Pentium Pro was good (at least as long as you stayed away from 16 bit code), but the Athlon showed that you can do stuff that only PA-RISC and Alpha had before with x86 ( in particular, superscalar FPU with parallel mul/add/memory op).
At that point, it was clear that you can shove EVERYTHING into x86. And who cares about 8 (now 16) registers, if you never touch them anyway and there are 200+ rename register under the hood?
It is the first time a x86 CPU has EVER been supplied with that much bandwith.
Yes, it is shared memory, but with low GPU load, the x86 CPU has >160GB/s bandwith available. Thats enough to provide a load and story for each core and clock cycle.
Tracking rate. At 1000 degree/s tracking rate if you want to keep the target stable in focus at closest appoach. I doubt that thing has gyros to give it that much rotational momentum...
2- No, its NOT the same as CNG. Because the energy density of adiabatic expansion is a LOT less than the one of natural gas combustion. So a much higher loss fraction during the compression cycle AND combustion waste heat cannot be used to pre-heat the compressed gas to counter valve freezing and whatever.
But with stars being point sources, with a telescope you can determine shifts of the center of brightness between stars very very accurately, especially with a sat telescope outside the atmosphere.
There ARE, of course limits, and the method becomes useless for stars many 1000 ly away (in the sense that the indirect methods, despite all their error possibilities, become more accurate), but 400ly is still doable.
Reading up, it turnes out the whole argumentation is exactly the other way round than I would hav eexpected.
You can meassure the distance of stars in multiple ways, most depend on assumptions that can be pretty hard to get right. There is ONE way, though, to accurately determine the distance:
By simple geometry. If you observe the star 6 months apart, you get a trianble with a base of 2AU, which is enough accurately triangulate the distance of stars up to a some 100 ly away.
This was exactly the method used 2 decade ago.
Not this new guy used a very indirect way (measuring the brightness we see the star, guessing its real brightness by looking at spectra and then deciding how far away it must have been), gets a 30% different number and claims his, indirect and error-laden, way is yielding the more correct of the results.
If you have been working for a decade with the same code, a certain blindness to issues will be inevitable. (In german, there is a word for it: Betriebsblindheit)
He might be right simply because of the outside perspective.
There is a reason they usually power down in the winter, when the french nuclear power plants have higher load from homes (Over here, there is no AC peak int he summer).
IIRC, LHC uses something between 250 and 350 MW power to run.
Drop those full size USB ports, and add a (micro) SD card slot.
It is totally ridiculous that all NEXUS devices are missing that one, even the new Nexus 10.I want to watch movies in a plane, or review my pictures away from my PC (where a 2560x1600 screen really would help). So fuck the cloud and fuck the tiered pricing system that askes for $100 more for adding $20 worth of flash - while STILL limiting the total capacity to amounts that are ridiculously low for a device of that cost.
Full sized USB I can understand for missing : Those plugs are huge. They would literally be the thickest thing in the tablet.
If you like your fenix, you would be blown away by something like a Quark Mini CR2, or a Jetbeam RRT-01 with 18350 cells (0.001 lumen to 800 lumen with seamless control via a dimmer ring).
Also, yes, those LEDs are getting cheaper, quickly. 5 Years ago you had to pay $15-20 for a led that put out 150 lumen at 50 lumen/W. Now you can get a led that puts out >1000 lumen at >100 lumen/W for $6.
Sorry, but that is NOT an "maximum for the existing technology". The value you give is like saying "lightspeed is the maximum speed for a Space Shuttle".
Intel HD3000 has less than 4%! Its just that that ATI and Nvidea are distributed over scores of indidual card types that each have a couple % that this ends up on top.
Slashdot Mirror
Don't forget the AMD Athlon.
Pentium Pro was good (at least as long as you stayed away from 16 bit code), but the Athlon showed that you can do stuff that only PA-RISC and Alpha had before with x86 ( in particular, superscalar FPU with parallel mul/add/memory op).
At that point, it was clear that you can shove EVERYTHING into x86. And who cares about 8 (now 16) registers, if you never touch them anyway and there are 200+ rename register under the hood?
You seem to be delusional.
Hurray to Google for re-inventing ActiveX. May they have just as much success as Microsoft with it.
Those class definitions refer to write speed. Even class 4 cards are typcially >>10MB/s when reading.
Do not ignore one thing of the PS4:
It is the first time a x86 CPU has EVER been supplied with that much bandwith.
Yes, it is shared memory, but with low GPU load, the x86 CPU has >160GB/s bandwith available. Thats enough to provide a load and story for each core and clock cycle.
Tracking rate. At 1000 degree/s tracking rate if you want to keep the target stable in focus at closest appoach. I doubt that thing has gyros to give it that much rotational momentum...
I wonder if they were actually able to take pictures or something. I mean, there is no way that the thing was ever designed for such tracking rates...
2- No, its NOT the same as CNG. Because the energy density of adiabatic expansion is a LOT less than the one of natural gas combustion. So a much higher loss fraction during the compression cycle AND combustion waste heat cannot be used to pre-heat the compressed gas to counter valve freezing and whatever.
Sure you arent mixing up Seven Samurai and The Hidden Fortress?
A couple millon.
But with stars being point sources, with a telescope you can determine shifts of the center of brightness between stars very very accurately, especially with a sat telescope outside the atmosphere.
There ARE, of course limits, and the method becomes useless for stars many 1000 ly away (in the sense that the indirect methods, despite all their error possibilities, become more accurate), but 400ly is still doable.
Reading up, it turnes out the whole argumentation is exactly the other way round than I would hav eexpected.
You can meassure the distance of stars in multiple ways, most depend on assumptions that can be pretty hard to get right. There is ONE way, though, to accurately determine the distance:
By simple geometry. If you observe the star 6 months apart, you get a trianble with a base of 2AU, which is enough accurately triangulate the distance of stars up to a some 100 ly away.
This was exactly the method used 2 decade ago.
Not this new guy used a very indirect way (measuring the brightness we see the star, guessing its real brightness by looking at spectra and then deciding how far away it must have been), gets a 30% different number and claims his, indirect and error-laden, way is yielding the more correct of the results.
Tard.
Man, reading your post makes me feel like I am in Bizarro land.
So having a decent client and seamless integration, as opposed to a shitty webinterface, is invasive nowadays? Good god.
Yup.
If you have been working for a decade with the same code, a certain blindness to issues will be inevitable. (In german, there is a word for it: Betriebsblindheit)
He might be right simply because of the outside perspective.
There is a reason they usually power down in the winter, when the french nuclear power plants have higher load from homes (Over here, there is no AC peak int he summer).
IIRC, LHC uses something between 250 and 350 MW power to run.
IIRC, older android editions on the Nexus7 used to automount USB-sticks added via an adaptor cable. Google disabled this (fuckers).
Becaause you never recharge your battery on travels, right?
Drop those full size USB ports, and add a (micro) SD card slot.
It is totally ridiculous that all NEXUS devices are missing that one, even the new Nexus 10.I want to watch movies in a plane, or review my pictures away from my PC (where a 2560x1600 screen really would help). So fuck the cloud and fuck the tiered pricing system that askes for $100 more for adding $20 worth of flash - while STILL limiting the total capacity to amounts that are ridiculously low for a device of that cost.
Full sized USB I can understand for missing : Those plugs are huge. They would literally be the thickest thing in the tablet.
If you like your fenix, you would be blown away by something like a Quark Mini CR2, or a Jetbeam RRT-01 with 18350 cells (0.001 lumen to 800 lumen with seamless control via a dimmer ring).
Also, yes, those LEDs are getting cheaper, quickly. 5 Years ago you had to pay $15-20 for a led that put out 150 lumen at 50 lumen/W. Now you can get a led that puts out >1000 lumen at >100 lumen/W for $6.
Sorry, but that is NOT an "maximum for the existing technology". The value you give is like saying "lightspeed is the maximum speed for a Space Shuttle".
When I was there is was raining. Quite disappointing.
Sorry, mail analogy is wrong.
http:/// == postcard, nothing to open there.
Sorry, Nokia managed that very well on their own...
Thats more than misleading!
Intel HD3000 has less than 4%! Its just that that ATI and Nvidea are distributed over scores of indidual card types that each have a couple % that this ends up on top.
I never could get into planescape, because of the amount of text (dialogue and other) you had to work yourself through.
I like to read, but not 100ks of words in low-res fonts in a small part of a VGA window.
The masks alone would cost more.