You can do similar studies on a fairly large number of animal subjects (e.g. stick a Nokia webphone in a rat cages -- there's where corporate funding is useful).
Simultaneously, you could measure how long it takes the rats to figure out ebay and see if there is an increased amount of cancer cells in the brains of the rats when they keel over.
More seriously, you could attempt to model the RF output of the phones on a human's head to that of a rat and see useful data in a little less time.
I went to a slightly cheaper route, which has some disadvantages but works fine for most things. I bought two Wavelan PCMCIA cards, one for my laptop and one for my desktop PC. I bought a $50 ISA PCMCIA card reader (A PC-700 from Buy.com if I remember correctly). Configuration on both machines was a snap and network performance is great. This doesn't work as an access point, I use Ad-hoc networking mode for both cards. The Wavelan interface runs over IPMasq.
If you want to save a few bucks and you really don't need an access point (or you want to use an existing IP Masq box) this will work fine for you.
The real missing link is here. If nothing else, this/. article gave me a few minutes amusement looking for the "missing link". Now where are those damn dirty apes?
At the beginning of this semester I bought a pair of the Sony MDR-NC20 headphones (for around $130 + shipping). The sound out of them is pretty decent and they are extremely comfortable (they cover your entire ear, so it looks like you have compound bug eyes for ears).
The drawbacks: they only cancel up to around 1500Hz, though it does an excellent job against computer fans and hard drives in my room and in lab. It definitely gets rid of the annoying lower-pitched whines. These particular headphones leave a small static-ish hiss in place of the noise. It's not annoying when you're actually playing music (I don't hear it if the music is loud enough to be recognizable), but it's there.
I haven't used them at work yet, where the big machine room is (on the order of 100 mainframes and thousands of hard disks), but I'll post when I've tried them there.
Of course these won't help you much if you want to talk on the phone at the same time. It simply isn't possible.:-)
Re:OK -- So when's /. going to HTTPS ???
on
The Encryption Wars
·
· Score: 1
It is, however, extremely CPU-intensive (yes, you'll notice the speed difference on pentiums). Now consider serving a few hundred connections with SSL on top of the normal load and you'll realize why e-commerce websites simply cannot encrypt everything. One trick is to keep the images unencrypted, but you can only buy yourself so much, particularly when I find myself downloading 200k - 500k web pages from Slashdot.
What you say about most browsers having SSL support now is correct. Even lynx has patches for SSL.
I'm upset to hear that people still do things like that to hardware.
Any clue as to when they'll be implementing the local digital version of Halloween? Just think, perfect reproductions of your hardware-bashing bananza!
I'd also fear that the auto industry is back at the same point! The reason is that many luxury automobiles have or will have 100 - 150 microcontrollers inside them, with networks ranging from a few kilobits (who pops those power locks?) to tens of megabits (multimedia systems, those LCD panels and everything else that keeps kids quiet. Yes, people buy these things.) What comes with lots of microcontrollers? Lots of software.
The car companies are under more pressure than software companies at this time. Their products can kill a person if they fail. Some cars are going to have steer and brake-by-wire (with no mechanical backup, as you see in many of today's higher-end cars) in a few years. If these systems fail, you're putting the driver, passengers and other people in danger. As a company, you put yourself in a position for many big lawsuits if you get the system (hardware, software, network) wrong. You had better bet that they are going to use a good process to make these things right, because finding out in the field (like you might with unnamed software companys' mass beta-level software distributions).
What scares me, however, is the interconnection of the different networks. Yes, there is a firewall between my engine's fuel injector, the brake system's network, the multimedia/communications network (*with* integrated cell phones -- cars that phone home when the airbags deploy are already out there). These systems have to be designed to be robust if something breaks, but they may also have to worry about attacks from outside. Good software design methodologies, like the upper levels of the CMM model described (but not named?) in the article, can help you get there. It's not a guarantee, but it definitely makes me more likely to trust that the cars on the road will operate correctly (to some definition of correctly).
Yes, there is still nmake in there. Maybe you didn't install all of the command line utilities?
Personally, I use GNU make from the cygwin package to fire off an nmake session. It's a little hokey, but if you can send the build profile off to nmake, it works well. By itself, nmake is very painful to use.
There is a < TABLE > tag in HTML (since version 2 or so), look at how the default slashdot web page is rendered. Tables? You bet!
Granted, you don't get fine control over the separator lines between cells, but the idea is there.
In addition, there is a math markup language extension for HTML (Mozilla supports this, the exact name escapes me right now). I haven't tried it, I don't believe Netscape 4.x supports it, but there are plugins. So, if you want those features from HTML, they're there.
I do agree, however, that there is no one-to-one coorespondence between tags or even functionality between the two systems, and the only conversion programs that can realistically be written are going to miss a lot and will produce documents that need serious help.:)
As far as an FFT engine goes, I've found that the FFTW library (it's GNU!) is very easy to use and not particularly processor intensive. There's no reason to rewrite code when people have made the "Fastest Fourier Transform in the West!":)
I used it on a real-time guitar sound effects processing program (never released) that I wrote a year ago and never had any problems using it.
If you haven't heard enough of the voting machines, there is an interesting lecture from a CMU faculty member who has worked for 20 years certifying voting machines (electrical, mechanical, paper, punchcard, optical scanning, etc...).
This lecture describes not only how you can cheat these systems, but how and why it's so hard to get an accurate count (not to mention why it's nearly impossible to count the same number of votes twice , mechanically OR manually).
Yes, some of the predictions that he made are somewhat dated (as the lecture happened on the 15th), but it is a very interesting overview of what can and *does* happen to your votes that you may not have thought about. It's a little long, but worth the time!
Frequently when NT goes "out to lunch" for a couple of seconds or so, it's actually trying to go out on the network and find *something*. When I have shortcuts to network shares on my desktop, the desktop takes significantly longer to load (although the CPU load is near zero during that time waiting on the net).
You might want to look for some obscure mapped drive or lingering network connection (NT likes to remember old ones, for some reason). Having a fast machine only makes those idle waits more apparent, because the "real" work gets done faster.
I have also seen this as an employee. My current manager wanted to hire a particular Ph.D. He could not get approval from the higher-ups for several months, because they needed justification that a Ph.D would really help the group.
By the time they finally cleared the approval, my manager learned that the potential employee had taken a job at another company. Even if a manager wants to hire, the additional salary and company politics may make it difficult on everyone.
On the 2nd to last page, it notes some numbers about non-resident and women recipients of PhDs.
48% and 43% of CS PhD's went to non-residents and only 12% went to women.
This doesn't surprise me after looking at the graduate student pictures in the engineering builing hallway. This particular school (CMU) has seen a marked increase (from 10% to nearly 40%) women in the freshman CS class in the past three years. Maybe these statistics could be just as fluid in a few years?
For older (read: non-pipelined) machines, you might use the inverse (clocks per instruction or CPI, a 68HC11 might take 3 clock cycles to execute an ADD instruction). It's generally an average over a plausible mix of instructions that would be a significant part of a normal software program.
Well, how do you make CPI into IPC? First, you pipeline the processor. Theoretically, you could get your CPI or IPC to near 1.0. However, certain things such as loads from memory might require you to wait a clock or two, pausing the entire pipeline. That'll keep your IPC from being big (big generally = better for IPC).
Okay, great. You did your best trying to get it near 1.0 and you couldn't quite do it. How about starting multiple instructions in each clock? Of course there are complications, you can't do an ADD that requires an answer before the multiply that supplies that answer has completed. So there are limitations. But for many instructions which are not inter-dependent, you can issue them all at the same time. Surprise! Now you can get an IPC > 1.0.
Yes, this is a grossly simplified answer computer architecture answer condensed in a can, but maybe it clears things up a little.
In the early 1970's, Burroughs Corporation (one of the consituents of today's Unisys Corporation) had a machine called the B1700. It was intended to be a "universal host" machine. Basically, before it ran a particular job (or between time slices of running jobs, if you so desired), the machine would load a set of microcode for the type of machine that it was supposed to be operating as.
I know that people ran it in PDP11 mode, as well as an IBM 360 mode (this was reportedly faster than a native 360 according to some sources).
The machine could essentially change its apparent instruction set on the bit level (up to 24-bit wide instructions). Over the summer, I started reading a book on the architecture (which was orange and I can't remember the name, but I think it's out of print). The details were rather confusing to me, but the high-level idea of the machine is very cool.
So, what you're asking about could be done in hardware, as well as software. Search on google for more information with "burroughs B1700".
Code reuse and plagerism are two different terms, which should not be confused.
The difference is that you're not actually copying someone's code when you use a library or a tool. For instance, you don't (or shouldn't) claim that you wrote the standard ANSI C library that is linked with your software. That is code reuse and as long as the license for the library permits that nobody will bother you.
However, if you decide to take the contents of the ANSI C library, insert them in your code and claim that they're your own (with or without credit), that's plagerism. It can be legal if your license permits that and there are no other restrictions. Similarly, if you integrate someone else's tool into your work, you had better make sure the licenses are okay and that you have given credit where credit is due.
We try to keep one registered republican in the house just so we can get their mailings. Yes, they're mis-guided, but if you read them just right, they can be hysterical.
On the other hand, if you happen to give $25 to the DNC, you're also on their mailing list for life. That means christmas cards from Clinton/Gore and other goodies that are neat to have hanging around. Generally we don't get vile spam from the dems, though.
Not only that, but you get a parking spot right by the front door. And, if you're sharing lab equipment with other people, they're not awake to take the machines at that time!
Flex-time truly is a wonderful thing. A number of other coworkers like to work a bit later. It's generally not a problem, because there are enough hours in the day to meet with them, as needed. I don't mind staying later if I'm in an interesting conversation.:-)
Therefore, by your logic, you have been dead for 30 years. I assure you that the people who walked on the moon before were alive. No sense in NASA sending four corpses to bounce on the moon!
It's interesting to note that they didn't say which JVM they were using. The IBM JVM and Sun's 1.3 JVM are supposed to be worlds better than the Blackdown 1.2 JVM on Linux. Not to mention, there is quite a bit of tweaking that can be done with just JVM settings (e.g. set Min/Max heap equal to keep heap reallocations down).
Also, I just have to note that nobody mentioned the testing methodology. Were they running the JVM for a long time? Or did they just fire it up for a few seconds? In the long case, the JVM will get into a steady state. In the initial case, most servers will be creating new objects and threads until they have adjusted to the load. A good rampup time is around 20 minutes.
Him: Your transceiver looks great!
Her: Thanks! Is that a Spot on your shirt?
Him: Why yes it is! It has 256MB of RAM!
Her:
if ( date == borg )
Oh!
else
[sighs]
Slashdot Mirror
Simultaneously, you could measure how long it takes the rats to figure out ebay and see if there is an increased amount of cancer cells in the brains of the rats when they keel over.
More seriously, you could attempt to model the RF output of the phones on a human's head to that of a rat and see useful data in a little less time.
If you want to save a few bucks and you really don't need an access point (or you want to use an existing IP Masq box) this will work fine for you.
The real missing link is here. If nothing else, this /. article gave me a few minutes amusement looking for the "missing link". Now where are those damn dirty apes?
The drawbacks: they only cancel up to around 1500Hz, though it does an excellent job against computer fans and hard drives in my room and in lab. It definitely gets rid of the annoying lower-pitched whines. These particular headphones leave a small static-ish hiss in place of the noise. It's not annoying when you're actually playing music (I don't hear it if the music is loud enough to be recognizable), but it's there.
I haven't used them at work yet, where the big machine room is (on the order of 100 mainframes and thousands of hard disks), but I'll post when I've tried them there.
Of course these won't help you much if you want to talk on the phone at the same time. It simply isn't possible. :-)
What you say about most browsers having SSL support now is correct. Even lynx has patches for SSL.
Any clue as to when they'll be implementing the local digital version of Halloween? Just think, perfect reproductions of your hardware-bashing bananza!
The car companies are under more pressure than software companies at this time. Their products can kill a person if they fail. Some cars are going to have steer and brake-by-wire (with no mechanical backup, as you see in many of today's higher-end cars) in a few years. If these systems fail, you're putting the driver, passengers and other people in danger. As a company, you put yourself in a position for many big lawsuits if you get the system (hardware, software, network) wrong. You had better bet that they are going to use a good process to make these things right, because finding out in the field (like you might with unnamed software companys' mass beta-level software distributions).
What scares me, however, is the interconnection of the different networks. Yes, there is a firewall between my engine's fuel injector, the brake system's network, the multimedia/communications network (*with* integrated cell phones -- cars that phone home when the airbags deploy are already out there). These systems have to be designed to be robust if something breaks, but they may also have to worry about attacks from outside. Good software design methodologies, like the upper levels of the CMM model described (but not named?) in the article, can help you get there. It's not a guarantee, but it definitely makes me more likely to trust that the cars on the road will operate correctly (to some definition of correctly).
Personally, I use GNU make from the cygwin package to fire off an nmake session. It's a little hokey, but if you can send the build profile off to nmake, it works well. By itself, nmake is very painful to use.
Granted, you don't get fine control over the separator lines between cells, but the idea is there.
In addition, there is a math markup language extension for HTML (Mozilla supports this, the exact name escapes me right now). I haven't tried it, I don't believe Netscape 4.x supports it, but there are plugins. So, if you want those features from HTML, they're there.
I do agree, however, that there is no one-to-one coorespondence between tags or even functionality between the two systems, and the only conversion programs that can realistically be written are going to miss a lot and will produce documents that need serious help. :)
I used it on a real-time guitar sound effects processing program (never released) that I wrote a year ago and never had any problems using it.
This lecture describes not only how you can cheat these systems, but how and why it's so hard to get an accurate count (not to mention why it's nearly impossible to count the same number of votes twice , mechanically OR manually).
The lecture is in .ASF and powerpoint format... sorry. Click http://wean1.ulib.org/cgi-bin/metawin-lectures.pl? target=Lectures/Distinguished%20Lectures /2000 and select the bottom row (Michael Shamos's "What's happening in Florida")! For some reason, slashdot is mangling the URL, so there are some spaces on it when you click there... just remove them in your netscape window.
Yes, some of the predictions that he made are somewhat dated (as the lecture happened on the 15th), but it is a very interesting overview of what can and *does* happen to your votes that you may not have thought about. It's a little long, but worth the time!
You might want to look for some obscure mapped drive or lingering network connection (NT likes to remember old ones, for some reason). Having a fast machine only makes those idle waits more apparent, because the "real" work gets done faster.
By the time they finally cleared the approval, my manager learned that the potential employee had taken a job at another company. Even if a manager wants to hire, the additional salary and company politics may make it difficult on everyone.
The same queen who heartlessly wrings the necks of pheasants, no less! http://mo re. abcnews.go.com/sections/world/dailynews/britain001 120.html.
48% and 43% of CS PhD's went to non-residents and only 12% went to women.
This doesn't surprise me after looking at the graduate student pictures in the engineering builing hallway. This particular school (CMU) has seen a marked increase (from 10% to nearly 40%) women in the freshman CS class in the past three years. Maybe these statistics could be just as fluid in a few years?
body transplant: the last ditch effort after yer plastic surgeon finally fails you
For older (read: non-pipelined) machines, you might use the inverse (clocks per instruction or CPI, a 68HC11 might take 3 clock cycles to execute an ADD instruction). It's generally an average over a plausible mix of instructions that would be a significant part of a normal software program.
Well, how do you make CPI into IPC? First, you pipeline the processor. Theoretically, you could get your CPI or IPC to near 1.0. However, certain things such as loads from memory might require you to wait a clock or two, pausing the entire pipeline. That'll keep your IPC from being big (big generally = better for IPC).
Okay, great. You did your best trying to get it near 1.0 and you couldn't quite do it. How about starting multiple instructions in each clock? Of course there are complications, you can't do an ADD that requires an answer before the multiply that supplies that answer has completed. So there are limitations. But for many instructions which are not inter-dependent, you can issue them all at the same time. Surprise! Now you can get an IPC > 1.0.
Yes, this is a grossly simplified answer computer architecture answer condensed in a can, but maybe it clears things up a little.
I know that people ran it in PDP11 mode, as well as an IBM 360 mode (this was reportedly faster than a native 360 according to some sources).
The machine could essentially change its apparent instruction set on the bit level (up to 24-bit wide instructions). Over the summer, I started reading a book on the architecture (which was orange and I can't remember the name, but I think it's out of print). The details were rather confusing to me, but the high-level idea of the machine is very cool.
So, what you're asking about could be done in hardware, as well as software. Search on google for more information with "burroughs B1700".
The difference is that you're not actually copying someone's code when you use a library or a tool. For instance, you don't (or shouldn't) claim that you wrote the standard ANSI C library that is linked with your software. That is code reuse and as long as the license for the library permits that nobody will bother you.
However, if you decide to take the contents of the ANSI C library, insert them in your code and claim that they're your own (with or without credit), that's plagerism. It can be legal if your license permits that and there are no other restrictions. Similarly, if you integrate someone else's tool into your work, you had better make sure the licenses are okay and that you have given credit where credit is due.
On the other hand, if you happen to give $25 to the DNC, you're also on their mailing list for life. That means christmas cards from Clinton/Gore and other goodies that are neat to have hanging around. Generally we don't get vile spam from the dems, though.
Flex-time truly is a wonderful thing. A number of other coworkers like to work a bit later. It's generally not a problem, because there are enough hours in the day to meet with them, as needed. I don't mind staying later if I'm in an interesting conversation. :-)
Therefore, by your logic, you have been dead for 30 years. I assure you that the people who walked on the moon before were alive. No sense in NASA sending four corpses to bounce on the moon!
Also, I just have to note that nobody mentioned the testing methodology. Were they running the JVM for a long time? Or did they just fire it up for a few seconds? In the long case, the JVM will get into a steady state. In the initial case, most servers will be creating new objects and threads until they have adjusted to the load. A good rampup time is around 20 minutes.
A sqwwaaaaakin' good time! irc.rabidpenguin.org
Him: Your transceiver looks great!
Her: Thanks! Is that a Spot on your shirt?
Him: Why yes it is! It has 256MB of RAM!
Her:
if ( date == borg )
Oh!
else
[sighs]
. . . and the rest is up to your imagination
A sqwwaaaaakin' good time! irc.rabidpenguin.org