They glow because they're doped--rubies were shown on a PBS show a few years back. Lab created gems are generally doped with elements to make them distinguishable somehow--don't think scientists and engineers don't get bribed to produce a perfect gem.
Yes, single crystal 'gems' created in the lab are for all intents and purposes, perfect, they have to be to be used in any experiment concerning the creation of semiconducting devices.
It probably would NOT change the gemstone market due to cost of growing diamonds. IIRC, CVD (Chemical Vapor Deposition) is currently the only way to produce diamonds for manufacturing. This is in no way as cheap or easy to do is pulling a 'perfect' silicon ingot out of a molten bath.
Re:What exactly counts as 'legacy'?
on
Legacy-Free PCs
·
· Score: 1
They'd probably end up with something similar (though lower power) to DEC's Alpha. The ultimate irony though, Alpha (the ISA) didn't even survive halfway through it's lifetime of 25 years while x86 is still freakin' alive!!!
Re:Let's hear it for legacy free!
on
Legacy-Free PCs
·
· Score: 1
You're not the only one. USB's limited to 100 reports per second, PS/2'll do 200. Frankly, until they get that up, or they strap a mouse to firewire and give it a better response rate, I'm sticking to PS/2 and that's that.
Sometimes elections don't take the form just going to the ballot and dropping a slip of paper in, instead it takes the form of bloody revolution.
Instead of a simple change of faces in government, you get to recreate your government. Look at the Chinese. Thousands of years of revolution--tearing down one dynasty to push up the next. Even when civil servents tested for their place in government to work for the emperor, there were revolutions.
In the case of a brutal dictatorship, it becomes a case of how often and how hard you can beat the people before they rise up against you. Even a dog would realize it's going to die someday, should we expect less from people?
Moreover, what happens if the people DO want some sort of monarchy or dictatorship (keep in mind, the form of government doesn't necessarily equate to good or bad, it's a question of HOW the government governs)? Do they vote for it? How would you vote for that? Do you think the existing government would stand there and take it? What happens when you loose confidence in someone? Can you suddenly vote no confidence one day and have them thrown out and replaced? The ability to vote is nice, but only if that ability grants true power--if it's nothing more then a delegation of rights and responsibilities, it's just a carrot dangled before the ignorant to keep them plodding in the direction you want them to walk.
Actually 1600x1200 is the resolution I try to stick to on my Hitachi 751. With a better video card, it would be my ONLY resolution--'cept for bootup of course.
At better then 21", I'd want 2048x1536 (Eizo/Nanao F980 preferably).
In terms of native resolution, a 17" LCD IS equivalent to a 17" CRT. If that 17" LCD had a native resolution of 1600x1200 and priced the same as my favorite Hitachi and Mitsubishi CRT's I might actually think of buying one, but until that time, I'll spend my money elsewhere.
I think the point of Rambus was equal latancy, much higher bandwidth. The problem was that Rambus' first word latancy sucked the life out of them for business apps, but it's great if you're streaming a whole array--donno how it performs traversing a binary tree or linked list though.
I think he's suggesting that companies should fight it out and make a name for their product. Not unreasonable considering pharm. companies spend more on advertising then on R&D, and the amount of R&D that goes on at universities and colleges all over the world.
Chances are good they'd support it through a an I/O chip (which would be pretty cool to tie PCI Express to their Hypertransport bus--1 chip switching data for say 6 PCI Express devices tied directly to the CPU Ram Controller which uses another HyperT pipe to manage the AGP slot). It's what I'd expect anyway.
You do realize the mex 'compiler' uses whatever compiler it's got to compile C right? I've used it before for particularly time intensive loops (there was a better then 3x speed up for that program) by writing the loop in C and then using mex-gcc to compile the code. I know it works with borland also, as I've ported that program to a Win2k box using borland's free commandline compiler. Use the mex -setup (I think). More details are on mathwork's site.
So, what's performance like for an Itanium2 in a 128 CPU machine used for financial computing? Will it match a comparable Sun box?;-) Remember, Sun's strength isn't what they can do with 1 CPU, it's what they can do with hundreds and thousands of them.
The motherboard manufacturers themselves have been saying they're waiting for AMD to release Athlon 64 to the market, and with Opteron due on April 22nd, I expect AMD to support it themselves (they've done it before with the Athlon and Athlon MP, they'll do it again for Opteron). Which means there will be very little lag if any at all.
Even though Apple has control over systems, they'll have to get chips from IBM first to design chipsets around--or get contractors to do that. So unless Apple's already decided, I doubt they're going to get chips until IBM starts sampling to anybody who wants some to test for a platform, and who knows exactly when that will be, and it will take even more time to design the rest of the platform if Apple wants their own platform solution rather then adapt some of their interface to the reference solution.
Well, you can do final test in BATCHES (statisically this works out fine, especially if you strive for the 6 sigma quality control), or you can perform final test on EACH individual unit. Testing individual units can have more wear and tear on your test equipment (boards, connectors, probes, etc), as well as possibly requiring more man power, while also lowering your shipping volume in say, a month.
Contrast this with batch testing, where you can test the same number of units, but statistically, it's representing a batch many times the size, thus allowing your shipping volume to bring your test cost per unit down.
R&D&T, costs only so much, as volume climbs towards infinity, the effect that R&D&T has on each unit drops to zero.
It's due to the fact that this SATA drive is 10k RPM and the design requires a smaller platter, physically, (for now) then what you'd see in a 7200 RPM HD--thus resulting in smaller platter sizes (in GB or MB).
Well, when you pay someone to build something for you, isn't that contracting? Doesn't that mean whatever got built is YOURS? If the Gov't paid for the infrastructure to be built, it doesn't matter who did it, it only means that it belongs to the Gov't, doesn't it? In which case, it would be perfectly fair for them to say, "Alright, everybody can rent them for $10 a pop, and we'll be contracting with other guys to maintain those lines." Wouldn't it?
Well, if it was a mission critical server, I'd agree.
However, a 4-way machine stopped being a mission critical server to me a long time ago, now it's just an expensive workstation--and on an expensive workstation, if something I do takes 10 minutes vs 30 minutes, I'll take the faster platform and just store my important data on the server.
Most likely this is a combination of SSE/SSE2 and the rate at which the P4's capable of accessing RAM. After all, unless spec's using lots and lots of SSE2, I'd attribute the P4's SpecFP scores mostly to it's RAM access capabilities.
Remember, with FP, it's generally not like you're working with JUST 1 number at a time and then jumping all over the RAM for the next, more likely, you're accessing a huge array of numbers where you can simply drag down data until you reach your end point (which explains why P4's tied to RDRAM still perform best in SpecFP--they avoid the critical first word).
If everyone banded together and bought a majority share of the stocks and held it as a co-op, the employees would effectively run the company as they'd have a larger voice on the board, wouldn't they? Doesn't sound so crazy to me.
Slashdot Mirror
They glow because they're doped--rubies were shown on a PBS show a few years back. Lab created gems are generally doped with elements to make them distinguishable somehow--don't think scientists and engineers don't get bribed to produce a perfect gem.
Yes, single crystal 'gems' created in the lab are for all intents and purposes, perfect, they have to be to be used in any experiment concerning the creation of semiconducting devices.
It probably would NOT change the gemstone market due to cost of growing diamonds. IIRC, CVD (Chemical Vapor Deposition) is currently the only way to produce diamonds for manufacturing. This is in no way as cheap or easy to do is pulling a 'perfect' silicon ingot out of a molten bath.
They'd probably end up with something similar (though lower power) to DEC's Alpha. The ultimate irony though, Alpha (the ISA) didn't even survive halfway through it's lifetime of 25 years while x86 is still freakin' alive!!!
You're not the only one. USB's limited to 100 reports per second, PS/2'll do 200. Frankly, until they get that up, or they strap a mouse to firewire and give it a better response rate, I'm sticking to PS/2 and that's that.
Sometimes elections don't take the form just going to the ballot and dropping a slip of paper in, instead it takes the form of bloody revolution.
Instead of a simple change of faces in government, you get to recreate your government. Look at the Chinese. Thousands of years of revolution--tearing down one dynasty to push up the next. Even when civil servents tested for their place in government to work for the emperor, there were revolutions.
In the case of a brutal dictatorship, it becomes a case of how often and how hard you can beat the people before they rise up against you. Even a dog would realize it's going to die someday, should we expect less from people?
Moreover, what happens if the people DO want some sort of monarchy or dictatorship (keep in mind, the form of government doesn't necessarily equate to good or bad, it's a question of HOW the government governs)? Do they vote for it? How would you vote for that? Do you think the existing government would stand there and take it? What happens when you loose confidence in someone? Can you suddenly vote no confidence one day and have them thrown out and replaced? The ability to vote is nice, but only if that ability grants true power--if it's nothing more then a delegation of rights and responsibilities, it's just a carrot dangled before the ignorant to keep them plodding in the direction you want them to walk.
I'd add The Razor's Edge to the list of Bill Murray movies with interesting messages.
Actually 1600x1200 is the resolution I try to stick to on my Hitachi 751. With a better video card, it would be my ONLY resolution--'cept for bootup of course.
At better then 21", I'd want 2048x1536 (Eizo/Nanao F980 preferably).
In terms of native resolution, a 17" LCD IS equivalent to a 17" CRT. If that 17" LCD had a native resolution of 1600x1200 and priced the same as my favorite Hitachi and Mitsubishi CRT's I might actually think of buying one, but until that time, I'll spend my money elsewhere.
That's a Win9X trait.
I think the point of Rambus was equal latancy, much higher bandwidth. The problem was that Rambus' first word latancy sucked the life out of them for business apps, but it's great if you're streaming a whole array--donno how it performs traversing a binary tree or linked list though.
How much did you pay for the powder over the years though? Nickels and dimes add up.
I think he's suggesting that companies should fight it out and make a name for their product. Not unreasonable considering pharm. companies spend more on advertising then on R&D, and the amount of R&D that goes on at universities and colleges all over the world.
Just my interpretation, I could be wrong though.
Chances are good they'd support it through a an I/O chip (which would be pretty cool to tie PCI Express to their Hypertransport bus--1 chip switching data for say 6 PCI Express devices tied directly to the CPU Ram Controller which uses another HyperT pipe to manage the AGP slot). It's what I'd expect anyway.
They probably underestimated the design in the first place.
In theory, practice must follow theory, in practice, things aren't so simple.
I thought it was a gov't resolution to eventually provide high speed connections to every household in Canada, and the gov't watches over CA*Net3?
You do realize the mex 'compiler' uses whatever compiler it's got to compile C right? I've used it before for particularly time intensive loops (there was a better then 3x speed up for that program) by writing the loop in C and then using mex-gcc to compile the code. I know it works with borland also, as I've ported that program to a Win2k box using borland's free commandline compiler. Use the mex -setup (I think). More details are on mathwork's site.
So, what's performance like for an Itanium2 in a 128 CPU machine used for financial computing? Will it match a comparable Sun box? ;-) Remember, Sun's strength isn't what they can do with 1 CPU, it's what they can do with hundreds and thousands of them.
Just a couple of nitpicks.
The motherboard manufacturers themselves have been saying they're waiting for AMD to release Athlon 64 to the market, and with Opteron due on April 22nd, I expect AMD to support it themselves (they've done it before with the Athlon and Athlon MP, they'll do it again for Opteron). Which means there will be very little lag if any at all.
Even though Apple has control over systems, they'll have to get chips from IBM first to design chipsets around--or get contractors to do that. So unless Apple's already decided, I doubt they're going to get chips until IBM starts sampling to anybody who wants some to test for a platform, and who knows exactly when that will be, and it will take even more time to design the rest of the platform if Apple wants their own platform solution rather then adapt some of their interface to the reference solution.
Well, you can do final test in BATCHES (statisically this works out fine, especially if you strive for the 6 sigma quality control), or you can perform final test on EACH individual unit. Testing individual units can have more wear and tear on your test equipment (boards, connectors, probes, etc), as well as possibly requiring more man power, while also lowering your shipping volume in say, a month.
Contrast this with batch testing, where you can test the same number of units, but statistically, it's representing a batch many times the size, thus allowing your shipping volume to bring your test cost per unit down.
R&D&T, costs only so much, as volume climbs towards infinity, the effect that R&D&T has on each unit drops to zero.
It's due to the fact that this SATA drive is 10k RPM and the design requires a smaller platter, physically, (for now) then what you'd see in a 7200 RPM HD--thus resulting in smaller platter sizes (in GB or MB).
Well, when you pay someone to build something for you, isn't that contracting? Doesn't that mean whatever got built is YOURS? If the Gov't paid for the infrastructure to be built, it doesn't matter who did it, it only means that it belongs to the Gov't, doesn't it? In which case, it would be perfectly fair for them to say, "Alright, everybody can rent them for $10 a pop, and we'll be contracting with other guys to maintain those lines." Wouldn't it?
^_^ Maybe I'm just massively over simplifying.
Well, if it was a mission critical server, I'd agree.
However, a 4-way machine stopped being a mission critical server to me a long time ago, now it's just an expensive workstation--and on an expensive workstation, if something I do takes 10 minutes vs 30 minutes, I'll take the faster platform and just store my important data on the server.
I wonder if something like a Matrox RT2500 would help you any.
Most likely this is a combination of SSE/SSE2 and the rate at which the P4's capable of accessing RAM. After all, unless spec's using lots and lots of SSE2, I'd attribute the P4's SpecFP scores mostly to it's RAM access capabilities.
Remember, with FP, it's generally not like you're working with JUST 1 number at a time and then jumping all over the RAM for the next, more likely, you're accessing a huge array of numbers where you can simply drag down data until you reach your end point (which explains why P4's tied to RDRAM still perform best in SpecFP--they avoid the critical first word).
BWAHAHAHA
For that price, it better come with a hooker!
If everyone banded together and bought a majority share of the stocks and held it as a co-op, the employees would effectively run the company as they'd have a larger voice on the board, wouldn't they? Doesn't sound so crazy to me.