Right, but the south koreans get quad core, some countries get different amounts of RAM etc.
If all you heard was the launch announcement of 'quad core*' and ignored the asterix of different countries getting different products you'd be confused by the whole thing.
As a developer by the way, this is a fucking nightmare. I work at a university, so we have, every year and every christmas people with phones from all over the world trying to use our mobile app. We need to test on the indian version, the korean version the chinese versions, the hong kong version, the taiwanese version, etc. etc. etc. And we need someone to keep track of what all the different versions are. I know the guys at big blue bubble in town who make mobile games have a big lab but I think they only care about europe and north america rather than everywhere else too.
So there just wasn't any other way to get this stimulative effect besides the Apollo program?
Spending is spending.
If you can find work for people to do, and then pay them for honestly doing it, and that work can in some way have some sort of positive benefit it will be better than just giving people things and hoping they stop being poor.
Now that means you have to recognize education as a form of productive work, it means you have to be willing to capitalize on under utilized labour, it means you have to have valid benchmarks for achievement. Hiring 1000 random people off the street and asking them to be teachers in classrooms with 100 students each might but 100 000 kids in school but it's unlikely to give them a useful education. It means when you have an under utilized labour market you have to be willing to tax or borrow the money to get something out of that labour and so on.
Governments are largely giant insurance systems - that's good, healthcare, police, army etc. are all basically forms of insurance. But they are also able to create markets for products and drive investment and innovation, that's good too (and in fact is in many cases a part of their spending as an insurance system, think police cars and fire trucks - innovation and demand for a new product to serve a useful roll, also, they aren't reinventing the wheel when they don't have to). Governments, as giant insurance systems, are actually a good place for risk. If any random company lost 40 billion dollars tomorrow (including Apple or Exxon) it would be a disaster for that company, big enough companies can survive of course, but a lot of investors would lose a lot of money and so on. Just about all of the western governments, including greece, could lose 40 billion dollars tomorrow and it would be inconvenient but not catastrophic (well, except that greece is trapped in the Euro but lets not get into that, they could survive an added 40 billion in debt, they'd just be stuck with 8-12% interest on it). It's also very hard for a government to actually lose 40 billion dollars in a rich country, it can very inefficiently use 40 billion dollars, but 40 billion dollars trying to build a tunnel to china and failing would still have put people to work for 40 billion dollars and driven up consumer demand for all the stuff they bought, so the government would have spent 40 billion, taxed back 15 or 20 billion, and benefited some from the spillover effect. And be left with a hole in the ground that goes no where. If the apollo programme had been a complete failure (all the rockets blew up for example), or if it turned out that for whatever reason you could never actually get any equipment that would be functional on the moon (people or otherwise) then at least all the people put to work trying would have had jobs, no small subset of the population would have borne the burden of eating the lost investment.
There are more complicated layers of course, about what to do in various states of employment, when just giving stuff away is the right course of action (emergencies for example), there's spending money to prevent disasters rather than recovering from them, which then looks like you've wasted money on a problem that never materialized. And sometimes you are only putting enough money into a problem to prevent it from getting worse.
I wouldn't expect most CIO's to be broadcasting to the world their costing estimates or disclosing the full extent of their IT infrastructure etc.
Besides, it really does matter on a company by company basis - what works for your company may not work for mine, so as much as generic studies like this can be a useful piece of the puzzle they aren't the final word for any specific outfit.
This might be because samsung is marketing a dual core and quad core phone under the same brand, despite the obvious difference in capability. That is, without a doubt, my biggest gripe with Samsung in the industry. A Galaxy S III should be the same everywhere, or failing that a Galaxy S III DC, or QC should be clearly the same everywhere. Having different versions of the same product is unnecessarily confusing.
Who says the game engine has to be PC only? They provide the basic API and the development tools (XNA), there's a good market for them selling some 3D engines and content authoring tools that would be very useful for both 360 and PC (and WP8). Where Epic (unreal engine) will only talk to you if you aren't concerned about price MS is big enough to have tiered support options and so on that most of the game engine companies can't afford, they have 'free and no support' or 'more money than you can afford' support.
Also, MS has been pretty good about doing what they do on the PC. They don't make game engines, but DX11 is a big testbed for future technology and they do actually support all this stuff pretty well from the development side. I grant it's second fiddle to the Xbox, but that doesn't mean we don't get what we need from them.
It seems unlikely this will change until the next generation of consoles arrives.
Sure. And that will likely be next christmas. You're right that games can't really be full blow DX 11 and still work on consoles. But even activision is axing DX9 support with Black ops 2.
Also, it seems like these days it's not so much the graphics limiting gaming performance as the rest of the code.
Depends very much on the game. The stuff I work on is mostly single CPU limited because we have one major algorithm that has to be done in order, can't be neatly pipelined etc, it just can't be run in parallel. We can spit everything else off onto its own thread, but we're still limited to the performance of one CPU.
Lots of games are limited by GPU performance, but that only applies if you have a low to mid range card. That's sort of the challenge with the whole PC market, lots of different configurations.
Piracy isn't killing PC gaming really, it's has killed certain parts of the business, anything that could migrate easily to the app store or free to play or to some sort of online service model (diablo) but there's still money to be made.
Oh, and HDD speed matters a lot to the experience, SSD's are great, but you can't count on people having them.
It's typically MOBO capacitors that fail on the MOBO itself, and the labour for a mobo install + ram + CPU suddenly means you may as well just buy a new machine.
PSU's and HDD's fail regularly too.
The PC industry very deliberately changes standards every couple of years, if your Core 2 Quad fails today it's actually a real pain to find a store that has a new CPU cooler in stock that will fit. You can order one on line, but then you're basically wasting a full day if not multiple days of employee time without a computer.
Why replace the existing computer in its entirety when for the cost of a replacement part you can keep the system running almost indefinitely these days?
On a server, because the diagnostic time equates to downtime, and trying to play whack a mole on a failed part on a live system isn't worth the loss.
On a regular PC, because after 5 years you can get a new computer that's 4 or 5x better performance for 500 bucks. At that point an hour of detection and correction time + the parts + lost employee time is made up for by reduced complaints due to losses.
Also, the rate of failure starts to go up over time. Once some parts fail it's usually worth dealing with it sooner rather than having to make a service call on other parts again in a week.
Where I work, the main DB servers are on a 3 year refresh, as are the customer facing computers. Everything else is 5-7 years.
The rule of thumb I give people is that computers have about a 50% failure rate for hardware at 6 years. If you really care about uptime you end up wanting to replace every 3 or so, BEFORE stuff starts to die. For computers that can be replaced without data loss and where 2 hours of downtime doesn't cause mass panic 5 years is fairly reasonable.
Desktop and laptop PCs have simply passed the point where even an entry-level model is sufficient for everyday home and business tasks like reading e-mail, web browsing, working on office documents and database applications, and playing audio/video files.
Sort of. Software hasn't caught up to taking advantage of hardware, in large part because you get glued to compatibility with 7 year old hardware and you can't take advantage of new hardware.
Though I fully accept that some problems just can't get much 'better' by throwing CPU cycles at it. Windows XP was the first 'good enough for everything' operating system from microsoft, and by about 2003-2004 you could do the vast majority of generic tasks reasonably well on affordable hardware. Quadrupling the speed of the computer won't make me type faster.
The problem of course isn't 99% of the applications on my computer. It's the 1% of applications that actually use the most power (or whatever that happens to be, maybe it's really and 80/20 problem). My car doesn't need to drive 100 Km/h 80% of the time, since I usually just drive too and from work, but I wouldn't buy a car that can't do highway speeds at all (unless it was a second car) sort of thing. If you want to play games, or if HTML 6 or some other new application comes along that really takes advantage of a faster computer you'll see people upgrade. Making windows kinda transparent and applications open 25% faster is improvement, it just, as you say, isn't all that useful.
I suspect the next big 'killer' app is going to look at lot like small business for home, which is going to be things like backup and networking tools for multiple computers, a 'home server'. Stuff geeks do now, but non geeks need. Those problems are more storage related than performance however. Games naturally remain the driver of performance for home users.
The open question is whether Intel will buy nVidia or whether they will try and continue in house development of GPU's, which haven't worked out particularly well so far.
Even future Windows Updates will break Valve's game engine for no good reason.
That doesn't even make sense. Much as I have been trying to persuade them, microsoft is not interested in making or selling game engines. There's a big difference between the directx API and a game engine, and MS isn't in the latter business at all, despite the fairly compelling case for there to be some serious software companies selling game engines at tiered pricepoints with tiered support.
I can now safely say that Valve isn't going to get the early access to Windows 9 they would need to stay competitive.
Again, that doesn't make any sense. Anyone who pays for MSDN access can get early access to windows 9, and, somewhat surprisingly, directx 8 code still works on windows 8, so there's not real reason to believe that source engine games are suddenly going to stop working any more than anything else could suddenly stop working.
Except for anti-malware/antivirus vendors. Those folk really have nowhere else to go
They're usually security companies, not just AV companies. There's still a market for intrusion detection, forensics, recovery etc. There is even a place for malware protection on linux and mac if they ever pick up market share, and there's a place for AV on mobile devices these days too.
People who want money worry about solving problems not placing blame.
But yes, google has probably done the metrics and figured it's not worth it to their bottom line, my point was those of us who are smaller outfits won't be able to jump on the 'ditch anything 3 years old' bandwagon quite as easily.
So far, non graphics stuff on gpus only has a few niche areas
uh... GPU accelerated video transcoding is the big mainstream app, along with physX on games. If the desktop actually runs through the 3D subsystem it can take advantage of the 3D hardware. I'm not convinced that will be faster than 2D for simple things, but when you start layering on effects like transparency, layering, morphing etc. it makes a big difference.
These tasks are only worth doing on mid to high end gpus, so doing them on these combo chips targeted at tablets and the like will not produce interesting results.
None of this statement is true. GPU accelerating the transparency effects around windows don't require a lot of power, and the benefit is noticeable, not all that important, but it does exist.
Afaik, There is nothing in windows 8 that makes integrated gpu elements necessary on desktop or tablet hardware, so what's the big deal? Directcompute is already available on x86 and has been for quite some time.
It's not necessary. It just speed things up. DirectCompute doesn't work worth shit on x86.
Even if these chips have exclusive functionality others lack,
These 'features' are really the difference between directx 11 and directx 9. They're not magical.
what features does windows 8 enable with them and how do these features break otherOS?
They don't. Well, UEFI might screw some of the linux distros, but that's about it. Windows 8 native gpu acceleration article for your perusal
This is stuff that you can do if you have the right hardware today, and the software takes advantage of it. The point is that windows 8 will take advantage of it for the entire desktop experience. Which makes it not all that relevant since you don't spend a lot of time just flipping through apps on the desktop, but it is there, and it does work.
To be clear, I'm not suggesting any of this is particularly useful, but it's real and it does work, it just doesn't matter all that much. If you really care about your computer feeling fast the difference between a 3 second load screen and 0.5 seconds makes a huge difference, even though it is not, in the grand scheme of things, much different. My computer with a SATA II SSD can go from off to applications running in 19 seconds (with login). With UEFI and windows 8 and a SATAIII SSD that number should be down to about 8-10 seconds, but it's not going to magically make me type faster.
Casual gamers simply aren't able or don't want to spend hours learning a game's mechanics, then spend more than an hour on a single level.
Serious gamers don't want the vast majority of their games very long either. That's why a 'big' game these days is 20 hours, and a typical one is 7-10. Looking at my list of games completed in the last 4 months (19 games completely) about half took less than 15 hours, give or take how you want to count Dawnguard.
Serious gamers play one or two games seriously (MMO, online FPS etc.) and then the rest of it they just chew through and move on.
Some of it is presentation and attitude, people think Torchlight, battlefield 3, max payne 3 etc are 'serious' games, but they aren't, you can play them in stints of 40 minutes or so at a time, and then be done all of them in under 15 hours each. Some of it is also self delusion, where a 'casual' gamer who plays farmville 3 hours a day, everyday, doesn't want to call themselves a 'gamer' because those are nerds who spend 3 hours a day playing games. And believe me, we have a lot of data on the so called 'casual' gamer, and they spend a lot of time playing games.
No, they're porting to Linux as a hedge against the success of Windows 8.
I sort of covered that in my 3rd paragraph. It's not windows 8 they're worried about, that's almost certainly doomed at this point. Windows 9 on the other hand....
But even then, they need to look at what value they can add as a store, and what they do if the whole Windows ecosystem implodes upon itself from microsoft doing things badly.
Ultimately this is a problem all of the big software vendors need to think about. Microsoft going after their market could be a disaster (think office to IBM lotus notes or Corel word perfect, internet explorer to netscape etc.), and microsoft's windows business collapsing, which, if we want to look at 5 year time horizons starts to look like a serious possibility - look at what happened to nokia in the last 5 years - both pose serious challenges. When you just make games you can always just make games for the next platform, and you're somewhat compartmentalized to what happens on 3 year cycles,and can start afresh after that, or at least use your art assets on a different platform, but when you have a big long term ecosystem you have to worry about just what is going to happen at both Apple and Microsoft.
So Valve needs a new platform for their game engines, games and game store to run on because Windows 8 is not going to work.
Uh... valve has another platform for their games and game engines, (the 360, and a half assed job of PS3) if they were just a game company, like the rest of us, they wouldn't really care. There will still be a market for games 5 years from now. That might be on the WiiU, the xbox3, PS4, an android console etc. The market for their games isn't going away, but yes, Steam runs into a lot of challenges from windows 8. If the windows store is hugely successful they're screwed, and if Windows 8 spells the end of the Windows 90% marketshare they're screwed too.
It does do 'gpu accelerated' desktop though, and that requires a GPU that supports the functions you're calling.
Again, it's not like these things can't exist now, it's just hard to convey that.
The on chip CPU thing is a combination of power and definitely getting a GPU part that isn't soul crushingly horrible, and will actually support the DX11 directcompute stuff that windows 8 uses to speed things up.
there is nothing intrinsic about windows 8 that requires GPU-like, crazy-scale SIMD instructions within the cpu.
Depends what you mean by intrinsic. Windows 8 can use GPU acceleration on just about everything (which is odd terminology, aren't most 3d games by definition gpu accelerated? Well sure, but that technology hasn't really filtered into the windowing system in Windows and things like MS office).
Unfortunately the term 'gpu accelerated' is a bit vague, since anything on your screen is through the GPU it's inherently unclear. The way I would describe it is a matter of executing the code on the GPU rather than using any sort of fixed function, and the trick is making sure your GPU can support whatever instructions you're going to support.
You need to come to grips with the simple fact that the purchasing power of fiat money decreases with time.
Which is irrelevant to this discussion. We're not talking about 1990 dollars versus 2012 dollars. We're talking about 2012 dollars versus 2012 dollars.
The discussion isn't 'oh games were so much cheaper in the past why are they so expensive now' because aliquis wasn't accounting for inflation. This is a ' if I buy from a retailer you get 20 dollars, so why don't I just buy directly from you for 20 dollars and skip the 40 dollars going to publishers and retailers'.
You're confusing a monetary issue with a business model one. Lots of companies (notably paradox and Stardock) sell games, or expansion packs for games, for less money online than they did with boxed copies. The upcomming HOI3 expansion is 20 bucks on gamersgate, if they were doing a big retail push it would be 30 everywhere because of the added distribution costs from using the retail chain.
I grant, there is some nominal price rigidity around games, and people don't want to lower the price knowing that customers will pay 50-60 dollars for a new game, but valve with their steam sales has had some success (and some fucking of developers) with selling even new games for a lot less money.
First: Reduced performance for significantly lower power consumption. There's no top range Ivy brige CPU right now (extreme edition) because the Sandy bridge are fast enough for that problem.
The second issue is going to be less CPU and more motherboard, UEFI specifically. While supported on 64 bit vista and Windows server 2008 and later you need a legacy mode for windows XP, it's 64 bit only and a few other inconveniences for windows 7 and up.
Third: The GPU (APU in AMD speak) is in the same IC as the CPU. You see that in laptops today, but it's all part of having a GPU accelerated operating system - 'designed for' as in it will definitely do that, you can do that on some systems today, but you'll definitely be able to do it on the new stuff.
Obviously it's marketing waffle - but there's nothing wrong with trying to sell products to consumers, talking about support for full SATA III SSD support, UEFI and GPU acceleration isn't really a good public branding strategy.
I would consider 'set top box' and 'console' the same thing for this discussion.
The proper set top boxes from the cable companies is a whole other problem. In that they'd be going head to head with the DVR and PVR market. But going head to head with cisco on that is a really really poor plan. Cisco just spent what, 5 billion dollars on NDS to get control of basically the whole set top box market, Valve is big, but it has no relationships with the cable companies, and I can't see them particularly wanting to do business with valve outside the US.
Now true, the NDS set top boxes all run java talking to linux servers, but they don't have the hardware for gaming on the box, and they don't really seem inclined to go that route, or if they do, they aren't telling me about it.
Slashdot Mirror
Right, but the south koreans get quad core, some countries get different amounts of RAM etc.
If all you heard was the launch announcement of 'quad core*' and ignored the asterix of different countries getting different products you'd be confused by the whole thing.
As a developer by the way, this is a fucking nightmare. I work at a university, so we have, every year and every christmas people with phones from all over the world trying to use our mobile app. We need to test on the indian version, the korean version the chinese versions, the hong kong version, the taiwanese version, etc. etc. etc. And we need someone to keep track of what all the different versions are. I know the guys at big blue bubble in town who make mobile games have a big lab but I think they only care about europe and north america rather than everywhere else too.
So there just wasn't any other way to get this stimulative effect besides the Apollo program?
Spending is spending.
If you can find work for people to do, and then pay them for honestly doing it, and that work can in some way have some sort of positive benefit it will be better than just giving people things and hoping they stop being poor.
Now that means you have to recognize education as a form of productive work, it means you have to be willing to capitalize on under utilized labour, it means you have to have valid benchmarks for achievement. Hiring 1000 random people off the street and asking them to be teachers in classrooms with 100 students each might but 100 000 kids in school but it's unlikely to give them a useful education. It means when you have an under utilized labour market you have to be willing to tax or borrow the money to get something out of that labour and so on.
Governments are largely giant insurance systems - that's good, healthcare, police, army etc. are all basically forms of insurance. But they are also able to create markets for products and drive investment and innovation, that's good too (and in fact is in many cases a part of their spending as an insurance system, think police cars and fire trucks - innovation and demand for a new product to serve a useful roll, also, they aren't reinventing the wheel when they don't have to). Governments, as giant insurance systems, are actually a good place for risk. If any random company lost 40 billion dollars tomorrow (including Apple or Exxon) it would be a disaster for that company, big enough companies can survive of course, but a lot of investors would lose a lot of money and so on. Just about all of the western governments, including greece, could lose 40 billion dollars tomorrow and it would be inconvenient but not catastrophic (well, except that greece is trapped in the Euro but lets not get into that, they could survive an added 40 billion in debt, they'd just be stuck with 8-12% interest on it). It's also very hard for a government to actually lose 40 billion dollars in a rich country, it can very inefficiently use 40 billion dollars, but 40 billion dollars trying to build a tunnel to china and failing would still have put people to work for 40 billion dollars and driven up consumer demand for all the stuff they bought, so the government would have spent 40 billion, taxed back 15 or 20 billion, and benefited some from the spillover effect. And be left with a hole in the ground that goes no where. If the apollo programme had been a complete failure (all the rockets blew up for example), or if it turned out that for whatever reason you could never actually get any equipment that would be functional on the moon (people or otherwise) then at least all the people put to work trying would have had jobs, no small subset of the population would have borne the burden of eating the lost investment.
There are more complicated layers of course, about what to do in various states of employment, when just giving stuff away is the right course of action (emergencies for example), there's spending money to prevent disasters rather than recovering from them, which then looks like you've wasted money on a problem that never materialized. And sometimes you are only putting enough money into a problem to prevent it from getting worse.
And are probably not published at all.
I wouldn't expect most CIO's to be broadcasting to the world their costing estimates or disclosing the full extent of their IT infrastructure etc.
Besides, it really does matter on a company by company basis - what works for your company may not work for mine, so as much as generic studies like this can be a useful piece of the puzzle they aren't the final word for any specific outfit.
This might be because samsung is marketing a dual core and quad core phone under the same brand, despite the obvious difference in capability. That is, without a doubt, my biggest gripe with Samsung in the industry. A Galaxy S III should be the same everywhere, or failing that a Galaxy S III DC, or QC should be clearly the same everywhere. Having different versions of the same product is unnecessarily confusing.
Smartphones are so fast nowadays that whatever you buy is good enough to do 90% of the things people want a smartphone to do.
That's because smart phones are basically 5 year old PC's with small screens.
But for some people the new network (LTE) will be radically different, especially if the 3G in your area has serious congestion issues.
Linux is still a platform that's supported and will be for years to come.
So is windows. Linux has version numbers too you know. The L4D2 thing was specifically for ubuntu 12.04 for example.
Who says the game engine has to be PC only? They provide the basic API and the development tools (XNA), there's a good market for them selling some 3D engines and content authoring tools that would be very useful for both 360 and PC (and WP8). Where Epic (unreal engine) will only talk to you if you aren't concerned about price MS is big enough to have tiered support options and so on that most of the game engine companies can't afford, they have 'free and no support' or 'more money than you can afford' support.
Also, MS has been pretty good about doing what they do on the PC. They don't make game engines, but DX11 is a big testbed for future technology and they do actually support all this stuff pretty well from the development side. I grant it's second fiddle to the Xbox, but that doesn't mean we don't get what we need from them.
It seems unlikely this will change until the next generation of consoles arrives.
Sure. And that will likely be next christmas. You're right that games can't really be full blow DX 11 and still work on consoles. But even activision is axing DX9 support with Black ops 2.
Also, it seems like these days it's not so much the graphics limiting gaming performance as the rest of the code.
Depends very much on the game. The stuff I work on is mostly single CPU limited because we have one major algorithm that has to be done in order, can't be neatly pipelined etc, it just can't be run in parallel. We can spit everything else off onto its own thread, but we're still limited to the performance of one CPU.
Lots of games are limited by GPU performance, but that only applies if you have a low to mid range card. That's sort of the challenge with the whole PC market, lots of different configurations.
Piracy isn't killing PC gaming really, it's has killed certain parts of the business, anything that could migrate easily to the app store or free to play or to some sort of online service model (diablo) but there's still money to be made.
Oh, and HDD speed matters a lot to the experience, SSD's are great, but you can't count on people having them.
It's typically MOBO capacitors that fail on the MOBO itself, and the labour for a mobo install + ram + CPU suddenly means you may as well just buy a new machine.
PSU's and HDD's fail regularly too.
The PC industry very deliberately changes standards every couple of years, if your Core 2 Quad fails today it's actually a real pain to find a store that has a new CPU cooler in stock that will fit. You can order one on line, but then you're basically wasting a full day if not multiple days of employee time without a computer.
Why replace the existing computer in its entirety when for the cost of a replacement part you can keep the system running almost indefinitely these days?
On a server, because the diagnostic time equates to downtime, and trying to play whack a mole on a failed part on a live system isn't worth the loss.
On a regular PC, because after 5 years you can get a new computer that's 4 or 5x better performance for 500 bucks. At that point an hour of detection and correction time + the parts + lost employee time is made up for by reduced complaints due to losses.
Also, the rate of failure starts to go up over time. Once some parts fail it's usually worth dealing with it sooner rather than having to make a service call on other parts again in a week.
Where I work, the main DB servers are on a 3 year refresh, as are the customer facing computers. Everything else is 5-7 years.
The rule of thumb I give people is that computers have about a 50% failure rate for hardware at 6 years. If you really care about uptime you end up wanting to replace every 3 or so, BEFORE stuff starts to die. For computers that can be replaced without data loss and where 2 hours of downtime doesn't cause mass panic 5 years is fairly reasonable.
Desktop and laptop PCs have simply passed the point where even an entry-level model is sufficient for everyday home and business tasks like reading e-mail, web browsing, working on office documents and database applications, and playing audio/video files.
Sort of. Software hasn't caught up to taking advantage of hardware, in large part because you get glued to compatibility with 7 year old hardware and you can't take advantage of new hardware.
Though I fully accept that some problems just can't get much 'better' by throwing CPU cycles at it. Windows XP was the first 'good enough for everything' operating system from microsoft, and by about 2003-2004 you could do the vast majority of generic tasks reasonably well on affordable hardware. Quadrupling the speed of the computer won't make me type faster.
The problem of course isn't 99% of the applications on my computer. It's the 1% of applications that actually use the most power (or whatever that happens to be, maybe it's really and 80/20 problem). My car doesn't need to drive 100 Km/h 80% of the time, since I usually just drive too and from work, but I wouldn't buy a car that can't do highway speeds at all (unless it was a second car) sort of thing. If you want to play games, or if HTML 6 or some other new application comes along that really takes advantage of a faster computer you'll see people upgrade. Making windows kinda transparent and applications open 25% faster is improvement, it just, as you say, isn't all that useful.
I suspect the next big 'killer' app is going to look at lot like small business for home, which is going to be things like backup and networking tools for multiple computers, a 'home server'. Stuff geeks do now, but non geeks need. Those problems are more storage related than performance however. Games naturally remain the driver of performance for home users.
So did everyone else.
The open question is whether Intel will buy nVidia or whether they will try and continue in house development of GPU's, which haven't worked out particularly well so far.
Pretty much. The 'future is fusion' is both a serious technical concept and a marketing phrase.
Eventually the GPU will just be another part of the CPU, like happened to math co processors.
Last figures I saw had the X360 having a sell through of 8 games per console,
Those are more like per year figures, with PS3 and 360 almost dead even now.
Even future Windows Updates will break Valve's game engine for no good reason.
That doesn't even make sense. Much as I have been trying to persuade them, microsoft is not interested in making or selling game engines. There's a big difference between the directx API and a game engine, and MS isn't in the latter business at all, despite the fairly compelling case for there to be some serious software companies selling game engines at tiered pricepoints with tiered support.
I can now safely say that Valve isn't going to get the early access to Windows 9 they would need to stay competitive.
Again, that doesn't make any sense. Anyone who pays for MSDN access can get early access to windows 9, and, somewhat surprisingly, directx 8 code still works on windows 8, so there's not real reason to believe that source engine games are suddenly going to stop working any more than anything else could suddenly stop working.
Except for anti-malware/antivirus vendors. Those folk really have nowhere else to go
They're usually security companies, not just AV companies. There's still a market for intrusion detection, forensics, recovery etc. There is even a place for malware protection on linux and mac if they ever pick up market share, and there's a place for AV on mobile devices these days too.
You're talking a lot of nonsense unfortunately.
Fault decides who should take corrective action.
No, it decides who idealists complain about.
People who want money worry about solving problems not placing blame.
But yes, google has probably done the metrics and figured it's not worth it to their bottom line, my point was those of us who are smaller outfits won't be able to jump on the 'ditch anything 3 years old' bandwagon quite as easily.
So far, non graphics stuff on gpus only has a few niche areas
uh... GPU accelerated video transcoding is the big mainstream app, along with physX on games. If the desktop actually runs through the 3D subsystem it can take advantage of the 3D hardware. I'm not convinced that will be faster than 2D for simple things, but when you start layering on effects like transparency, layering, morphing etc. it makes a big difference.
These tasks are only worth doing on mid to high end gpus, so doing them on these combo chips targeted at tablets and the like will not produce interesting results.
None of this statement is true. GPU accelerating the transparency effects around windows don't require a lot of power, and the benefit is noticeable, not all that important, but it does exist.
Afaik, There is nothing in windows 8 that makes integrated gpu elements necessary on desktop or tablet hardware, so what's the big deal? Directcompute is already available on x86 and has been for quite some time.
It's not necessary. It just speed things up. DirectCompute doesn't work worth shit on x86.
Even if these chips have exclusive functionality others lack,
These 'features' are really the difference between directx 11 and directx 9. They're not magical.
what features does windows 8 enable with them and how do these features break otherOS?
They don't. Well, UEFI might screw some of the linux distros, but that's about it. Windows 8 native gpu acceleration article for your perusal
http://arstechnica.com/information-technology/2012/07/windows-8-gpu-acceleration-good-news-for-metro/
This is stuff that you can do if you have the right hardware today, and the software takes advantage of it. The point is that windows 8 will take advantage of it for the entire desktop experience. Which makes it not all that relevant since you don't spend a lot of time just flipping through apps on the desktop, but it is there, and it does work.
To be clear, I'm not suggesting any of this is particularly useful, but it's real and it does work, it just doesn't matter all that much. If you really care about your computer feeling fast the difference between a 3 second load screen and 0.5 seconds makes a huge difference, even though it is not, in the grand scheme of things, much different. My computer with a SATA II SSD can go from off to applications running in 19 seconds (with login). With UEFI and windows 8 and a SATAIII SSD that number should be down to about 8-10 seconds, but it's not going to magically make me type faster.
Casual gamers simply aren't able or don't want to spend hours learning a game's mechanics, then spend more than an hour on a single level.
Serious gamers don't want the vast majority of their games very long either. That's why a 'big' game these days is 20 hours, and a typical one is 7-10. Looking at my list of games completed in the last 4 months (19 games completely) about half took less than 15 hours, give or take how you want to count Dawnguard.
Serious gamers play one or two games seriously (MMO, online FPS etc.) and then the rest of it they just chew through and move on.
Some of it is presentation and attitude, people think Torchlight, battlefield 3, max payne 3 etc are 'serious' games, but they aren't, you can play them in stints of 40 minutes or so at a time, and then be done all of them in under 15 hours each. Some of it is also self delusion, where a 'casual' gamer who plays farmville 3 hours a day, everyday, doesn't want to call themselves a 'gamer' because those are nerds who spend 3 hours a day playing games. And believe me, we have a lot of data on the so called 'casual' gamer, and they spend a lot of time playing games.
No, they're porting to Linux as a hedge against the success of Windows 8.
I sort of covered that in my 3rd paragraph. It's not windows 8 they're worried about, that's almost certainly doomed at this point. Windows 9 on the other hand....
But even then, they need to look at what value they can add as a store, and what they do if the whole Windows ecosystem implodes upon itself from microsoft doing things badly.
Ultimately this is a problem all of the big software vendors need to think about. Microsoft going after their market could be a disaster (think office to IBM lotus notes or Corel word perfect, internet explorer to netscape etc.), and microsoft's windows business collapsing, which, if we want to look at 5 year time horizons starts to look like a serious possibility - look at what happened to nokia in the last 5 years - both pose serious challenges. When you just make games you can always just make games for the next platform, and you're somewhat compartmentalized to what happens on 3 year cycles,and can start afresh after that, or at least use your art assets on a different platform, but when you have a big long term ecosystem you have to worry about just what is going to happen at both Apple and Microsoft.
So Valve needs a new platform for their game engines, games and game store to run on because Windows 8 is not going to work.
Uh... valve has another platform for their games and game engines, (the 360, and a half assed job of PS3) if they were just a game company, like the rest of us, they wouldn't really care. There will still be a market for games 5 years from now. That might be on the WiiU, the xbox3, PS4, an android console etc. The market for their games isn't going away, but yes, Steam runs into a lot of challenges from windows 8. If the windows store is hugely successful they're screwed, and if Windows 8 spells the end of the Windows 90% marketshare they're screwed too.
It does do 'gpu accelerated' desktop though, and that requires a GPU that supports the functions you're calling.
Again, it's not like these things can't exist now, it's just hard to convey that.
The on chip CPU thing is a combination of power and definitely getting a GPU part that isn't soul crushingly horrible, and will actually support the DX11 directcompute stuff that windows 8 uses to speed things up.
there is nothing intrinsic about windows 8 that requires GPU-like, crazy-scale SIMD instructions within the cpu.
Depends what you mean by intrinsic. Windows 8 can use GPU acceleration on just about everything (which is odd terminology, aren't most 3d games by definition gpu accelerated? Well sure, but that technology hasn't really filtered into the windowing system in Windows and things like MS office).
Unfortunately the term 'gpu accelerated' is a bit vague, since anything on your screen is through the GPU it's inherently unclear. The way I would describe it is a matter of executing the code on the GPU rather than using any sort of fixed function, and the trick is making sure your GPU can support whatever instructions you're going to support.
You need to come to grips with the simple fact that the purchasing power of fiat money decreases with time.
Which is irrelevant to this discussion. We're not talking about 1990 dollars versus 2012 dollars. We're talking about 2012 dollars versus 2012 dollars.
The discussion isn't 'oh games were so much cheaper in the past why are they so expensive now' because aliquis wasn't accounting for inflation. This is a ' if I buy from a retailer you get 20 dollars, so why don't I just buy directly from you for 20 dollars and skip the 40 dollars going to publishers and retailers'.
You're confusing a monetary issue with a business model one. Lots of companies (notably paradox and Stardock) sell games, or expansion packs for games, for less money online than they did with boxed copies. The upcomming HOI3 expansion is 20 bucks on gamersgate, if they were doing a big retail push it would be 30 everywhere because of the added distribution costs from using the retail chain.
I grant, there is some nominal price rigidity around games, and people don't want to lower the price knowing that customers will pay 50-60 dollars for a new game, but valve with their steam sales has had some success (and some fucking of developers) with selling even new games for a lot less money.
Three things.
First: Reduced performance for significantly lower power consumption. There's no top range Ivy brige CPU right now (extreme edition) because the Sandy bridge are fast enough for that problem.
The second issue is going to be less CPU and more motherboard, UEFI specifically. While supported on 64 bit vista and Windows server 2008 and later you need a legacy mode for windows XP, it's 64 bit only and a few other inconveniences for windows 7 and up.
Third: The GPU (APU in AMD speak) is in the same IC as the CPU. You see that in laptops today, but it's all part of having a GPU accelerated operating system - 'designed for' as in it will definitely do that, you can do that on some systems today, but you'll definitely be able to do it on the new stuff.
Obviously it's marketing waffle - but there's nothing wrong with trying to sell products to consumers, talking about support for full SATA III SSD support, UEFI and GPU acceleration isn't really a good public branding strategy.
I diverged from my own one line summary, that didn't help readability or convey my point particularly well.
I would consider 'set top box' and 'console' the same thing for this discussion.
The proper set top boxes from the cable companies is a whole other problem. In that they'd be going head to head with the DVR and PVR market. But going head to head with cisco on that is a really really poor plan. Cisco just spent what, 5 billion dollars on NDS to get control of basically the whole set top box market, Valve is big, but it has no relationships with the cable companies, and I can't see them particularly wanting to do business with valve outside the US.
Now true, the NDS set top boxes all run java talking to linux servers, but they don't have the hardware for gaming on the box, and they don't really seem inclined to go that route, or if they do, they aren't telling me about it.