfactory.
On the bright side, though, the rayleigh-fading nature of the channel (in a nutshell, the dynamics of reflected signals cause signal strength to vary wildy at the receiver depending on location) means that antennas one-half wavelength apart will maxmize the relative energy difference of the different signals at different antennas. Thus, you're hoping (usually correctly, unless the receiver is on a satellite or a large open area) that each signal will be most powerful at a different antenna making the signal-estimation phase much quicker (ie, if the received signal x is way more powerful than y or z on antenna 1, x is easy to estimate which in turn makes y and z easier to estimate).
Slashdot Mirror
You cite Rense.com?!? Are you fucking kidding me? How can you trust any website that prints trash like this?
Tim
Yeah, I mean god forbid people be allowed to watch what *they* want to watch instead of what elitist fucks like yourself want to force them to watch...
Tim
Smart man.
:) ) will want to show that same pride in his production code and work to make it great (not perfect, perfect is the enemy of good enough) and on schedule rather than just going through the motions.
I got into my current company after leaving college without a degree mostly by sending in a sapce-invaders game I wrote in 80x86 assembly language along with my resume. Of course, this was a device-driver and embedded systems job, so showing off some m@d 5ki11z (did I even do that right?:) ) with assembly language was particularly relevant and gave me a starting point in my interviews (and helped my resume stand out, too).
It never hurts to be a bit of a show-off during an interview. I do much different work now (DSP programming), so asking for a portfolio of actual code (as opposed to a thesis or something) would be rather difficult, but I still like it when a candidate likes to show off and demonstrates some pride in his accomplishments. I take it as a sign that he/she (who am I kidding, he
Tim
The problem with China is that the PRC doesn't allow Chinese banks (outside of Hong Kong) to hold foreign assets. Under this arrangement, improting foreign products can become painfully difficult for the population to buy imported products, but still export on the same terms.
Now, this arrangement may (and probably will, if they continue it indefinitely) lead to a consumer revolt, it doesn't effect their export prices or economy at all in the short-term (as for the long term, see Japan) as long as they keep the Yuan pegged to the US Dollar.
Tim
The problem with Indian and Chinese workers, as I see it (through my interactions with 30+ of each), is that their EE degree (in my specific case) is seen more as a ticket out than an undying passion.
.com days, but to a much lesser degree, because you could still earn a decent living being an accountant or finance guy or chemist or whatever if you had the capacity to do it. But in India and China, the pay-scales are so out of whack and have been for a while now, that there is a totally undermotivated, wrongly-educated workforce (though still certainly intelligent enough to handle the job) in place and its a crap-shoot when hiring.
This doesn't mean that there aren't very intelligent Indian or Chinese coders out there - In fact, I work with several of them. However, I also work with several that are reasonably intelligent people (they do have that MS or PhD after all) but are horrid developer because they lack the passion for for engineering.
Look at it this way - in the US or Western Europe (though to a degree less so), if you're a smart guy you can go into almost any field you want and bring home a decent salary. In India or China, your choices for bringing home mad bank are limited to getting a job in EE or CS. Being an accountant in China sucks ass, so even though running numbers may be your passion you're not going to do it if you feel you have some aptitude in programming or circuit design.
So what you end up with is not a less educated or intelligent workforce, but one that didn't go through a natural filtering system. We saw this here in the US during the
Tim
Nyquist is the sample-rate guy (the Nyquist theorem).
Shannon capacity is the theoretical bit-rate you can stuff through a channel of a given physical bandwidth with a given signal to noise ratio:
C = W * Log2(1+S/N)
where C is capacity (bits per second), W is bandwidth (Hz), S is signal power, and N is noise power.
As far as Turbo Codes go, without getting too technical, its an extension of the principle that you can increase the efficiency of a channel while transmitting at a given power by attaching some redundant bits (redundant signal dimensions is probably a better way to look at it, though) to the signal. I'm not too familiar with the particulars of Turbo Coding, but it is a lot like Viterbi coding where these redundant bits are dependent on the data bits and when you detect an error (the redundant bit doesn't match the proper sequence), you back-trace through your data and find the most likely non-errored sequence and adjust your data bits accordingly.
Tim
Thank you for saving me the time of pointing out in excruciating detail that most (all, maybe) commercial modems wouldn't work if we needed 10x the Nyquist to reconstruct the waveform.
Really folks - you only need a touch more than than 2x the Nyquist. If the highest frequency we're interested in is 2 Mhz, then all we need to reconstruct a sine wave at that frequency is a sample rate of 4MHz + some infinitely small amount.
If you disagree, please take up your argument with the ghost of Claude Shannon.
Tim
It depends which kind of status meeting. If its just a meeting to fill in the PM, then yeah, 25 minutes is good. But if its more engineering oriented, taking a while isn't necessarily a bad thing.
For a project I was recently working on, we did a combo meeting - the first 15-30 minutes was spent filling in the PM on our progress, and the next 30-45 minutes was spent helping out people who were stuck on their current task. The second half wasn't mandatory, you could leave if you wanted to, but I think it was pretty useful given the kind of work we do (DSP programming on a proprietary chip) can at times resemble voodoo - sometimes other people have run across a similar situation in the past and can sum up a solution in 5 minutes where you might spend 5 weeks to come to the same answer.
Tim
Going back to Uni is also not an option (at least in my state, north-rhine westfalia, other states already did that or announced to do it in the next years), because it costs EUR 650, in addition to the ~ EUR 150-200 that it cost before, for "long-time" students
:)
:) oh and fun and family and friends and education and personal satisfaction and all that stuff too ).
I'm sorry, this kills me - 850 Euro? $1100? And you're bitching and whining?
If you want to go back to university, take out a freaking loan and just do it! It isn't all that much - even if you went for four years, that's only 5K Euro in debt (after accumulating interest). You'll probably be paying around 120 Euro/month to pay that off in 5 years after finishing - I don't understand the big deal.
I took out about $60K in loans to go to school here in the US (and got about another $60K in gratis government assistance), and 4 years later they're about half paid off and I still net way more each month after making my $700 payments than I would have if I didn't take those loans. And I didn't even bother to graduate
So less whining, more action. If you want to do anything - make money, get more education, whatever - action always accomplishes more than whining. And don't worry about debt, there's more to life than having chickenshit money right now (like having a lot of money later
Remember, fortune favors the bold.
Tim
That's not really true - Windows, for instance, runs in protected mode, where the chip intercepts memory access and hands it off to the OS for translation which then hands it back to the chip for execution. This way, the OS (or the old DOS memory managers, if you remember them) translate your address into a flat memory model. Under this system, each application gets its own little memory sandbox - In theory, I shouldn't be able to corrupt the memory of another program unless I have both cunning and malicious intent.
Now, what this means is that if I write a Win32 application-layer program that attempts to directly modify, say, the interrupt vector table, I'm not writing to the real interrupt vector table that the hardware uses when it receives physical interrupt signals, I'm writing to a version of it in my slice of memory, and what happens next is anybody's guess.
In otherwords, access to memory-mapped system resources like screen-buffers and interrupt vectors is arbitrated by the OS memory manager. What I want to do may work, it may not. I imagine Linux is more friendly in this regard, but I really don't know. DOS, on the otherhand, was anarchy - there was no memory manager (unless you intentionally ran one with your application) so all requests from the application layer were carried out without any intervention.
Tim
Screw you. I've been at this since the late 60's. Coding in assembler is fine for tinie, winie stuff, not applications which accomplish most of the world's functionality
Oh really?
How about the computer that makes your car engine work efficiently? Assembly.
The physical layer of virtually every communications device on the planet? Assembly.
Avionics systems? Quite a bit of assembly.
Higher level languages have their place - I use Matlab and C extensively - but in any system that is competing on cost (like modems) or has size/weight/power constraints (avionics, engines) assembly is the language of choice for production because it uses less silicon to get the same result.
Oh, and the guys writing the compilers had better have a good grasp of assembly as well if they plan on making decent optimizations.
Tim
When I started my current job, the compiler for one of our processing engines was only kinda-sorta done. Common instructions were supported, but most mode settings and some other things weren't. (Un)Fortunately, the compiler did let us hack in the correct opcode, leading to code like this:
------------
...stuff I can read... //I'll put a useful comment here
...stuff I can read... //and here
...stuff I can read...
0x6782; //comments?
0x9567; //we don't need no stinking comments!
0x8102;
...stuff I can read...
0x7230; //It should be completely obvious what this does, no comment here
...stuff I can read...
------------
Even as somebody who *prefers* to program on the metal, I would never want to revisit that hell.
TimFirst off, while I am not a Linux expert, I imagine that Linux would allow you more freedom to actually do stuff like mess with interrupt vectors and system resources (aasuming you are running at the correct privelege layer) whereas from my minimal exposure to Win32 driver-layer assembly, it seemed you are rather restricted in some regards.
:)
IMHO, DOS is the perfect OS to learn assembly-language programming on - no restrictions at all (and fast reboots when you fuck up). But maybe I'm just biased because I program DSPs with no operating system whatsoever
Tim
Along those lines, I'm a former physics student (dropout) now doing DSP/digital-communications work at the physical layer.
I started college at an EE, switched to CS, switched to Physics, left, got a job doing testing, a couple days later (after a bit of showing off) changed into driver development, a month or two later turned into an embedded systems programmer, a year or so later moved into TC-layer development (layer 1/2 stuff), and after that sorta morphed into the DSP engineer I am now.
Life can take weird turns - I went full circle. So don't pigeonhole yourself.
Tim
I test drove an S-2000 last time I was looking for a car. Its pretty quick, its great with the top down, the speakers sucked (I hope they fixed that), but to me, the real problem was that it was insanely noisy.
The powerband is over 7000 RPM, so its got a very squeally nature to it. A high pitched whine. I couldn't stand it. Maybe if I bought one, I'd have built the internal filter for that frequency, but it could have driven me nuts, too.
Tim
Oh, and 1 more thing:
If you include a couple of papers that you co-authored, we will pull them out and the engineers you interview with will have read them. And, personally, I will ask you about it.
So if its something you got a co-authorship on when helping out a crazy-genius professor in college but don't fully understand yourself, cop to that immediately when asked about it (the honesty is appreciated, and its a fairly common situation - it won't count against you, it will just realign expectations with reality). If you don't, I (and many of my colleagues) will expect you to be all-knowing god on that subject, and in some cases you may be interviewing with someone who *is* an all-knowing on the subject - in fact, if you have a co-authorship for a paper on, say, Delta-Sigma DAC techniques, you will be scheduled to interview with our resident delta-sigma guru. You may be able to play the bullshit with some of us who have a more basic understanding of foo, but the expert will catch you.
Tim
Only 7 pages? :)
:)
A few months ago, we received a 27 (!) page resume (single-spaced) for an entry-level DSP engineering position.
Now, admittedly, he was obviously a smart guy - with 80+ IEEE papers (listed individually, making for the bulk of that 27 pages). But there was just no way we could possibly hire a guy like that for an entry level position - for a senior researcher, fine, but we have enough of those. Seemingly, he was interested in doing actual engineering work but there was nothing there to indicate that he spent time doing anything other than churn out papers.
We normally schedule about 6 hours of interviews - this guy lasted 15 minutes before him and my boss realized they were looking for very different things
And here's a hint - if you have voluminous output in published journals (or the patent office - we get similar resumes stuffed with patents, though not to such an absurd degree), cite a couple that are relevant - for instance, we're a DSL company - all your papers on cell networks are nice (and semi-relevant), but if you highlighted your 3 papers on wireline communications that you co-authored and included a line about having 77 more papers published you would save everyone involved a lot of time - if we're that interested (for verification purposes or morbid curiousity), we'll go to IEEE.org or uspto.gov and search for your name. Listing that much individually just makes you seem arrogant. (on the otherhand, if you've only published a couple of things, go ahead and list them individually regardless of subject - it'll at least give us a good starting point in an interview and a chance to show-off).
Tim
Yes you can. You can define a binary system with however many bits you like. It may not look pretty in a hexadecimal representation, but there's no reason you can't do it.
In fact, I work on a processor in which one engine has 20 bit words, another has 18 bit words, and another does multiplies out to 29 bits (15 bit x 14 bit).
Tim
Don't they use unemployment claims to measure that?
No, they don't. Not the OECD numbers, anyways (which is the 6%).
There are two methods used:
1) Calling people up and asking if they are employed or looking for work.
2) Polling a realtively large sample of employers about their payroll-size. On a quarterly basis, this one isn't very good because it misses new startups, but over the long term its probably the more accurate number. Lately, this one has been 6%, and the first one has been somewhat lower.
Tim
Oh BTW, neat idea, but it's really hard to make inductors out of silicon.
I'm not an LSI guy (though I do work closely with our LSI team), but IIRC from a college course a million (ok, 4:) ) years ago, there is a way to simulate an inductor using capacitors and op-amps. I'm pretty sure that analog-LSI people do this all the time when they make our uber-cool programmable A/D-D/A chips with tunable filters and adaptive hybrids.
But from my fuzzy recollection, it seemed like really convoluted deign - you'd probably waste more power trying to simulate the inductor than you would save by having your sim-inductor.
Tim
It may tick some people off, but it does work.
I haven't seen the August numbers yet, but in July YahooBB! (well, SoftbankBB now) absolutely creamed NTT in new subscribers. It wasn't even close - something like 2-1 over NTT East and West combined.
Acca and E-Access, on the other hand, are in for a world of hurt precisely because they are small, and the even smaller providers are going to get hit worse. If you're really interested in the details, I'll explain further (I really shouldn't on a public BB) - email me: thetimdog@hotmail.com.
Tim
While I agree that the stuff you did sounds like cheating, you could also define it as asymmetric advantage.
Or, to put it back in the warfare setting: what's the difference between having the rifle fir without pulling the trigger and having an attack helicopter provide support? What's the difference between obscuring the sensor and damn good body armor? What's the difference between a god-mode cheat and divine intervention (ok, I'm going way out on a limb here)?
Well, let's analyze:
1) Totally defeats the training for the shooter, but gives the defender something more to think about. This should enhance the effect of the shootee's training - total suprise is the best way to go - don't even get into the enemy's field of fire.
2) Is pretty realistic. Modern body armor is pretty damn good and the MILES system of one-shot-one-kill doesn't necessarily apply. Both sides should be prepared for this and prepare accordingly.
3) Is just plain cheating. WTF was your CO thinking?
Tim
Err, DMT ADSL (which is like 99.9% of all ADSL connections), as per the G.992 standard, is most certainly a baseband signal using an OFDM modualtion with bin 0 defined at DC (each bin has a width of 4.3125 KHz). And while no actual data is transmitted below bin 6 (6*4.3125 KHz), the signal is processed in a baseband fashion (though I suppose it would be possible to demodulate and then decode, you'd have to be a real masochist to do so).
Tim
You can check my earlier post on it. I'm not retyping all of that :)
TimThe idea behind MIMO is that wireless signal are orthgonal in space. Think about it this way:
Lets say I have three transmit antennas, sending three signals (x,y, and z). I also have three (or more) receive antennas (1, 2, and 3).
The first thing I do is train my receivers by having the transmitters transmit a known signal one at a time, such that when transmitter x is transmitting, at receiver 1 I receive a signal (a1*x), at receiver 2 I get (a2*x), and at receiver 3 I get (a3*x). When y transmits, I get (b[1,2,3]*y) and similarly for z (c[1,2,3]*z) at my receivers. Since I know what x,y, and z are supposed to be at this point, I can calculate values for a[1,2,3], b[1,2,3], and c[1,2,3].
After the training sequence is complete, I begin transmitting on all three at once and assume a,b, and c haven't changed that much (a reasonable assumption as long you're not moving much) leaving me with:
Rx1 = a1*x + b1*y + c1*z
Rx2 = a2*x + b2*y + c2*z
Rx3 = a3*x + b3*y + c3*z
Given that the only unknowns in this equation are x,y and z, its perfectly solvable.
Unfortunately, its not quite that simple. The signals will mutually interfere, so a long process begins where you estimate one signal and start subtracting it and then estimate and subtract the other signals based on your result and then keep repeating process until your estimation is satisfactory.
On the bright side, though, the rayleigh-fading nature of the channel (in a nutshell, the dynamics of reflected signals cause signal strength to vary wildy at the receiver depending on location) means that antennas one-half wavelength apart will maxmize the relative energy difference of the different signals at different antennas. Thus, you're hoping (usually correctly, unless the receiver is on a satellite or a large open area) that each signal will be most powerful at a different antenna making the signal-estimation phase much quicker (ie, if the received signal x is way more powerful than y or z on antenna 1, x is easy to estimate which in turn makes y and z easier to estimate).
Tim