Actually all the registers were in fact, thirty-two bit, which is what I was referring to. Both address and register. Well, okay, some of the oddball things like the status/ccr register might have been 16-bit, but that was just a bunch of bitflags anyways..
The busses were another story, and there were indeed 8, 16, and 32bit variants for data, and 24 and 32bit variants for address busses for the 68K line (680{0,1,2,3,4,6}0).
Damn I'd still love to have a 68060-50 based A4000T though, but that probably still costs more than a high end PC..arg
"I would love to see some applications written to take advantage of those extra registers! (Linux apps aside.)"
<Abe Simpson> When I was your age, sonny, we used Motorola 68K CPUs which had sixteen registers (address and data) instead of eight like in those x86 units. And they were all 32-bit in all models of 68K, I tell you! And never you mind that linear 32-bit addressing... Why, back then you could go to a movie, eat at a fancy restaurant, and buy a gallon of gas all for a nickel... Zzzzzz..</Abe Simpson>
Actually I just use my "OCT" library (eight 16-bit "words", combined into a single variable, 128-bits long) to handle situations like that.
3.4028 x 10^38 ought to be enough seconds for anybody (10,790,283,070,806,014,188,970,529,154,990 non-leap-years)
Let's see that's um ten thousand billion billion.. uh billion? non-leap-years. (365 day periods, each day being 86400 seconds)
For shorter periods, I use just unsigned 32-bit based off of a 1-Jan-2000 00:00:00 epoch, with the zero value being reserved for an error. (This updates from my previous library which used 1990 as an epoch). That should take me to something like 2136, and gives me the handy fact that my time-equivalent function returns TRUE if it worked, and FALSE otherwise. (if(time(&Now))... )
I have a linksys unit myself, (BEFSR11, it was cheaper to get that and a 5-port switch than the SR41 by itself at the time) and it used to have difficulties requiring a reset when it's firmware was yonger, but the latest firmware (1.45.7 for me, I should check for more updates) seems to be completely stable. Your unit will probably lose those glitches as the firmware matures and updates become available, also.
The unit itself only consumes 0.7A at 7.5VDC - five watts, and is fanless and smaller than an 8-port switch. It's also very simple, which suggests to me that it's a lot less likely to have some sort of exploit that can root it. If an exploit was discovered, it would take an embedded devices expert to actually make use of it, and even if THAT happened, the router contains next to nothing in the way of sensitive data, and further attacks would be necessary to cause any serious damage to my datafiles. Also, it's capabilities to launch other attacks would be insignifigant, especially compared to a P2+ Linux or BSD box.
The linksys box also handles PPPoE, which is necessary with some high speed ISPs up here, which frees me from the burden of having to compile PPPoE support into a kernel, and/or installing that broken Access Manager onto a Windows box.
One last thing, it boots almost instantly, and is ready right away. Most of my PC boxes take at least 30 seconds to boot, Linux AND Windows (2k) both.
"Look at it this way: would you rather have the wristwatch that is hand crafted to perfection, works better, and will last forever, or would you rather buy the watch that came off of the assembly line, always loses time, and will break on you in a year or two?"
The watch from the assembly line is a tenth of the price. Many people can't look beyond the "tenth of the price" issue, and you can't always blame them. Quality isn't always obvious, up front. A shame really.. many good products die because their quality goes unappreciated under a hefty price tag.
Personally though, I'd rather buy that handcrafted watch that lasts forever. Better to buy one two hundred dollar thing, that it will last forever, than fifty ten-dollar items, that only last six months..
Don't forget that the chips were better too; the Agnus could handle 2 megs of chip and the Denise could handle a wider variety of modes (same bandwidth as before though; lost color information in the higher rezes), built-in hardware flickerfixer, etc.
The 3000 was awesome. I bought one a couple of years after the 4000 came out, used, just because it was such a groundbreaker (and much cheaper than the 4K). I still have it, and it still works, although I managed to break the ALT key off of the keyboard during a move...arg
"I think that concern has been answered by the nForce series of MB chipsets. I've built several nForce2 based systems, and they are rock solid."
This is good to hear, as I've been considering an nForce2 based AMD upgrade for my aging P3-1000mhz system. While I applaud the performance of a P4-3.2ghz/875P system, I'd really rather not have to have a mortgage on my computer~
After tax, in CDN$, a Barton-core Athlon XP 2500, with 512 megs of DDR, and nForce2 mobo, is about $450-470. Compared to a 3.2ghz P4 CPU all by itself, $550 before tax, well... I'm not feeling price/performance happy on the Intel side.
Normally I'd be pro-Intel, for compatibility reasons, but the cost difference cannot be ignored here. A stable, well performing system for half the cost is very persuasive.
(Be it so noted that Intel has always done badly with their low-cost CPUs, 8088, 386sx, 486sx, pre-300A Pentium celerons, etc, anyways.)
I believe the Motorola 68000 CPU came out in like 1979, and it featured 24 address lines and 32-bit address registers, so was thusly ready for 16MB of RAM out of the box, and was transparently extendable to 4GB, with the addition of 8 more address lines on the 68020.
The only downside that was, at least in the Amiga community, that some programmers who fancied themselves clever, used the upper, unused, 8 bits of the address registers for flags, and thus their programs died horribly on 68020s, which could actually physically connect to the full 32-bit address range.
The 68K was a fine chip, with linear address space, and 8 general purpose data registers, and 7 general purpose address registers (plus one special purpose address register). It's such a shame we ended up with that kludgy intel beast. Sort of funny to watch a P4 or Athlon XP chip run MSDOS 5.0 with no emulation, though.;p
"If they do, the file formats for PDA's et al move away from MS's (FAT) standard - something that mas long-term repercussions for MS"
Actually, PalmOS units use their own filesystem, at least in their core memory. It's more like a memory allocation system too, than an actual formal filesystem. The flash/SD card things might have VFAT on those, but you don't _need_ to use those anyways..
"To promote the progress of science and useful arts"
Isn't this promoting progress? Wouldn't FAT dying a horrible death be a good thing? Surely we could come up with something better for interoperability... Heck, even a filesystem that works on the same cluster-allocation system, but simply has more space for filenames, instead of the long/short filename kludge (which is the core of the patent issue) which is hardly an optimal solution to begin with!
7.1) Bigger is better. Why use a color-indexed png or gif for that 7-color, badly drawn MS paint illustration, when a 24-bit bmp or jpg at 100% is so much bigger! This also applies to using reducing width= and height= values to make thumbnails in galleries, instead of quality and size-reduced images that are actually smaller and faster to load...
Slashdot Mirror
Actually all the registers were in fact, thirty-two bit, which is what I was referring to. Both address and register. Well, okay, some of the oddball things like the status/ccr register might have been 16-bit, but that was just a bunch of bitflags anyways..
..arg
The busses were another story, and there were indeed 8, 16, and 32bit variants for data, and 24 and 32bit variants for address busses for the 68K line (680{0,1,2,3,4,6}0).
Damn I'd still love to have a 68060-50 based A4000T though, but that probably still costs more than a high end PC
"I would love to see some applications written to take advantage of those extra registers! (Linux apps aside.)"
<Abe Simpson> When I was your age, sonny, we used Motorola 68K CPUs which had sixteen registers (address and data) instead of eight like in those x86 units. And they were all 32-bit in all models of 68K, I tell you! And never you mind that linear 32-bit addressing... Why, back then you could go to a movie, eat at a fancy restaurant, and buy a gallon of gas all for a nickel... Zzzzzz..</Abe Simpson>
Actually I just use my "OCT" library (eight 16-bit "words", combined into a single variable, 128-bits long) to handle situations like that.
... )
3.4028 x 10^38 ought to be enough seconds for anybody (10,790,283,070,806,014,188,970,529,154,990 non-leap-years)
Let's see that's um ten thousand billion billion.. uh billion? non-leap-years. (365 day periods, each day being 86400 seconds)
For shorter periods, I use just unsigned 32-bit based off of a 1-Jan-2000 00:00:00 epoch, with the zero value being reserved for an error. (This updates from my previous library which used 1990 as an epoch). That should take me to something like 2136, and gives me the handy fact that my time-equivalent function returns TRUE if it worked, and FALSE otherwise. (if(time(&Now))
I have a linksys unit myself, (BEFSR11, it was cheaper to get that and a 5-port switch than the SR41 by itself at the time) and it used to have difficulties requiring a reset when it's firmware was yonger, but the latest firmware (1.45.7 for me, I should check for more updates) seems to be completely stable. Your unit will probably lose those glitches as the firmware matures and updates become available, also.
The unit itself only consumes 0.7A at 7.5VDC - five watts, and is fanless and smaller than an 8-port switch. It's also very simple, which suggests to me that it's a lot less likely to have some sort of exploit that can root it. If an exploit was discovered, it would take an embedded devices expert to actually make use of it, and even if THAT happened, the router contains next to nothing in the way of sensitive data, and further attacks would be necessary to cause any serious damage to my datafiles. Also, it's capabilities to launch other attacks would be insignifigant, especially compared to a P2+ Linux or BSD box.
The linksys box also handles PPPoE, which is necessary with some high speed ISPs up here, which frees me from the burden of having to compile PPPoE support into a kernel, and/or installing that broken Access Manager onto a Windows box.
One last thing, it boots almost instantly, and is ready right away. Most of my PC boxes take at least 30 seconds to boot, Linux AND Windows (2k) both.
"Look at it this way: would you rather have the wristwatch that is hand crafted to perfection, works better, and will last forever, or would you rather buy the watch that came off of the assembly line, always loses time, and will break on you in a year or two?"
The watch from the assembly line is a tenth of the price. Many people can't look beyond the "tenth of the price" issue, and you can't always blame them. Quality isn't always obvious, up front. A shame really.. many good products die because their quality goes unappreciated under a hefty price tag.
Personally though, I'd rather buy that handcrafted watch that lasts forever. Better to buy one two hundred dollar thing, that it will last forever, than fifty ten-dollar items, that only last six months..
Don't forget that the chips were better too; the Agnus could handle 2 megs of chip and the Denise could handle a wider variety of modes (same bandwidth as before though; lost color information in the higher rezes), built-in hardware flickerfixer, etc.
The 3000 was awesome. I bought one a couple of years after the 4000 came out, used, just because it was such a groundbreaker (and much cheaper than the 4K). I still have it, and it still works, although I managed to break the ALT key off of the keyboard during a move...arg
"I think that concern has been answered by the nForce series of MB chipsets. I've built several nForce2 based systems, and they are rock solid."
This is good to hear, as I've been considering an nForce2 based AMD upgrade for my aging P3-1000mhz system. While I applaud the performance of a P4-3.2ghz/875P system, I'd really rather not have to have a mortgage on my computer~
After tax, in CDN$, a Barton-core Athlon XP 2500, with 512 megs of DDR, and nForce2 mobo, is about $450-470. Compared to a 3.2ghz P4 CPU all by itself, $550 before tax, well... I'm not feeling price/performance happy on the Intel side.
Normally I'd be pro-Intel, for compatibility reasons, but the cost difference cannot be ignored here. A stable, well performing system for half the cost is very persuasive.
(Be it so noted that Intel has always done badly with their low-cost CPUs, 8088, 386sx, 486sx, pre-300A Pentium celerons, etc, anyways.)
I believe the Motorola 68000 CPU came out in like 1979, and it featured 24 address lines and 32-bit address registers, so was thusly ready for 16MB of RAM out of the box, and was transparently extendable to 4GB, with the addition of 8 more address lines on the 68020.
The only downside that was, at least in the Amiga community, that some programmers who fancied themselves clever, used the upper, unused, 8 bits of the address registers for flags, and thus their programs died horribly on 68020s, which could actually physically connect to the full 32-bit address range.
The 68K was a fine chip, with linear address space, and 8 general purpose data registers, and 7 general purpose address registers (plus one special purpose address register). It's such a shame we ended up with that kludgy intel beast. Sort of funny to watch a P4 or Athlon XP chip run MSDOS 5.0 with no emulation, though. ;p
"If they do, the file formats for PDA's et al move away from MS's (FAT) standard - something that mas long-term repercussions for MS"
Actually, PalmOS units use their own filesystem, at least in their core memory. It's more like a memory allocation system too, than an actual formal filesystem. The flash/SD card things might have VFAT on those, but you don't _need_ to use those anyways..
"To promote the progress of science and useful arts"
Isn't this promoting progress? Wouldn't FAT dying a horrible death be a good thing? Surely we could come up with something better for interoperability... Heck, even a filesystem that works on the same cluster-allocation system, but simply has more space for filenames, instead of the long/short filename kludge (which is the core of the patent issue) which is hardly an optimal solution to begin with!
Long ago? 1800-1900s? Hey, my mom knows shorthand, dammit.. (old style shorthand; not this silly 'intarweb!!111' crap)
You forgot:
7.1) Bigger is better. Why use a color-indexed png or gif for that 7-color, badly drawn MS paint illustration, when a 24-bit bmp or jpg at 100% is so much bigger! This also applies to using reducing width= and height= values to make thumbnails in galleries, instead of quality and size-reduced images that are actually smaller and faster to load...
Forgot that the damned less-than sign is an HTML special character. D'oh!
Actually, on my old debian-potato system that I haven't bothered upgrading, sleep doesn't take floating point arguments.
However, on my debian-woody system, with a newer version of sleep, it does.
..plus it features authors! lol
# kitt.sh - change the '1' to desired delay
:; do for led in num caps scroll caps; do setleds -L +$led
#(requires GNU shell utils 2.0.11~)
#(based on code presented in this thread)
while