Andoid and iOS. With windows 8/wp8 they'll be a mostly unified platform but not yet.
Gaming time on 360/PS is taken up almost entirely by AAA titles you aren't going to win timeshare, which means sales. The place to compete is gaming time that isn't on one of those platforms.
How do they sell 'more hardware'? 100% of computers have video cards in them. If you're cheap you get intel, if you want gaming you get Nvidia/AMD in a 60/40 split more or less. And they're already in the HPC space (and have been for several years).
Unless you have some great way for how an open source driver could let you run database queries on a GPU that a closed source can't, there aren't a lot of places they want to pickup sales.
There aren't enough people who care about open source to shift the 60/40 split in the gaming space. And those guys don't want to have an open source driver because if there's a huge layer of liability if the video cards die due to a driver bug (which by the way, they do, but at least when nvidia releases the driver they're the ones liable, so it's pretty cut and dried).
Right now you can't do enough better on a PC to warrant doing major PC exclusive graphics compared to the 360/PS3. Higher resolution, better FPS sure, but not significantly better. Intel is reasonably closing the gap on PS3/Xbox 2 level performance, but that puts then at about 0.1 of a good graphics card*.
Of course the 'next gen' consoles are in the making now, and that means we'll see consoles about on par with what you can do with a decent rig today. So then the next cross platform titles (which will be Xbox 3, PS4, PC) won't run anywhere near viably on intel integrated graphics.
Sure, intel just wiped out their competition in the low end casual/non gamer market, but that was never a great market anyway.
*Yes high end GPU's are easily 10x the PS3/Xbox2 GPU's, but you have a lot more going on on the PC, which is a certain degree of inefficiency, and you have high resolutions and framerates, and some visual candy that isn't on the consoles which takes up the power.
These are a hell of a lot better than that. They aren't good, but they will manage to play skyrim for example (albeit at a relatively low resolution for a decent framerate). http://www.anandtech.com/show/5626/ivy-bridge-preview-core-i7-3770k/15
These are probably the wrong direction for the product though. They don't viably compete with a discrete GPU, so people who can, would rather not have to buy an integrated GPU at all, and for business it's so powerful it's letting employees game on work computers, which isn't good.
On the other hand, for the home user market, it's good enough that people can do whatever they want on the machines and they'll manage.
do you have a proven track record of being able to produce 100 000 units? What's your typical cost/cost overun on a project that size?
Business is business. Engineering is part of business, but if you're advertising the greatest thing ever for 100 dollars that is supposedly 10x than what everyone else is selling for 1000 people are rightfully skeptical that you can actually deliver the product on time, and on budget. That doesn't mean you can't, and yes in any business advertising (or in the case of the US government lobbying) matters tremendously, but there can be non obvious factors at play.
As with anything you might really have been trumped by 'strategic concerns' (you weren't going to create enough jobs, in the right districts, or pay the right campaign kickbacks), but you might have just not seemed honest, being the only honest one in a room full of crooks.
Generally if there's something on there end causing a problem their own monitoring tools will find it, and they go out and fix it. If you call up and say I have X ms of latency, and everyone else in the area has 0.1 or 0.25X then it's clear the problem is with your end. For AT&T that number is apparently 200ms in one area.
Sometimes their own tools don't catch it, but then the guy who comes out looks at your setup, it all tests fine (the mentioned tracerouting in other post for example), and decides if there's a problem elsewhere in the system and is now assigned to fix it.
300ms seems quite high for most residential DSL though, especially to their own test server, unless you're really rural.
Um.... gpu's work on floats. I specifically know about the problem because I discovered it using GPU's for something. I only noticed it because I had the same problem ray tracing.
Floats are floats. the IEEE specifies how they behave. GPU's happen to be able to crunch a lot of them compared to a CPU, even when it's actually implementing the floating point specification correctly (early GPU's didn't quite). It's a matter of the capacity to represent decimal numbers in binary. When you only have X bits for mantissa, Y bits for an exponent there's only so much you can do. It's not a matter of a particular brand of hardware having precision, they're all the same, or supposed to be.
And yes, fixed point was the way to go, but for today, floats are probably your best bet performance wise, you just need to be aware that they will have errors because you just can't represent some numbers differently in floating point.
Oh and what have you, yourself done? Risked everyone else's lives by essentially putting on a blind fold, hence the analogy. If *you* are paying attention to your phone you are paying less attention to the road, risking everyone else's lives, now, the question is how much distraction is an acceptable risk, radio, screaming children, screaming spouse, texting?
If you aren't paying attention to the road, you shouldn't be on it. In that sense texting while driving is already covered, but then you get to my issue of 'awareness' you now have no excuse to say you weren't aware it was actually distracting you. By writing new laws about it they'd get more press and more awareness.
Should you not be charged with attempted murder for shooting into a crowded room and happen to miss? The fact that you didn't kill anyone makes it a significantly lesser charge than say... actually killing someone. But do we really need to wait until after you've killed someone because you were texting to say 'you shouldn't have been texting'?
Should you be allowed to blind fold yourself when driving? Until you caused an accident it was perfectly safe right?
You shouldn't be allowed on the road if you're going to be reckless. What meets the standard of reckless is very much up to scientific analysis and public disclosure of that knowledge. Any idiot should be able to figure out that you should not be allowed to drive while blind folded. But people don't realize they're distracted when on a cell phone. You'd think they'd realize they're distracted when texting, but people definitely don't appreciate just how distracted they are when on a hands free device 'my hands are on the wheel!' is not the same as 'my attention is on the road'.
A floating point precision error. Floating points cannot represent quite a diverse collection of numbers, this is especially problematic when you're doing intersections with small objects. Say a ray projected from an object will, because of the minute errors in floating point, collide with the same object (which produces some cool patterns).
Floating points are kind of crappy. Not that I have a better option with viable performance on a desktop machine. That's not a division bug, that's just the nature of representing numbers in binary with a fixed number of bits.
Are you sure they don't? I did IT work for a guy who had a studio in toronto where he did guitar work for artists, which they did real time collaboration with other studios to integrate things. The artist could be in London or Los Angles or the like and they had some setup where they could collaboratively combine all of that stuff at once. I know his studio did voice work as well, but his specialty was guitar so I know they had some special hookup on the guiltar so it would transmit specially (i.e. the full quality of the guitar audio or something).
I was just selling him some office computers so I don't know the details of how the setup works sadly, or I'd say.
And even then, if your studio in London takes track A, your studio in LA takes track B, and your other studio in LA track C it's essentially the same problem.
A motorist retains right as a person. If you cannot bar a persons right to speech in a park on the street etc. can you ban it when they are driving? This goes to the falsely shouting fire in a crowded theatre argument, and you do not have a right to that. In this case any speech at all creates a significant danger to other motorists.
Think of it this way: A person is a motorist. You cannot take away their right to have "I hate/.ers" bumper stickers - that's a non infringing free speech.
Whether or not there is a 'right' to use roads has been clearly long settled. You could get into a question of rights if the government could prohibit you from taking a driving test because you're say black, or a woman. But as it is there is a universal right to be considered to the same standard as everyone else for driving or a passport or the like.
Whether or not laws against murder actually deter murder doesn't matter to whether or not their should be laws against murder. If an 27 year detention period is a detriment, but a 25 year one not, then you're having a useful discussion.
Texting while driving is really f'n dangerous to you, and to everyone else on the road. It kills people and causes significant material losses, so pretty obviously it warrants rules against it (those laws in many cases already exists, a study like this merely clarifies that they should apply to those existing laws). Beyond that it is a matter of degree as to what should or should not be done to keep people off the road. Texting while driving should be illegal, the question is what is the most effective enforcement mechanism to reduce it. A $7 fine? A $170 fine? a $1700 fine? Roadside seizure? (where the car is towed but still owned by the driver they can pick it up later), confiscation? Jail (for how long?).
If you can't reasonably decide if something is bad (e.g. swearing) then you can't reasonably compose laws about it, sure. But in the context of what you can do while driving, texting, talking on the phone (hands free or otherwise) it's clear it shouldn't be allowed.
This has significant implications for car manufacturers, which is less about criminal laws and more about regularly compliance. If you using a hands free phone while driving should be banned well car companies need to stop making hands free cell integration into their vehicles, *or* find a way that it will reduce the distraction effect.
Again though, if you have 25 studios all around the world, each one of which could be working on all or part of a track, it becomes very hard to manage thousands of separate pieces of data.
It's not that the potential security arrangements are impossible, they certainly aren't, record companies did business long before the internet, so that's even an option. It's that an effective, collaborative workflow for hundreds or thousands of employees around the world, or even on one large facility, it's a time wasting nightmare to be running through layers of security to send it from building 1a to building 9 100 metres away.
Lots of businesses manage to keep their employees connected, have files of various sorts accessible, because they have security they feel is capable of keeping all of that material secure. They might be wrong, and sony might have thought it had adequate protections. We can't on one hand argue for all of these collaborative tools that really do improve workflow and then say no one should use them because they're insecure. If they are insecure it's a matter of figuring out how secure we can make them.
No matter what though, if people have access to the data, they could release it.
Except that it's not really factually true, they're just trying to make it sound like democracy is better than it is. People may not be able to vote well on issues like tax reform, but they're probably competent enough on issues like "lets kill all the jews" or "lets ditch birth control". Dictators are generally smart enough to find the right people to take over a country, and to have organized a coup in the first place. They know who is important enough to ally with, payoff or the like, and if you look at several of the dictators in history:
Hitler: Took out poland in 5 week, france in 6, got to the gates of moscow in 6 months. The last two Kims in north korea: managed to keep themselves sufficiently popular with the right people that they died of old age, firmly in charge, and they chose their successors. Saddam Hussein: He did run the country for something like 30 years, and when push came to shove, he did the right thing and ditched his WMD's. (Gaddafhi was something similar, he ran the place for 40 years, and then even restored relations with the west).
The problem isn't that they aren't smart or don't pick good people to solve lots of problems. It's that they have an added problem of trying to keep in power which necessitates being evil.
Monarchy produces leadership who are, at least, prepared for the fact that they're going to be in charge, and encourages a well oiled bureaucratic process, because if "I am the state" I have to worry about big picture metrics - I need to know what they are, and how to interpret them, and I spent time learning those until my parent died, and if someone if failing to provide good results you know who they are.
In short: Dictators tend to be smart, but have misguided priorities. They're above average people, but their goals primarily are concerned with how they will stay in power. Monarchs are concerned with staying in power as well, but they're also concerned about their children being in power, and grand children, and they were hopefully prepared for it, they're bad (without a parliamentary balance) for everyone not in the upper class because they like wealth to distribution to be bad, and for it to stay that way. Democracies are concerned with the will of the moment, facts be damned. But it provides significant bonuses to live in a country where people believe they're doing the right thing by the will of the majority, unity of effort sort of thing. People are more willing to go along with poorly thought out ideas if they think they have broad support, and in turn because they go along with it, it helps those ideas congeal into something workable.
People on average really aren't capable of judging the skills of potential leaders, or whether or not their '999' plan is good or bad. Even if someone else does the analysis they still don't get it. The point then is to balance the people elected to power, with people appointed to power, and with people who inherit power etc. The UK seemed to sort of circle around the sweetspot for about a century, as did Prussia/germany for a while. A Monarchy is an institution that balances the reality that the rich have a lot of say in government, and it is concerned with sustainability for multiple generations, but it needs to be balanced with the realities, desires (even if they're unrealistic desires) and available funds of the masses. At the same time it has the appointed leadership (judges) who are explicitly supposed to be experts. When you only have rule by the masses people end up electing george bushes, rich santorums, adolf hitlers, *and* there's nothing to reign in the crazy. The US system isn't a democracy, it's a democratic system, but those idiots are balanced off against people who are supposed to be experts (ultimately the supreme court). The problem is that the supreme court only serves to ask if something is legal, not if it's good. All of the people tasked with assessing if a plan is good, for example congressional research services or the like, don't have to be listene
They seem to have a much bigger product lineup than they used to. Windows isn't just a kernel, it's a whole software suite, and windows 8 seems like it's a lot of development time. The whole azure/cloud service thing, and all of the overhead that goes with that is a lot of work, the security products etc. Even add in.net, silverlight, skydrive, their server stuff, that's a lot more than the company did 15 years ago.
Admittedly, I look at software through the lens of the game business, but to make a piece of software today takes a lot more people than a piece of software 10 or 15 years ago that you can still only sell for 60 bucks. It's not even necessarily programmers, but technical design people, scientists etc. If you're going to sell 500 million copies of something (and windows will sell a lot more than 500 million copies), the difference between very little things makes a very big difference, and you spend a lot of time fussing over details. I'll be very interested to see windows 8, given the scale of the departure of their user experience from anything before. You have to figure there has been a mountain of research done on everything from supporting proper english, to the precise layout and organization of their new mail client, and the core technical stuff about how to have an API for contacts (people I think they call it), how to build applications in whatever the new framework is around which the new UI exists and works etc.
And ya, it's hard to know what the new microsoft is even trying to do, let alone how many people they have doing it. At one point you were sure they were trying to take over the world. Now it's not even clear if they have a strategy other than reacting to a changing marketplace by changing for the sake of being in a changing marketplace. Good business is either guessing what the future will want, and being positioned for it, or making the future with into your vision of what it should be. Microsoft doesn't seem to have a clear sense of what it's trying to do, or why or how it's going to get it done.
Maybe it seems like they've been bulking up on lawyers because/. covers lawyers more than developers, and because developers that work for microsoft get almost no individual press for their work.
That's in part because MS doesn't let you sell your code on the side, and in part because you work for Microsoft, as part of a team, and externally the team gets credit. You may be clearly listed as part of a team, and credited for the team, but mostly we only see one point person for a team (who blogs or gives talks) and everyone else we don't know about. Which, given the range of products MS has would be prohibitively hard to keep track of from the outside anyway.
That's a pretty big assumption. And it assumes that the information he would have is valuable.
He's almost certainly not the first FTC person ever hired by microsoft, directly or indirectly. They probably already know the inner workings of the FTC. The question is whether or he specifically has knowledge of a business plans from google that could damage them somehow, but benefit MS.
More likely he merely adds more man hours to their access to the FTC and their ability to do the proper paper work in the proper way, for the proper people, and so they can support more paperwork.
Obviously he wasn't all that effective at his last job if they ignored his recommendations. But he does know what the review process is, and that's useful to have on staff so they can file the right paperwork.
Well 2 years is a two fold issue. First, as said elsewhere, you need to feed yourself for 2 years. Secondly, in 2 years hopefully most of the existing short term contracts have gone through the system, and it's all new stuff. You can't lobby for a bid already in progress sort of thing (where you were on the inside looking evaluating a bid, and then turn around and tell the company what to do to change it so it's successful).
Sort of by definition you will still know most of the people there and where they fit in the hierarchy, but that's not a huge problem.
That's mostly because those of us who make games are still writing for 360/PS3 and the CPU isn't the limiting factor so much on the PC side. I'm doing some research atm, and the new CPU's are even clock for clock 20% faster per generation if not more on the same task without any optimization. You just don't see the benefit because you aren't CPU limited. Which was my point about replacing with a lower tier. If you're happy with a q6600 general performance then the replacement line isn't i7-920 - i7 2600 it's to an i5 of some sort which are like 170 bucks compared to 300. I think a new i5 2300 will run about 2x as fast as a q6600, and it's 190 bucks retail including retail packaging and cooler, not counting mobo obviously. The q6600 was a 300 dollar part at retail, though I think I paid 400 for mine, but that might have been a USD-CDN conversion thing. You could probably to match performance of a q6600 with an i3 g8xx which are less than $100 retail, and only dual core (but are clock for clock much faster, and clocked higher). It might be a bit slower, hard to say without spending more than 5 minutes doing analysis. You could completely replace your ram, MOBO and CPU for probably 300 bucks at that point, and you'd be getting something brand new, which supports newer memory standards (see below). The way I price it out on newegg, and I'm not in the US so I'm not sure if it's lying to me (I'm using newegg us not newegg canada, which doesn't lie to me): A gigabyte 1155 board, 105 dollars, an i 13 g860 is 100 bucks and 8 gigs of ram is 75. You'll have a tough time convincing me that being able to replace the system for actually slightly better overall system performance for about 1/3rd the price of what you paid is somehow bad for you if a part fails. A new mobo, ram and CPU is actually less than you would have paid for the CPU.
The big thing is *memory*. Not ram though, nothing uses enough ram for 6/8/12 to make much difference. USB3 and SATAIII. The difference in user experience between an extremely fast raid 0's pair of drives on q6600 and an SATAIII SSD is enormous. It's one of those defining technologies that made smart phones viable. They're worth having in a computer that you actually use, but it's hard to convey how much better a full speed SATA III SSD is compared to say... anything else ever. Booting windows in 16 or 17 seconds (I take about 2 seconds to log in), and then being able to load say a game like WoW or SWTOR in a matter of seconds completely changes the experience. And they aren't hacking that back into a core 2 mobo.
Your statement on the LGA 775 is correct generally, it lasted a long time, and you bought right at the end of its lifecycle (as did I). But there were different product range sockets available even then (771's for example). The LGA 1366 - 1156 is the same thing. the 1156 isn't a replacement for the 1366 nor the reverse. They're for different markets and different CPU's. The LGA 1366 is superceded by the LGA 2011, but it stuck around for about 4 years as well, and the LGA 1155 supercedes the LGA 1156 in the mid range, and again, stuck around about 4 years. So yes, intel has released several sockets, but they're for different product ranges, they don't all supercede each other.
If you want to go from a q6600 on a 775 to a i7 920 on a 1366 to a i7 2600 on a LGA 2011 then sure you're talking about the last part to use a socket, to the first part to use a socket, each with about a 4 year life cycle. So you expect there to be some overlap. How far into the future LGA 2011 will last I don't know, afaik it will support ivy bridge though, which should take us through 2013 at least. But you'd also be sticking in the product range of being at the bottom of the high end range. You could quite reasonably drop down a tier at least from the q6600 and have the same CPU power.
To try and illustrate High end 775 (2004-2008) *you are here* - 1366 (2008 - 2012) -> LGA 2011 (nov 2011 -> unknown, but some ivy bridge) Mid range 771 -
This is in part because every university has its own programme arrangement, meaning the greatest textbook ever from some other university will only cover half of what is to be covered in your course, or it will cover something in a confusing way.
At highschool and lower this doesn't need to happen. Everyone, or nearly everyone should be on the same national curriculum, with the same textbooks. That will never happen in some countries though (like the US). You can even leave some of the day/week/month to 'teacher time' where they decide what to cover.
Unless it's a big class you don't get enough in royalties per book for that to be a particularly good plan. And if it is a big enough class it's important that the textbook fits with whatever the broader strategy of the university is, since you're training a huge portion of the student body with it.
If one of your parts fails, either you had the foresight to buy replacements when they were being cleared out from inventory, or you buy a lower end version of a current product that will have the same performance for less money.
I'm running one q6600 and one q6700 right now. I have some spare ram lying around. But if either CPU, or a mobo dies my best bet is to probably just buy a low end i5 series or whatever is in the sweet spot for price drops rather than trying to repair the old stuff. If you can't recognize a product cycle, or what it does to prices, you probably can't recognize whether or not a q6600 is adequate or inadequate for your current usage, because you don't understand enough about what the products do and how that translates to a user experience.
This isn't some giant conspiracy against you trying to force you to upgrade 'prematurely'. This is a business that runs on1.5-2 year product cycles, and they still make hundreds of millions of parts within a product cycle. Those production lines shut down, new ones come online. If you only need to match the performance of your existing parts you buy lower end versions of future parts. It's not realistic to think that a core 2 quad, with what 550 million transistors, to an i7 920 with 700 and some, to a i7 -2600 (sandy bridge) with just shy of a billion are all going to just play nice on the same CPU sockets or necessarily with the same RAM. They're all ~300 dollar CPU's (which is why I chose them for comparison purposes).
There's a sweet spot where prices drop enough that you buy future spare parts, and then it becomes sufficiently antiquated it's impractical to buy spares unless it's an emergency. Most consumer electronics have a ~50% failure rate at 6 years anyway as capacitors give out.
This applies to any business, you just need to pay attention to the cycle for whatever business you're in. They probably stopped fabbing core 2 parts 4 years ago, and all of the periphery parts around the same time, so anything you're buying is that old.
Besides, they want you to replace the whole cpu/ram/mobo combo at once. They are sold separately for good reasons, but they all go together. If one part dies they're all designed for each other. And it's not like intel doesn't have new products at the similar and lower price points that will provide you the same performance, especially 5 years on. If your MOBO dies, your CPU is probably not far behind, and the ram not far behind that. If you don't need better performance just buy a lower end i5 (the q6600 was a mid range, buy a low end)
True, but, like most students in this day and age, there's a risk you can take on borrowing against your future so that you'll have a better future. Granted, like most of our students in general, that doesn't work out too well.
And society did help you with learning a new skillset. Compulsory education to age 16/18 is a huge step in the right direction. It means the maximum retraining time from any profession to any other is about 10 years (4 years bachelors, 2 years masters 4 years PhD) and more realistically, the training time from one profession to another is more like 2 years since you don't want to change from say being a tradeschool carpenter to medicine when you're 40.
Oh and some places do offer job training as part of being on unemployment (countries/provinces/states/localities, and some companies will offer retraining or severance packages which can be used for such purposes). So your statement isn't universal.
Slashdot Mirror
Andoid and iOS. With windows 8/wp8 they'll be a mostly unified platform but not yet.
Gaming time on 360/PS is taken up almost entirely by AAA titles you aren't going to win timeshare, which means sales. The place to compete is gaming time that isn't on one of those platforms.
How do they sell 'more hardware'? 100% of computers have video cards in them. If you're cheap you get intel, if you want gaming you get Nvidia/AMD in a 60/40 split more or less. And they're already in the HPC space (and have been for several years).
Unless you have some great way for how an open source driver could let you run database queries on a GPU that a closed source can't, there aren't a lot of places they want to pickup sales.
There aren't enough people who care about open source to shift the 60/40 split in the gaming space. And those guys don't want to have an open source driver because if there's a huge layer of liability if the video cards die due to a driver bug (which by the way, they do, but at least when nvidia releases the driver they're the ones liable, so it's pretty cut and dried).
Come back again in a console generation.
Right now you can't do enough better on a PC to warrant doing major PC exclusive graphics compared to the 360/PS3. Higher resolution, better FPS sure, but not significantly better. Intel is reasonably closing the gap on PS3/Xbox 2 level performance, but that puts then at about 0.1 of a good graphics card*.
Of course the 'next gen' consoles are in the making now, and that means we'll see consoles about on par with what you can do with a decent rig today. So then the next cross platform titles (which will be Xbox 3, PS4, PC) won't run anywhere near viably on intel integrated graphics.
Sure, intel just wiped out their competition in the low end casual/non gamer market, but that was never a great market anyway.
*Yes high end GPU's are easily 10x the PS3/Xbox2 GPU's, but you have a lot more going on on the PC, which is a certain degree of inefficiency, and you have high resolutions and framerates, and some visual candy that isn't on the consoles which takes up the power.
These are a hell of a lot better than that. They aren't good, but they will manage to play skyrim for example (albeit at a relatively low resolution for a decent framerate). http://www.anandtech.com/show/5626/ivy-bridge-preview-core-i7-3770k/15
These are probably the wrong direction for the product though. They don't viably compete with a discrete GPU, so people who can, would rather not have to buy an integrated GPU at all, and for business it's so powerful it's letting employees game on work computers, which isn't good.
On the other hand, for the home user market, it's good enough that people can do whatever they want on the machines and they'll manage.
do you have a proven track record of being able to produce 100 000 units? What's your typical cost/cost overun on a project that size?
Business is business. Engineering is part of business, but if you're advertising the greatest thing ever for 100 dollars that is supposedly 10x than what everyone else is selling for 1000 people are rightfully skeptical that you can actually deliver the product on time, and on budget. That doesn't mean you can't, and yes in any business advertising (or in the case of the US government lobbying) matters tremendously, but there can be non obvious factors at play.
As with anything you might really have been trumped by 'strategic concerns' (you weren't going to create enough jobs, in the right districts, or pay the right campaign kickbacks), but you might have just not seemed honest, being the only honest one in a room full of crooks.
Generally if there's something on there end causing a problem their own monitoring tools will find it, and they go out and fix it. If you call up and say I have X ms of latency, and everyone else in the area has 0.1 or 0.25X then it's clear the problem is with your end. For AT&T that number is apparently 200ms in one area.
Sometimes their own tools don't catch it, but then the guy who comes out looks at your setup, it all tests fine (the mentioned tracerouting in other post for example), and decides if there's a problem elsewhere in the system and is now assigned to fix it.
300ms seems quite high for most residential DSL though, especially to their own test server, unless you're really rural.
Um.... gpu's work on floats. I specifically know about the problem because I discovered it using GPU's for something. I only noticed it because I had the same problem ray tracing.
Floats are floats. the IEEE specifies how they behave. GPU's happen to be able to crunch a lot of them compared to a CPU, even when it's actually implementing the floating point specification correctly (early GPU's didn't quite). It's a matter of the capacity to represent decimal numbers in binary. When you only have X bits for mantissa, Y bits for an exponent there's only so much you can do. It's not a matter of a particular brand of hardware having precision, they're all the same, or supposed to be.
And yes, fixed point was the way to go, but for today, floats are probably your best bet performance wise, you just need to be aware that they will have errors because you just can't represent some numbers differently in floating point.
Oh and what have you, yourself done? Risked everyone else's lives by essentially putting on a blind fold, hence the analogy. If *you* are paying attention to your phone you are paying less attention to the road, risking everyone else's lives, now, the question is how much distraction is an acceptable risk, radio, screaming children, screaming spouse, texting?
If you aren't paying attention to the road, you shouldn't be on it. In that sense texting while driving is already covered, but then you get to my issue of 'awareness' you now have no excuse to say you weren't aware it was actually distracting you. By writing new laws about it they'd get more press and more awareness.
Should you not be charged with attempted murder for shooting into a crowded room and happen to miss? The fact that you didn't kill anyone makes it a significantly lesser charge than say... actually killing someone. But do we really need to wait until after you've killed someone because you were texting to say 'you shouldn't have been texting'?
Should you be allowed to blind fold yourself when driving? Until you caused an accident it was perfectly safe right?
You shouldn't be allowed on the road if you're going to be reckless. What meets the standard of reckless is very much up to scientific analysis and public disclosure of that knowledge. Any idiot should be able to figure out that you should not be allowed to drive while blind folded. But people don't realize they're distracted when on a cell phone. You'd think they'd realize they're distracted when texting, but people definitely don't appreciate just how distracted they are when on a hands free device 'my hands are on the wheel!' is not the same as 'my attention is on the road'.
A floating point precision error. Floating points cannot represent quite a diverse collection of numbers, this is especially problematic when you're doing intersections with small objects. Say a ray projected from an object will, because of the minute errors in floating point, collide with the same object (which produces some cool patterns).
Floating points are kind of crappy. Not that I have a better option with viable performance on a desktop machine. That's not a division bug, that's just the nature of representing numbers in binary with a fixed number of bits.
Are you sure they don't? I did IT work for a guy who had a studio in toronto where he did guitar work for artists, which they did real time collaboration with other studios to integrate things. The artist could be in London or Los Angles or the like and they had some setup where they could collaboratively combine all of that stuff at once. I know his studio did voice work as well, but his specialty was guitar so I know they had some special hookup on the guiltar so it would transmit specially (i.e. the full quality of the guitar audio or something).
I was just selling him some office computers so I don't know the details of how the setup works sadly, or I'd say.
And even then, if your studio in London takes track A, your studio in LA takes track B, and your other studio in LA track C it's essentially the same problem.
Completely unrelated.
A motorist retains right as a person. If you cannot bar a persons right to speech in a park on the street etc. can you ban it when they are driving? This goes to the falsely shouting fire in a crowded theatre argument, and you do not have a right to that. In this case any speech at all creates a significant danger to other motorists.
Think of it this way: A person is a motorist. You cannot take away their right to have "I hate /.ers" bumper stickers - that's a non infringing free speech.
Whether or not there is a 'right' to use roads has been clearly long settled. You could get into a question of rights if the government could prohibit you from taking a driving test because you're say black, or a woman. But as it is there is a universal right to be considered to the same standard as everyone else for driving or a passport or the like.
Whether or not laws against murder actually deter murder doesn't matter to whether or not their should be laws against murder. If an 27 year detention period is a detriment, but a 25 year one not, then you're having a useful discussion.
Texting while driving is really f'n dangerous to you, and to everyone else on the road. It kills people and causes significant material losses, so pretty obviously it warrants rules against it (those laws in many cases already exists, a study like this merely clarifies that they should apply to those existing laws). Beyond that it is a matter of degree as to what should or should not be done to keep people off the road. Texting while driving should be illegal, the question is what is the most effective enforcement mechanism to reduce it. A $7 fine? A $170 fine? a $1700 fine? Roadside seizure? (where the car is towed but still owned by the driver they can pick it up later), confiscation? Jail (for how long?).
If you can't reasonably decide if something is bad (e.g. swearing) then you can't reasonably compose laws about it, sure. But in the context of what you can do while driving, texting, talking on the phone (hands free or otherwise) it's clear it shouldn't be allowed.
This has significant implications for car manufacturers, which is less about criminal laws and more about regularly compliance. If you using a hands free phone while driving should be banned well car companies need to stop making hands free cell integration into their vehicles, *or* find a way that it will reduce the distraction effect.
Again though, if you have 25 studios all around the world, each one of which could be working on all or part of a track, it becomes very hard to manage thousands of separate pieces of data.
It's not that the potential security arrangements are impossible, they certainly aren't, record companies did business long before the internet, so that's even an option. It's that an effective, collaborative workflow for hundreds or thousands of employees around the world, or even on one large facility, it's a time wasting nightmare to be running through layers of security to send it from building 1a to building 9 100 metres away.
Lots of businesses manage to keep their employees connected, have files of various sorts accessible, because they have security they feel is capable of keeping all of that material secure. They might be wrong, and sony might have thought it had adequate protections. We can't on one hand argue for all of these collaborative tools that really do improve workflow and then say no one should use them because they're insecure. If they are insecure it's a matter of figuring out how secure we can make them.
No matter what though, if people have access to the data, they could release it.
Except that it's not really factually true, they're just trying to make it sound like democracy is better than it is. People may not be able to vote well on issues like tax reform, but they're probably competent enough on issues like "lets kill all the jews" or "lets ditch birth control". Dictators are generally smart enough to find the right people to take over a country, and to have organized a coup in the first place. They know who is important enough to ally with, payoff or the like, and if you look at several of the dictators in history:
Hitler: Took out poland in 5 week, france in 6, got to the gates of moscow in 6 months.
The last two Kims in north korea: managed to keep themselves sufficiently popular with the right people that they died of old age, firmly in charge, and they chose their successors.
Saddam Hussein: He did run the country for something like 30 years, and when push came to shove, he did the right thing and ditched his WMD's. (Gaddafhi was something similar, he ran the place for 40 years, and then even restored relations with the west).
The problem isn't that they aren't smart or don't pick good people to solve lots of problems. It's that they have an added problem of trying to keep in power which necessitates being evil.
Monarchy produces leadership who are, at least, prepared for the fact that they're going to be in charge, and encourages a well oiled bureaucratic process, because if "I am the state" I have to worry about big picture metrics - I need to know what they are, and how to interpret them, and I spent time learning those until my parent died, and if someone if failing to provide good results you know who they are.
In short: Dictators tend to be smart, but have misguided priorities. They're above average people, but their goals primarily are concerned with how they will stay in power. Monarchs are concerned with staying in power as well, but they're also concerned about their children being in power, and grand children, and they were hopefully prepared for it, they're bad (without a parliamentary balance) for everyone not in the upper class because they like wealth to distribution to be bad, and for it to stay that way. Democracies are concerned with the will of the moment, facts be damned. But it provides significant bonuses to live in a country where people believe they're doing the right thing by the will of the majority, unity of effort sort of thing. People are more willing to go along with poorly thought out ideas if they think they have broad support, and in turn because they go along with it, it helps those ideas congeal into something workable.
People on average really aren't capable of judging the skills of potential leaders, or whether or not their '999' plan is good or bad. Even if someone else does the analysis they still don't get it. The point then is to balance the people elected to power, with people appointed to power, and with people who inherit power etc. The UK seemed to sort of circle around the sweetspot for about a century, as did Prussia/germany for a while. A Monarchy is an institution that balances the reality that the rich have a lot of say in government, and it is concerned with sustainability for multiple generations, but it needs to be balanced with the realities, desires (even if they're unrealistic desires) and available funds of the masses. At the same time it has the appointed leadership (judges) who are explicitly supposed to be experts. When you only have rule by the masses people end up electing george bushes, rich santorums, adolf hitlers, *and* there's nothing to reign in the crazy. The US system isn't a democracy, it's a democratic system, but those idiots are balanced off against people who are supposed to be experts (ultimately the supreme court). The problem is that the supreme court only serves to ask if something is legal, not if it's good. All of the people tasked with assessing if a plan is good, for example congressional research services or the like, don't have to be listene
They seem to have a much bigger product lineup than they used to. Windows isn't just a kernel, it's a whole software suite, and windows 8 seems like it's a lot of development time. The whole azure/cloud service thing, and all of the overhead that goes with that is a lot of work, the security products etc. Even add in .net, silverlight, skydrive, their server stuff, that's a lot more than the company did 15 years ago.
Admittedly, I look at software through the lens of the game business, but to make a piece of software today takes a lot more people than a piece of software 10 or 15 years ago that you can still only sell for 60 bucks. It's not even necessarily programmers, but technical design people, scientists etc. If you're going to sell 500 million copies of something (and windows will sell a lot more than 500 million copies), the difference between very little things makes a very big difference, and you spend a lot of time fussing over details. I'll be very interested to see windows 8, given the scale of the departure of their user experience from anything before. You have to figure there has been a mountain of research done on everything from supporting proper english, to the precise layout and organization of their new mail client, and the core technical stuff about how to have an API for contacts (people I think they call it), how to build applications in whatever the new framework is around which the new UI exists and works etc.
And ya, it's hard to know what the new microsoft is even trying to do, let alone how many people they have doing it. At one point you were sure they were trying to take over the world. Now it's not even clear if they have a strategy other than reacting to a changing marketplace by changing for the sake of being in a changing marketplace. Good business is either guessing what the future will want, and being positioned for it, or making the future with into your vision of what it should be. Microsoft doesn't seem to have a clear sense of what it's trying to do, or why or how it's going to get it done.
Maybe it seems like they've been bulking up on lawyers because /. covers lawyers more than developers, and because developers that work for microsoft get almost no individual press for their work.
That's in part because MS doesn't let you sell your code on the side, and in part because you work for Microsoft, as part of a team, and externally the team gets credit. You may be clearly listed as part of a team, and credited for the team, but mostly we only see one point person for a team (who blogs or gives talks) and everyone else we don't know about. Which, given the range of products MS has would be prohibitively hard to keep track of from the outside anyway.
That's a pretty big assumption. And it assumes that the information he would have is valuable.
He's almost certainly not the first FTC person ever hired by microsoft, directly or indirectly. They probably already know the inner workings of the FTC. The question is whether or he specifically has knowledge of a business plans from google that could damage them somehow, but benefit MS.
More likely he merely adds more man hours to their access to the FTC and their ability to do the proper paper work in the proper way, for the proper people, and so they can support more paperwork.
Obviously he wasn't all that effective at his last job if they ignored his recommendations. But he does know what the review process is, and that's useful to have on staff so they can file the right paperwork.
Well 2 years is a two fold issue. First, as said elsewhere, you need to feed yourself for 2 years. Secondly, in 2 years hopefully most of the existing short term contracts have gone through the system, and it's all new stuff. You can't lobby for a bid already in progress sort of thing (where you were on the inside looking evaluating a bid, and then turn around and tell the company what to do to change it so it's successful).
Sort of by definition you will still know most of the people there and where they fit in the hierarchy, but that's not a huge problem.
That's mostly because those of us who make games are still writing for 360/PS3 and the CPU isn't the limiting factor so much on the PC side. I'm doing some research atm, and the new CPU's are even clock for clock 20% faster per generation if not more on the same task without any optimization. You just don't see the benefit because you aren't CPU limited. Which was my point about replacing with a lower tier. If you're happy with a q6600 general performance then the replacement line isn't i7-920 - i7 2600 it's to an i5 of some sort which are like 170 bucks compared to 300. I think a new i5 2300 will run about 2x as fast as a q6600, and it's 190 bucks retail including retail packaging and cooler, not counting mobo obviously. The q6600 was a 300 dollar part at retail, though I think I paid 400 for mine, but that might have been a USD-CDN conversion thing. You could probably to match performance of a q6600 with an i3 g8xx which are less than $100 retail, and only dual core (but are clock for clock much faster, and clocked higher). It might be a bit slower, hard to say without spending more than 5 minutes doing analysis. You could completely replace your ram, MOBO and CPU for probably 300 bucks at that point, and you'd be getting something brand new, which supports newer memory standards (see below). The way I price it out on newegg, and I'm not in the US so I'm not sure if it's lying to me (I'm using newegg us not newegg canada, which doesn't lie to me): A gigabyte 1155 board, 105 dollars, an i 13 g860 is 100 bucks and 8 gigs of ram is 75. You'll have a tough time convincing me that being able to replace the system for actually slightly better overall system performance for about 1/3rd the price of what you paid is somehow bad for you if a part fails. A new mobo, ram and CPU is actually less than you would have paid for the CPU.
The big thing is *memory*. Not ram though, nothing uses enough ram for 6/8/12 to make much difference. USB3 and SATAIII. The difference in user experience between an extremely fast raid 0's pair of drives on q6600 and an SATAIII SSD is enormous. It's one of those defining technologies that made smart phones viable. They're worth having in a computer that you actually use, but it's hard to convey how much better a full speed SATA III SSD is compared to say... anything else ever. Booting windows in 16 or 17 seconds (I take about 2 seconds to log in), and then being able to load say a game like WoW or SWTOR in a matter of seconds completely changes the experience. And they aren't hacking that back into a core 2 mobo.
Your statement on the LGA 775 is correct generally, it lasted a long time, and you bought right at the end of its lifecycle (as did I). But there were different product range sockets available even then (771's for example). The LGA 1366 - 1156 is the same thing. the 1156 isn't a replacement for the 1366 nor the reverse. They're for different markets and different CPU's. The LGA 1366 is superceded by the LGA 2011, but it stuck around for about 4 years as well, and the LGA 1155 supercedes the LGA 1156 in the mid range, and again, stuck around about 4 years. So yes, intel has released several sockets, but they're for different product ranges, they don't all supercede each other.
If you want to go from a q6600 on a 775 to a i7 920 on a 1366 to a i7 2600 on a LGA 2011 then sure you're talking about the last part to use a socket, to the first part to use a socket, each with about a 4 year life cycle. So you expect there to be some overlap. How far into the future LGA 2011 will last I don't know, afaik it will support ivy bridge though, which should take us through 2013 at least. But you'd also be sticking in the product range of being at the bottom of the high end range. You could quite reasonably drop down a tier at least from the q6600 and have the same CPU power.
To try and illustrate
High end
775 (2004-2008) *you are here* - 1366 (2008 - 2012) -> LGA 2011 (nov 2011 -> unknown, but some ivy bridge)
Mid range
771 -
This is in part because every university has its own programme arrangement, meaning the greatest textbook ever from some other university will only cover half of what is to be covered in your course, or it will cover something in a confusing way.
At highschool and lower this doesn't need to happen. Everyone, or nearly everyone should be on the same national curriculum, with the same textbooks. That will never happen in some countries though (like the US). You can even leave some of the day/week/month to 'teacher time' where they decide what to cover.
Unless it's a big class you don't get enough in royalties per book for that to be a particularly good plan. And if it is a big enough class it's important that the textbook fits with whatever the broader strategy of the university is, since you're training a huge portion of the student body with it.
re read what I said.
If one of your parts fails, either you had the foresight to buy replacements when they were being cleared out from inventory, or you buy a lower end version of a current product that will have the same performance for less money.
I'm running one q6600 and one q6700 right now. I have some spare ram lying around. But if either CPU, or a mobo dies my best bet is to probably just buy a low end i5 series or whatever is in the sweet spot for price drops rather than trying to repair the old stuff. If you can't recognize a product cycle, or what it does to prices, you probably can't recognize whether or not a q6600 is adequate or inadequate for your current usage, because you don't understand enough about what the products do and how that translates to a user experience.
This isn't some giant conspiracy against you trying to force you to upgrade 'prematurely'. This is a business that runs on1.5-2 year product cycles, and they still make hundreds of millions of parts within a product cycle. Those production lines shut down, new ones come online. If you only need to match the performance of your existing parts you buy lower end versions of future parts. It's not realistic to think that a core 2 quad, with what 550 million transistors, to an i7 920 with 700 and some, to a i7 -2600 (sandy bridge) with just shy of a billion are all going to just play nice on the same CPU sockets or necessarily with the same RAM. They're all ~300 dollar CPU's (which is why I chose them for comparison purposes).
Spot on.
There's a sweet spot where prices drop enough that you buy future spare parts, and then it becomes sufficiently antiquated it's impractical to buy spares unless it's an emergency. Most consumer electronics have a ~50% failure rate at 6 years anyway as capacitors give out.
This applies to any business, you just need to pay attention to the cycle for whatever business you're in. They probably stopped fabbing core 2 parts 4 years ago, and all of the periphery parts around the same time, so anything you're buying is that old.
Besides, they want you to replace the whole cpu/ram/mobo combo at once. They are sold separately for good reasons, but they all go together. If one part dies they're all designed for each other. And it's not like intel doesn't have new products at the similar and lower price points that will provide you the same performance, especially 5 years on. If your MOBO dies, your CPU is probably not far behind, and the ram not far behind that. If you don't need better performance just buy a lower end i5 (the q6600 was a mid range, buy a low end)
Or safari.
And even then, who are your search choices that won't track you, and produce results that aren't terrible.
True, but, like most students in this day and age, there's a risk you can take on borrowing against your future so that you'll have a better future. Granted, like most of our students in general, that doesn't work out too well.
And society did help you with learning a new skillset. Compulsory education to age 16/18 is a huge step in the right direction. It means the maximum retraining time from any profession to any other is about 10 years (4 years bachelors, 2 years masters 4 years PhD) and more realistically, the training time from one profession to another is more like 2 years since you don't want to change from say being a tradeschool carpenter to medicine when you're 40.
Oh and some places do offer job training as part of being on unemployment (countries/provinces/states/localities, and some companies will offer retraining or severance packages which can be used for such purposes). So your statement isn't universal.