If it's in your house or an office, sure. If it's in a datacenter, you're on a fixed power and thermal budget and 1W of heat that you generate takes roughly another 1W to pull out, and you can't go over the power budget. So, running AMD in a data center that could just about fill its racks with Intel servers, will necessarily give you less processors and some empty, expensive space that you can't use. I can fully agree, though, that for home or office use, AMD is the way to go. I only wish hackintoshes ran AMD:)
Calculating the function accurately for ALL input values instead of just for 0.4 takes an unpredictable amount of time.
There's no reason to believe that this is true. The set of inputs is fixed - it's either 2^32 or 2^64 values, and not all of those bit patterns represent valid IEEE-754 "floats". So, the amount of time is finite, and the amount of memory is finite - it's a O(1) lookup, with memory cost proportional to the size of the domain. All you need for a completely accurate single precision FSIN is a 4GB table in RAM. An FPU implementation of course doesn't need such tables, but IMHO a table is a very reasonable starting point the everything else is to be compared to. Having implemented a few single precision sines in my time, I can fully attest that they return completely accurate single precision results, and that has been tested by iterating all possible input values. I don't think it'd be any different for a 64 bit FSIN implementation. Heck, if you've got a supercomputer time to borrow, you should be plenty able to quickly iterate those 2^64 inputs to verify that they produce fully accurate double precision outputs.
The fucking reason for all of that is precisely the baroque non-design of CSS, the lack of a formal specification (whatever passes for it nowadays is a joke), and the fact that noone really can truly claim how this monstrosity was supposed to work. Most "introductions" to CSS are using analogies that are really unhelpful. Whatever "models" model the behavior of CSS are really not very useful, because the number of cases where those abstractions break down hopelessly is so high as to make them useless. It's like teaching quantum mechanics using any of the historical analogies (balls joined by springs, oscillators, waves, whatever). The only true "spec" of the CSS, in practice, is the browser behavior, as implemented. That's just sad. The main problem is that all CSS users, and I mean all accurate down to 10ppm or so, have no formal CS education, and if they do, they didn't internalize enough of it to understand what's wrong, and what should be done to fix it. It's like the Lemmings game (cuz the real lemmings don't).
I'd classify it as a fuck-up if you need to re-read things to get the meaning out of what should be a simple sentence that refers to concepts most/. readers take for granted.
It's not complex, but it's about as intuitive as the PHP libraries, and it requires such a baroque approach to the simplest of things that I really think the whining is well-deserved. Just because you can learn how to use this abomination doesn't mean that it's any good. It seems to be tightly tied to whatever box model Opera A/S was using at the time. That box model was demonstrably decently performing back then, but that's no reason to *expose* such an optimization as an interface everyone has to use to lay out their pages, for crying out loud.
It is my very firm opinion that whoever came up with CSS should not be putting it on his/her resume, because I wouldn't be hiring them for this monstrosity. It is a mess that's so thoroughly convoluted and unintuitive to use as to make the people involved in the core design essentially persona non grata of the tech world. I wouldn't want those people designing anything that I myself would have to deal with, no, thank you.
The idea behind the CSS is sound. The implementation is beyond horrible. It tells you something when you need to go through less hoops to get the layout you want in TeX, than in CSS, and TeX's basic block model was not designed for most of what it's being used for nowadays. CSS's design is an abomination that perhaps resulted from not having much in the way of exposure to the high-level designs of any layout software, or perhaps in the designers having grandiose aims with not much experience to back it up. Either way, it's just bad. I'd call CSS barely functional, like you have with some alcoholics who can hold a job and go through life "mostly fine" - as in without quite killing themselves or anyone else around them.
As an example of an API/interface design, CSS is precisely the way nobody should be doing it.
Technically, these days you don't need a phone number to initiate a phone call. You can get outbound-only VOIP service where there literally isn't a number where someone could call you back. So sometimes it's accurate to say that the robocallers have no callback number - it literally doesn't exist. This is very different from unlisted phone numbers, of course, since those by definition exist but merely are not listed.
Oh, the person with the unlisted number has called me. If they did it purposefully, I see no reason they have any standing to hide behind an unlisted number. My privacy is as valuable as theirs. If they've pocket-dialed, tough luck. I'm still at the receiving end of the call.
Moreover, unlisted numbers aren't 128 bit hashes that noone has time to enumerate. It's not as if I can't call an unlisted number. Heck, it's easy to corral the unlisted numbers, since they are disjoint from the listed numbers. Start with a set that spans the range of valid 7-digit phone numbers in a given area code. Then remove the listed numbers. Then remove the numbers that get connection errors. You are left with unlisted numbers. Such scans, in the day and age of VOIP, are rather easy to do.
And, finally, many digital connection providers pass an unlisted number along with merely a bit indicating the the number is not to be presented at the terminal. So the information is there, and it doesn't take but an asterisk setup to leverage that.
So yeah, they telco is just stalling here, nothing new...:(
I've met a few self-proclaimed assholes who try very hard to maintain a hardy image and are otherwise harmless and haven't even attempted to hurt anyone, much less carried out their "recollections". The "not hiding the fact" can occasionally be a fabrication. The more I read, though, the more it appears that perhaps here the concerns are genuine and the lady is above suspicion.
What is there to take his word over her word, or even to take either's word at face value? Exceptional claims require exceptional proof, and all that... The safest stance, without lots of corroboration, is for me to assume that they've both got serious issues. Sometimes the simplest explanation might be the correct one. I'd love to be shown otherwise, but whatever I could find online was just rather unconvincing.
HDCP in a freakin' camera? I don't think so. And I'm sure one can call it whatever they want, once the things ship there can be fees paid, compliance tests done, etc.
For prototyping you can do whatever you want, if you've got the money to potentially wreck a $10k chip (some high-end FPGAs cost that much). For production, you really don't want any hair or dust under those hi-density BGA chips. Really, decent electronics manufacturing requires clean room conditions, no matter what you might think. When you buy anything made by brand-names like Apple, the inside is always much, much cleaner than the outside. In anything with an image sensor in it, you'd want it to be triple so:)
I must add that this is not about the entire Antarctica, but about the South Pole (Amundsen-Scott Station) and similar inland locations. The coastal areas aren't as problematic, IIRC.
I only used the word "microclimate" to mean that the air temperature and moisture are maintained at normal levels indoors - to prevent trivializing the problem as "it must be the cold". You get a razor cut anywhere else on Earth, and it's normally no biggie, you don't think about it after an hour or two at most. You get a razor cut on Antarctica, you'll be lucky when it's completely healed a month later. By complete healing I mean a state where the skin has mostly normal properties, same thickness, same resiliency, etc. - basically the things we take for granted.
Frankly said, a paid-for VID/PID is not necessary for very much these days. You can piggyback on some defunct vendor's VID and use whatever PID you fancy. I don't think that DRM-free HDMI requires any licensing? I agree with other points though. They really need someone manufacturing savvy to set up their garage. I can't really see soldering their stuff in uncontrolled atmosphere either, you need clean room conditions - not hard to rig up a garage for it, but you need to have some upfront engineering experience first. Or at least very, very good feel for the experimental method, since you'll be tweaking the process to no end and must be very quick thinking to validate the hypotheses as to why things don't work, and the fixes for them.
"right about unified memory architectures" So, that pretty much settles it for me: their shaders push the memory subsystem to its knees and at 1080p the CPUs don't have enough memory bandwidth left to run at full speed.
In fact, I'd think their problem will be dumping waste heat, as any sort of air cooling would need *enormous* radiators. They might need geothermal cooling, in fact. Especially if they'd use nuclear power with a steam cycle.
There's something to this. For reasons we absolutely don't understand, healing even the simplest of injuries takes forever in Antarctica - mind you, even for people who live completely indoors, in climate controlled environment not much different from any modern city dwelling. So, somehow, when humans are in truly alien environment, even if the microclimate is right, they somehow suffer for it. There's no reason to believe that this effect would be necessarily absent on Mars. I'd consider it a nice surprise if it wasn't present there, in fact.
A 70s-level of tech could be achieved with much less, though. A "bright renaissance man" of sorts, living on Earth, with access to the first-hand know-how of right people, could probably bootstrap a couple of rudimentary 10um semiconductor processes - including optics etc., starting with a very limited set of tooling and raw materials. My only worry is really about the materials - are there enough concentrated ores on Mars? From what little I've read it'd seem that Mars's and Earth's geology are totally different and Mars is the last place you would want to mine anything on... Perhaps I'm wrong?
The real bootstrapping force these days is computing power. A couple of modern laptops provide incredible productivity gains over what you had to do in the 70s. So I think that the bootstrap to something like tech from 40 years ago would be possible, but you'd need to have modern computing power handy to do all that. Writing compilers is much easier on a modern LCD than on a home-built green phosphor CRT:)
Slashdot Mirror
This is especially egregious given that all the arguments and results are representable in IEEE-754 with no loss of precision!
If it's in your house or an office, sure. If it's in a datacenter, you're on a fixed power and thermal budget and 1W of heat that you generate takes roughly another 1W to pull out, and you can't go over the power budget. So, running AMD in a data center that could just about fill its racks with Intel servers, will necessarily give you less processors and some empty, expensive space that you can't use. I can fully agree, though, that for home or office use, AMD is the way to go. I only wish hackintoshes ran AMD :)
Calculating the function accurately for ALL input values instead of just for 0.4 takes an unpredictable amount of time.
There's no reason to believe that this is true. The set of inputs is fixed - it's either 2^32 or 2^64 values, and not all of those bit patterns represent valid IEEE-754 "floats". So, the amount of time is finite, and the amount of memory is finite - it's a O(1) lookup, with memory cost proportional to the size of the domain. All you need for a completely accurate single precision FSIN is a 4GB table in RAM. An FPU implementation of course doesn't need such tables, but IMHO a table is a very reasonable starting point the everything else is to be compared to. Having implemented a few single precision sines in my time, I can fully attest that they return completely accurate single precision results, and that has been tested by iterating all possible input values. I don't think it'd be any different for a 64 bit FSIN implementation. Heck, if you've got a supercomputer time to borrow, you should be plenty able to quickly iterate those 2^64 inputs to verify that they produce fully accurate double precision outputs.
The fucking reason for all of that is precisely the baroque non-design of CSS, the lack of a formal specification (whatever passes for it nowadays is a joke), and the fact that noone really can truly claim how this monstrosity was supposed to work. Most "introductions" to CSS are using analogies that are really unhelpful. Whatever "models" model the behavior of CSS are really not very useful, because the number of cases where those abstractions break down hopelessly is so high as to make them useless. It's like teaching quantum mechanics using any of the historical analogies (balls joined by springs, oscillators, waves, whatever). The only true "spec" of the CSS, in practice, is the browser behavior, as implemented. That's just sad. The main problem is that all CSS users, and I mean all accurate down to 10ppm or so, have no formal CS education, and if they do, they didn't internalize enough of it to understand what's wrong, and what should be done to fix it. It's like the Lemmings game (cuz the real lemmings don't).
I'd classify it as a fuck-up if you need to re-read things to get the meaning out of what should be a simple sentence that refers to concepts most /. readers take for granted.
It's not complex, but it's about as intuitive as the PHP libraries, and it requires such a baroque approach to the simplest of things that I really think the whining is well-deserved. Just because you can learn how to use this abomination doesn't mean that it's any good. It seems to be tightly tied to whatever box model Opera A/S was using at the time. That box model was demonstrably decently performing back then, but that's no reason to *expose* such an optimization as an interface everyone has to use to lay out their pages, for crying out loud.
It is my very firm opinion that whoever came up with CSS should not be putting it on his/her resume, because I wouldn't be hiring them for this monstrosity. It is a mess that's so thoroughly convoluted and unintuitive to use as to make the people involved in the core design essentially persona non grata of the tech world. I wouldn't want those people designing anything that I myself would have to deal with, no, thank you.
The idea behind the CSS is sound. The implementation is beyond horrible. It tells you something when you need to go through less hoops to get the layout you want in TeX, than in CSS, and TeX's basic block model was not designed for most of what it's being used for nowadays. CSS's design is an abomination that perhaps resulted from not having much in the way of exposure to the high-level designs of any layout software, or perhaps in the designers having grandiose aims with not much experience to back it up. Either way, it's just bad. I'd call CSS barely functional, like you have with some alcoholics who can hold a job and go through life "mostly fine" - as in without quite killing themselves or anyone else around them.
As an example of an API/interface design, CSS is precisely the way nobody should be doing it.
It's the law. It's illegal for them not to provide that information to him. Sure, they don't owe him anything :)
Kerosene? No on the instances I manage. It's all coal and steam powered.
Technically, these days you don't need a phone number to initiate a phone call. You can get outbound-only VOIP service where there literally isn't a number where someone could call you back. So sometimes it's accurate to say that the robocallers have no callback number - it literally doesn't exist. This is very different from unlisted phone numbers, of course, since those by definition exist but merely are not listed.
Oh, the person with the unlisted number has called me. If they did it purposefully, I see no reason they have any standing to hide behind an unlisted number. My privacy is as valuable as theirs. If they've pocket-dialed, tough luck. I'm still at the receiving end of the call.
Moreover, unlisted numbers aren't 128 bit hashes that noone has time to enumerate. It's not as if I can't call an unlisted number. Heck, it's easy to corral the unlisted numbers, since they are disjoint from the listed numbers. Start with a set that spans the range of valid 7-digit phone numbers in a given area code. Then remove the listed numbers. Then remove the numbers that get connection errors. You are left with unlisted numbers. Such scans, in the day and age of VOIP, are rather easy to do.
And, finally, many digital connection providers pass an unlisted number along with merely a bit indicating the the number is not to be presented at the terminal. So the information is there, and it doesn't take but an asterisk setup to leverage that.
So yeah, they telco is just stalling here, nothing new... :(
I've met a few self-proclaimed assholes who try very hard to maintain a hardy image and are otherwise harmless and haven't even attempted to hurt anyone, much less carried out their "recollections". The "not hiding the fact" can occasionally be a fabrication. The more I read, though, the more it appears that perhaps here the concerns are genuine and the lady is above suspicion.
What is there to take his word over her word, or even to take either's word at face value? Exceptional claims require exceptional proof, and all that... The safest stance, without lots of corroboration, is for me to assume that they've both got serious issues. Sometimes the simplest explanation might be the correct one. I'd love to be shown otherwise, but whatever I could find online was just rather unconvincing.
HDCP in a freakin' camera? I don't think so. And I'm sure one can call it whatever they want, once the things ship there can be fees paid, compliance tests done, etc.
For prototyping you can do whatever you want, if you've got the money to potentially wreck a $10k chip (some high-end FPGAs cost that much). For production, you really don't want any hair or dust under those hi-density BGA chips. Really, decent electronics manufacturing requires clean room conditions, no matter what you might think. When you buy anything made by brand-names like Apple, the inside is always much, much cleaner than the outside. In anything with an image sensor in it, you'd want it to be triple so :)
I must add that this is not about the entire Antarctica, but about the South Pole (Amundsen-Scott Station) and similar inland locations. The coastal areas aren't as problematic, IIRC.
I only used the word "microclimate" to mean that the air temperature and moisture are maintained at normal levels indoors - to prevent trivializing the problem as "it must be the cold". You get a razor cut anywhere else on Earth, and it's normally no biggie, you don't think about it after an hour or two at most. You get a razor cut on Antarctica, you'll be lucky when it's completely healed a month later. By complete healing I mean a state where the skin has mostly normal properties, same thickness, same resiliency, etc. - basically the things we take for granted.
unless they decided to act sane and not follow the instruction of the judge
You must've missed that part :) It means the same even if I don't use the buzzwords.
Frankly said, a paid-for VID/PID is not necessary for very much these days. You can piggyback on some defunct vendor's VID and use whatever PID you fancy. I don't think that DRM-free HDMI requires any licensing? I agree with other points though. They really need someone manufacturing savvy to set up their garage. I can't really see soldering their stuff in uncontrolled atmosphere either, you need clean room conditions - not hard to rig up a garage for it, but you need to have some upfront engineering experience first. Or at least very, very good feel for the experimental method, since you'll be tweaking the process to no end and must be very quick thinking to validate the hypotheses as to why things don't work, and the fixes for them.
Today's JVM runs much better on an old Pentium than the JVM that was released concurrently with said Pentium.
"right about unified memory architectures" So, that pretty much settles it for me: their shaders push the memory subsystem to its knees and at 1080p the CPUs don't have enough memory bandwidth left to run at full speed.
What if the consoles have shared system memory and they're hitting the memory bandwidth with the graphics?
In fact, I'd think their problem will be dumping waste heat, as any sort of air cooling would need *enormous* radiators. They might need geothermal cooling, in fact. Especially if they'd use nuclear power with a steam cycle.
There's something to this. For reasons we absolutely don't understand, healing even the simplest of injuries takes forever in Antarctica - mind you, even for people who live completely indoors, in climate controlled environment not much different from any modern city dwelling. So, somehow, when humans are in truly alien environment, even if the microclimate is right, they somehow suffer for it. There's no reason to believe that this effect would be necessarily absent on Mars. I'd consider it a nice surprise if it wasn't present there, in fact.
A 70s-level of tech could be achieved with much less, though. A "bright renaissance man" of sorts, living on Earth, with access to the first-hand know-how of right people, could probably bootstrap a couple of rudimentary 10um semiconductor processes - including optics etc., starting with a very limited set of tooling and raw materials. My only worry is really about the materials - are there enough concentrated ores on Mars? From what little I've read it'd seem that Mars's and Earth's geology are totally different and Mars is the last place you would want to mine anything on... Perhaps I'm wrong?
The real bootstrapping force these days is computing power. A couple of modern laptops provide incredible productivity gains over what you had to do in the 70s. So I think that the bootstrap to something like tech from 40 years ago would be possible, but you'd need to have modern computing power handy to do all that. Writing compilers is much easier on a modern LCD than on a home-built green phosphor CRT :)