I personally believe that you are wrong.
One person can certainly learn enough technical
knowledge to invent things. In fact, I've been
structuring my life around learning enough to
do the things that are "too hard for one person".
At the moment (and I'm only 20), I write compilers
and OS-level code for fun, I design and build
innumerable electronic devices, I design
chemical processes (primarily for PC-boards currently),
I do theoretical math and physics, I'm learning
machining and woodworking, and I read medical texts on
the side. AND I've almost figured
out a homegrown process for SOI IC fabrication. *WHEW*
I'm a firm believer that merely dropping the attitude
"I can't learn it because I'm too [stupid/specialized/etc]"
is all that's really required. Human capability is
limitless if you push hard enough.
Check out FORTH. Now there's a clean language.
Unfortunately, it's as hard as programming backwards
in assembly. Fortunately, it's extremely fast
and dynamically compiled. Dynamic compilation is a tremendous
advantage, one that Adobe knows all about (comparing
PostScript and FORTH is almost funny, they're so
similar in many ways).
Then again, maybe I'm
biased, seeing as how I've written a FORTH code
generator, several FORTH translators and designed
a CPU to natively run FORTH instructions (at the
gate level, sooner or later I may build one in TTL
for fun).
The software configuration for that would be almost
trivial (except for routing issues). The real
problem lies in backbones. 802.11 works fine for
connecting groups (in cities, lets say). To do
more would require a more complex setup.
Personally, I'd think that something like 30+Ghz
line-of-signt microwave would work for connecting
cities, but I doubt there would be enough people
with the required knowledge to set that up in
every city. That, and although the actual transciever
equipment might not cost that much ($400 or so, homebuilt)
the comm towers to put it on would be a lot more difficult/expensive.
Then again, maybe somebody will take a page from Dave Gingery's books
and build the towers from pop-cans;)
Personally, I'd love to be involved in such a project,
but I'm so far out in the middle of nowhere that I'm
not sure of the use. Although I do have a printed circuit
board lab sitting around here, and it would be fun.
Freedom of information doesn't mean information is free. Just 'cause you can legally read the book doesn't mean you don't have to buy the book.
Funny, I check books out from my library all the time.
Sometimes I even go to my local Chapters store
and sit for hours and read books (they actually have
a policy that they are part library and not just
a bookstore, it's how they stay open on Sundays when
other bookstores must remain closed).
Mind you, I also find that I buy a LOT more books
from Chapters than any other bookstore... They
give me free access to read them at their store,
and my memory is very nearly photographic, and I
still buy books. Hrm...
I'm going to stick my neck out here and ask what
the hell is up with this moderation. This comment
is certainly inflamatory, but mostly to government
officials and such. Personally I think it should
be modded up as insightful.
Now just watch, somebody is going to mod me down
just for my trouble.
I suspect that the transmitting/receiving equipment could be built quite easily by somebody with EE training, HAM radio equipment, and a used satellite-TV dish (slightly modified).
Personally, I've wanted to hack the downstream on satellites for a while (passive scanning of satellite communication). It would be fun, free TV, and Hubble pictures right away!
Now, actually taking control of one, that would be too illegal for me.
My shot at this is now up on my homepage, I even dug up my old sources (it's a hack onto aavga). Somebody can probably get it working if they try hard enough.
At the time, I didn't think this worthy of posting to slashdot. Hehehe.
I beat them to it.
on
Textmode Quake 2
·
· Score: 5, Interesting
I had Quake2 in textmode long before this. I hacked the aaquake svgalib-emulator module so that it supported multiple video pages.
I'm updating my homepage right now with some screenshots, see it at my homepage.
First, let me say that my first suspicion in this case would be your multimeter. I've measured an aweful lot of NiCD AA's in my time, and anything well beyond 1.2-1.3V is unusual, to say the least.
Now, the lead-acid battery is a collection of cells, so there's no telling what the regular "rated" voltage is without knowing the number of cells involved, and the sepecific chemistry (some of the new gel-batteries are slightly different). It would seem that it was measured under certain nominal load conditions. This is often true for large batteries that must supply current under heavy load (car batteries come to mind). In this case, they may stack more cells than would seem appropriate.
The simplest way to explain this would be to connect a certain resistor (the load resistor) across the battery and measure the voltage across the resistor. It will be quite a bit lower than the maximum (unloaded) voltage.
To deliver power to a heavy load at a certain voltage, the internal resistance must be taken into consideration (it acts as a voltage divider, further elaboration can be found in any basic electronics text). So to achieve a certain voltage under load, it must actually deliver a higher potential unloaded.
An interesting one is that, as a cell/battery is discharged, the unloaded voltage stays roughly the same, while the loaded voltage drops. (Which is where the so-called "internal resistance" of the battery comes from. The voltage essentially stays the same, but the ability to supply current drops, hence a higher "internal resistance").
In any situation where the voltage really matters, a solid-state regulator should be used, since the actual voltage from most power sources is uncertain.
Basic electrochemistry. They aren't really batteries, they're "cells", meaning they output whatever voltage your choice of electrodes combines to (redox reactions). Lithiums do 3.0V if I remember correctly. Ni-Cd and Ni-MH do 1.2 and 1.25 respectively, iirc.
The Outer Limits did not one, but TWO episodes about this very thing (sort-of). It was a rare mutagenic condition caused by "designer" children, which caused them to become these hideous monster things.
Gataga looked at similar problems, although more about designer children than clones.
Oh, by the way, 50% failure is average. In-Vitro usually requires 2-4 embryos implanted at once. Often, neither will catch (usually at least one will).
Yes. My SO had radioactive iodine (I131) treatement for thyroid cancer. There were big warnings about how it could cause cancer (especially lymphoma and leukemia). On the other hand, it is extremely effective at preventing recurrence, so....
Actually, you often get what are called (I think, I forget if this is the correct name) bioactive tumours, which still perform one or two of their original functions.
So you'll get somebody with WAY to much adrenal function, or some pituitary hormone going berzerk. Most cancers don't work this way, but specific types do.
Actually, you're dead wrong. Nerve cells do regrow (although spinal cords tend to need a LOT of help, they're only learning how to do that one).
There was a recent discoveery made that brain cells actually do continue to divide and populate the brain (they tested this in monkeys by injecting them with dye, so that the dye would only show up in new tissues. Scanning the brain, they found large deposits of this dye). Check the archives, I'm sure it was on slashdot.
Not to mention that most people who've had nerve damage after surgery will tell you about the "electric shocks" they feel 6-8 months afterward (the nerves regrowing and reconnecting, the area starts becoming sensitive again).
Verilog (and VHDL, and several others) are indeed hardware description languages. What this student has done is to write new Verilog to implement the same interface (instruction set, etc) as the ARM, while using none of ARM's hardware language.
It's like Abiword reading/writing.doc files. The application is all original code, but it's using an interface from somebody else. If they can really bully him around, I'm very scared to be an EE.
Actually, SRAM (static RAM, like the CMOS) is much faster, that's why it's used for on-die caches. It's also many times as big, so we get the nasty compromises of DRAM (dyanmic RAM, it loses it's data every ~60msec, so you constantly need to scan through and refresh it).
Unfortunately, DRAM would be a really serious issue, since even a small (100K) chip can draw a good solid amp or two during a write or refresh operation (their power usage is very "spikey", meaning most memory chips have lots of big capacitors around them to handle it).
Funny, the unity-gain frequency of a MOSFET is given by Ft=(gm/(2*pi*(1/2*W*L*Cox)). So scaling down W[idth] or L[ength] down by a certain factor would raise the maximum operational speed of a MOSFET by the same amount. Some parts, such as dynamic RAM, don't really scale this way, but static RAM (such as used on on-die caches) is composed purely of MOSFETS, not the capacitors in DRAM's.
The main reason for the rearchitecting when the process is shrinked is generally the larger available number of transistors. A design involving 45 million electromechanical relays could take advantage of the shorter per-clock paths as easily as modern CMOS, but such a thing would be enormous. Chip yields tend to be based on wafer size moreso than complexity, so the smaller processes yield many extra available transistors.
A truly "atomic" operation is one that involves only two-level-logic (minterms or maxterms). Any design can be simplified to this, although a floating-point multiplier would be an amazingly large array of gates. Any operation that takes more than two propogation delays can be subdivided, although you will observe diminishing returns. There is no such thing as "infinitely fast", a clock cycle will always require 2*(propogation delay) plus a little for bookkeeping. So-called "Dynamic" logic does things in exactly this way.
Now, I/O to the actual die becomes the major bottleneck, since parasitic reactances on the pins start becoming a serious issue. So the actual caching strategies do need to be improved to keep the processor busy (rather than waiting on system RAM). I will note that several "alternative" solutions have come as a result of this, usually involving interspersing small CPU's with static RAM in a big die, or using FPGA's for computation (which is my personal favorite). Alternatively, explicit caching is difficult to program, but performs very well.
Actually, the real issue is the length (in terms of propogation delays) of the instruction pipeline. The P4 breaks up instructions into dozens of little steps, and does them all at once. The point is that these little steps each take less time to complete, so with the same speed transistors the clockspeed can be increased.
The problem with this idea is that the MHz goes up a lot, but the actual CPU speed may not (for instance, the P4 can do up to (roughly) 40 things at once - but if it fails branch-prediction, it can spend almost 40 clock cycles just regenerating the instruction path. Shorter pipelines are faster during heavily branched code, longer pipelines generally improve integer-math-heavy code.
I personally think that the parent post, as inflamatory as it is, should be modded up.
I'm personally somewhat younger, and my IQ, as if it matters anything, is somewhere in that range. And I agree wholeheartedly.
I grew up in such a way that everything looked like a building set to me. My favorite toys age 11-14 were solderless breadboard and 7400-series TTL logic. I've had virtually all the major building sets, Construx, Meccano, Robotix, Capsela, Lasy (that's a little obscure), used Lego considerably (my brother had some). I had a half-dozen Radio Shack lab kits. I've programmed since I was five.
Any parent thinking I'm a special case should reevaluate how they perceive their children. Intelligence is learned, during (generally) the first 6 years of life.
I don't know, I connected my C128 to my construx motor. I found that construx will mount breadboards fairly easily, so I designed a quick-and-dirty relay driver board and plugged it into the joystick port. It could only move back and forth but it was programmable.
Of course, this was like 9 years ago. God I feel old...
Of course, bipolar is HUGE and draws a lot of power. Makes for a real pain trying to use it on a handheld and learning that 10 7404's can drain 4 AA's in under an hour. And that's just the interface circuitry.
I still think it has it's place on a system running off mains though.
Hrm, Symbol technologies produces a handheld which is rated at a 12 foot drop to cement before it will break. They also made one with a half-VGA touchscreen rated at a 6 foot drop.
I've played with these devices. X86 compatible even (and MUCH more expensive than the average Palm, they ran about $5000 for the 8088 model). The heavy plastic and rubberized corners were truly a sight to behold.
Personally, I'd want my handheld to be able to handle a drop of at least 4 feet to cement (since that's the normal operating height) without breaking, since handhelds DO get dropped ocassionally (my current handheld is a TI-89 *wink*, which is actually pretty rugged).
Charge is in Coulombs, Capacity (oops, dated term, it's "Capacitance" now, reading too many 1940's-era electronics books;) is in Farads.
Furthermore, The 1970's were almost entirely dominated by TTL logic, which is almost immune to static (you can still kill it, but you need to be TRYING). In the case of TTL, current is what kills it.
Thing is, CMOS doesn't work like that. A MOS transistor has a thin layer of metal oxide which acts as a capacitor. When the voltage across it exceeds the dielectric ability to withstand voltage (I forget the term....), then it pops, and, just like that, it's ruined. Voltage is the killer for CMOS, which is why ESD has become an issue.
Slashdot Mirror
At the moment (and I'm only 20), I write compilers and OS-level code for fun, I design and build innumerable electronic devices, I design chemical processes (primarily for PC-boards currently), I do theoretical math and physics, I'm learning machining and woodworking, and I read medical texts on the side. AND I've almost figured out a homegrown process for SOI IC fabrication. *WHEW*
I'm a firm believer that merely dropping the attitude "I can't learn it because I'm too [stupid/specialized/etc]" is all that's really required. Human capability is limitless if you push hard enough.
Then again, maybe I'm biased, seeing as how I've written a FORTH code generator, several FORTH translators and designed a CPU to natively run FORTH instructions (at the gate level, sooner or later I may build one in TTL for fun).
Personally, I'd think that something like 30+Ghz line-of-signt microwave would work for connecting cities, but I doubt there would be enough people with the required knowledge to set that up in every city. That, and although the actual transciever equipment might not cost that much ($400 or so, homebuilt) the comm towers to put it on would be a lot more difficult/expensive.
Then again, maybe somebody will take a page from Dave Gingery's books and build the towers from pop-cans ;)
Personally, I'd love to be involved in such a project, but I'm so far out in the middle of nowhere that I'm not sure of the use. Although I do have a printed circuit board lab sitting around here, and it would be fun.
Funny, I check books out from my library all the time. Sometimes I even go to my local Chapters store and sit for hours and read books (they actually have a policy that they are part library and not just a bookstore, it's how they stay open on Sundays when other bookstores must remain closed).
Mind you, I also find that I buy a LOT more books from Chapters than any other bookstore... They give me free access to read them at their store, and my memory is very nearly photographic, and I still buy books. Hrm...
Now just watch, somebody is going to mod me down just for my trouble.
If you've wired logic by hand with 7400-series TTL (like I have), HDL's come quite naturally. They're NOTHING like a computer programming language.
Grab a copy of klogic or something similar and play with digital design before you learn an HDL. You'll do much better design as a result.
Personally, I've wanted to hack the downstream on satellites for a while (passive scanning of satellite communication). It would be fun, free TV, and Hubble pictures right away!
Now, actually taking control of one, that would be too illegal for me.
At the time, I didn't think this worthy of posting to slashdot. Hehehe.
I'm updating my homepage right now with some screenshots, see it at my homepage.
Now, the lead-acid battery is a collection of cells, so there's no telling what the regular "rated" voltage is without knowing the number of cells involved, and the sepecific chemistry (some of the new gel-batteries are slightly different). It would seem that it was measured under certain nominal load conditions. This is often true for large batteries that must supply current under heavy load (car batteries come to mind). In this case, they may stack more cells than would seem appropriate.
The simplest way to explain this would be to connect a certain resistor (the load resistor) across the battery and measure the voltage across the resistor. It will be quite a bit lower than the maximum (unloaded) voltage.
To deliver power to a heavy load at a certain voltage, the internal resistance must be taken into consideration (it acts as a voltage divider, further elaboration can be found in any basic electronics text). So to achieve a certain voltage under load, it must actually deliver a higher potential unloaded.
An interesting one is that, as a cell/battery is discharged, the unloaded voltage stays roughly the same, while the loaded voltage drops. (Which is where the so-called "internal resistance" of the battery comes from. The voltage essentially stays the same, but the ability to supply current drops, hence a higher "internal resistance").
In any situation where the voltage really matters, a solid-state regulator should be used, since the actual voltage from most power sources is uncertain.
Basic electrochemistry. They aren't really batteries, they're "cells", meaning they output whatever voltage your choice of electrodes combines to (redox reactions). Lithiums do 3.0V if I remember correctly. Ni-Cd and Ni-MH do 1.2 and 1.25 respectively, iirc.
There are more of us out there than people think.
Gataga looked at similar problems, although more about designer children than clones.
Oh, by the way, 50% failure is average. In-Vitro usually requires 2-4 embryos implanted at once. Often, neither will catch (usually at least one will).
Yes. My SO had radioactive iodine (I131) treatement for thyroid cancer. There were big warnings about how it could cause cancer (especially lymphoma and leukemia). On the other hand, it is extremely effective at preventing recurrence, so....
So you'll get somebody with WAY to much adrenal function, or some pituitary hormone going berzerk. Most cancers don't work this way, but specific types do.
There was a recent discoveery made that brain cells actually do continue to divide and populate the brain (they tested this in monkeys by injecting them with dye, so that the dye would only show up in new tissues. Scanning the brain, they found large deposits of this dye). Check the archives, I'm sure it was on slashdot.
Not to mention that most people who've had nerve damage after surgery will tell you about the "electric shocks" they feel 6-8 months afterward (the nerves regrowing and reconnecting, the area starts becoming sensitive again).
It's like Abiword reading/writing .doc files. The application is all original code, but it's using an interface from somebody else. If they can really bully him around, I'm very scared to be an EE.
Unfortunately, DRAM would be a really serious issue, since even a small (100K) chip can draw a good solid amp or two during a write or refresh operation (their power usage is very "spikey", meaning most memory chips have lots of big capacitors around them to handle it).
The main reason for the rearchitecting when the process is shrinked is generally the larger available number of transistors. A design involving 45 million electromechanical relays could take advantage of the shorter per-clock paths as easily as modern CMOS, but such a thing would be enormous. Chip yields tend to be based on wafer size moreso than complexity, so the smaller processes yield many extra available transistors.
A truly "atomic" operation is one that involves only two-level-logic (minterms or maxterms). Any design can be simplified to this, although a floating-point multiplier would be an amazingly large array of gates. Any operation that takes more than two propogation delays can be subdivided, although you will observe diminishing returns. There is no such thing as "infinitely fast", a clock cycle will always require 2*(propogation delay) plus a little for bookkeeping. So-called "Dynamic" logic does things in exactly this way.
Now, I/O to the actual die becomes the major bottleneck, since parasitic reactances on the pins start becoming a serious issue. So the actual caching strategies do need to be improved to keep the processor busy (rather than waiting on system RAM). I will note that several "alternative" solutions have come as a result of this, usually involving interspersing small CPU's with static RAM in a big die, or using FPGA's for computation (which is my personal favorite). Alternatively, explicit caching is difficult to program, but performs very well.
Sorry to flame.
The problem with this idea is that the MHz goes up a lot, but the actual CPU speed may not (for instance, the P4 can do up to (roughly) 40 things at once - but if it fails branch-prediction, it can spend almost 40 clock cycles just regenerating the instruction path. Shorter pipelines are faster during heavily branched code, longer pipelines generally improve integer-math-heavy code.
I'm personally somewhat younger, and my IQ, as if it matters anything, is somewhere in that range. And I agree wholeheartedly.
I grew up in such a way that everything looked like a building set to me. My favorite toys age 11-14 were solderless breadboard and 7400-series TTL logic. I've had virtually all the major building sets, Construx, Meccano, Robotix, Capsela, Lasy (that's a little obscure), used Lego considerably (my brother had some). I had a half-dozen Radio Shack lab kits. I've programmed since I was five.
Any parent thinking I'm a special case should reevaluate how they perceive their children. Intelligence is learned, during (generally) the first 6 years of life.
And is it any wonder I ended up in Engineering?
Of course, this was like 9 years ago. God I feel old...
I still think it has it's place on a system running off mains though.
I've played with these devices. X86 compatible even (and MUCH more expensive than the average Palm, they ran about $5000 for the 8088 model). The heavy plastic and rubberized corners were truly a sight to behold.
Personally, I'd want my handheld to be able to handle a drop of at least 4 feet to cement (since that's the normal operating height) without breaking, since handhelds DO get dropped ocassionally (my current handheld is a TI-89 *wink*, which is actually pretty rugged).
Furthermore, The 1970's were almost entirely dominated by TTL logic, which is almost immune to static (you can still kill it, but you need to be TRYING). In the case of TTL, current is what kills it.
Thing is, CMOS doesn't work like that. A MOS transistor has a thin layer of metal oxide which acts as a capacitor. When the voltage across it exceeds the dielectric ability to withstand voltage (I forget the term....), then it pops, and, just like that, it's ruined. Voltage is the killer for CMOS, which is why ESD has become an issue.