There is nothing stopping OSS from shipping 64bit builds, there are already 64bit builds of OOo and Firefox for platforms other than windows...
It's just that the benefits of shipping a 32bit binary outweigh the benefits of 64 in these cases... If support for 32bit was dropped, then recompiling as 64 would not be terribly difficult (and as you pointed out, third parties are already doing so).
Currently if you were to ship a 64bit firefox on windows, you would have to ensure users download the correct version, and you would lose access to most plugins such as flash.
You can't introduce a new architecture, no matter how superior it is... The prevalence of proprietary binary-only software will (and has already on several occasions) ensure that any new architecture will fail miserably if it's not compatible with x86 (itanium, power, alpha) or take off extremely slowly if it is (amd64).
I have IA64 and Alpha machines that make great linux boxes, and i can recompile virtually anything that runs on x86, but what i don't have is the proprietary crap like flash. Now imagine trying to use these machines where 90% of the software you want to run only comes as binaries... You would be faced with things that don't run at all, or run extremely slowly through emulation.
This is what happened to IA64, native code ran very quickly, but emulating x86 code was far slower than using real x86 hardware, if 90% of your apps are running in that mode then just using x86 hardware will be faster.
There is no incentive for commercial software producers to port their apps to IA64, because there are too few potential customers. There is no incentive for customers who want commercial software to use IA64 because there is little or no such software available.
So long as a large proportion of the software people want to run is commercially produced and shipped without source code, new architectures will fail.
Not all games work properly, many use drm schemes that require kernel level drivers, and some of the older games haven't had these kernel drivers ported to 64bit.
They have already kludged PAE support in... All they need to do, is remove the arbitrary licensing restriction that stops people from using it. Did you not read the article? It also states that prior to xp/sp2 PAE support was available by default and it was possible to use more than 4gb in a 32bit system.
32bit machines were around for years before MS bothered supporting them...
Their first 32bit OS came out in 1993 or 1994, and wasn't targeted at consumers. The first 64bit cpus and associated software were already out by then.
It took MS another 10 years before they came out with a 64bit os... I was quite surprised to see there weren't any 128-bit cpus available yet.
A lot of places are stuck with legacy applications and/or hardware that only works on 32bit versions of windows... I know several places that buy machines with 4GB, and install 32bit xp resulting in 1gb of memory that is completely unusable, and they have several reasons for wanting to do that ranging from old apps, to a desire not to support multiple versions across a large network.
There is even lots of proprietary software that is still being supported, and yet it doesn't support 64bit... A lot of commercial vpn clients for example.
The market for people who have 64bit systems which can hold far more than 4gb of ram, but for whatever reason want or need to run 32bit versions of windows is much larger...
A lot of systems have broken ACPI implementations... Broken in that they don't comply with Intel's documented spec for ACPI, they follow microsoft's specs instead.
Depending on the hardware, linux can have better battery life than windows. My experiences have varied massively depending on the machine.
OSX achieves good battery life because the hardware and software are designed together, from the brief testing i've done linux on my macbook pro has better battery life than windows but not as good as osx.
Just because something is google branded, doesn't mean the carriers won't supply it to their customers. There's no reason Google couldn't get their own branded phone manufactured and shipped to multiple carriers, virtually every carrier around offers blackberry, nokia, sony-ericson branded phones already.
Browser history is only kept for a finite period of time, so it has no reason to slow down at all beyond that point.
On a unix based system, the configuration for each app is kept separately and only loaded if required, on windows it's all bundled in to the registry so it needs to be parsed wether you're using that application or not. So no, simply having more applications and associated their configuration on your drive should NOT slow the system down.
Also, windows apps, especially games, like to load kernel drivers for drm and auto update processes etc... I have a windows machine that is used for nothing but games, and on startup it loads several drm drivers for the various games which have been installed. If you remove the drm crap, the associated game stops working, if you remove the auto updater then your apps will not get updated and could suffer from security holes, or you might find yourself unable to play online because all the other players have been updated to the latest version.
The fact that you consider it normal for windows to get gradually slower under typical usage is just absurd, why should people have to watch out for how many things they install incase the system will grind to a halt?
Also, there is the problem with uninstallers not removing stuff properly and leaving detritus littering the system... It comes from relying on the application vendor to write an uninstaller, a proper package manager that kept track of these things would work much better.
If the modem presents itself to the system in a standard way then sure, you could probably even install android on it and use it like a (very large) phone.
I doubt you could get 12 hours runtime, especially if you have an nvidia gpu that's capable of decoding hd video.. Not without a stupidly large battery anyway.
You could probably manage that with an ARM based laptop and a hardware video decoder, but it's not really worth putting such specialized hardware into a laptop like that.
Security is of little concern in this context... The system will only be running while the game is being played, after that it gets turned off and everything is gone. There would be no network services running except the game itself (assuming it had some kind of server mode) - having anything extra running would just waste resources that could be used for playing. Local security wouldn't be an issue, since it would inherently be a single user system.
Linux users aren't buying games because there aren't many available... There are plenty of people, myself included who use Linux for day to day tasks, and windows for games, or use windows for everything because they primarily play games, but would consider linux otherwise.
I find windows absolutely terrible, but often a necessary evil if you want to play games on your computer... Even doing nothing other than gaming, the system gradually breaks down... Once you have a bunch of games installed, along with a bunch of drm drivers shoehorned into the kernel at various points things get really messy. And the hassle of having to keep physical media around because games want you to physically insert the media to play... And NoCD cracks aren't practical, some prevent you playing online, most prevent you from updating (and modern games are poorly tested, extremely buggy and require updating)..
How about producing gaming livecds? A bootable linux livecd with ati and nvidia drivers preinstalled, configured to go straight into a particular game and nothing else... You could bring some of the convenience of console gaming, while not bogging down someone's live install.
And what about people who downloaded the linux binary to play the demo version? How about people who acquired the linux binary from mirror sites or third party packages?
Depends on your processor really... On Sparc, MIPS or PPC, having a 64bit browser will just consume more memory (tho it can potentially use more) and possibly run slower... On x86-64 a 64bit browser may run faster due to having access to a larger pool of registers, as well as being compiled for a more modern architecture which is closer to the cpu being used (most x86 software tends to be compiled for a 386 as the lowest common denominator).
Slashdot Mirror
Assuming your 64bit system had 1.5 times as much ram as the 32bit, would those scripts run faster than on the 32bit boxes?
Not necessarily...
I have a Compaq ppro box that can take 8gb, not a 64bit cpu...
I also have a dual xeon (older 32bit xeon) that can support 8gb..
There is nothing stopping OSS from shipping 64bit builds, there are already 64bit builds of OOo and Firefox for platforms other than windows...
It's just that the benefits of shipping a 32bit binary outweigh the benefits of 64 in these cases... If support for 32bit was dropped, then recompiling as 64 would not be terribly difficult (and as you pointed out, third parties are already doing so).
Currently if you were to ship a 64bit firefox on windows, you would have to ensure users download the correct version, and you would lose access to most plugins such as flash.
You can't introduce a new architecture, no matter how superior it is... The prevalence of proprietary binary-only software will (and has already on several occasions) ensure that any new architecture will fail miserably if it's not compatible with x86 (itanium, power, alpha) or take off extremely slowly if it is (amd64).
I have IA64 and Alpha machines that make great linux boxes, and i can recompile virtually anything that runs on x86, but what i don't have is the proprietary crap like flash. Now imagine trying to use these machines where 90% of the software you want to run only comes as binaries... You would be faced with things that don't run at all, or run extremely slowly through emulation.
This is what happened to IA64, native code ran very quickly, but emulating x86 code was far slower than using real x86 hardware, if 90% of your apps are running in that mode then just using x86 hardware will be faster.
There is no incentive for commercial software producers to port their apps to IA64, because there are too few potential customers. There is no incentive for customers who want commercial software to use IA64 because there is little or no such software available.
So long as a large proportion of the software people want to run is commercially produced and shipped without source code, new architectures will fail.
Not all games work properly, many use drm schemes that require kernel level drivers, and some of the older games haven't had these kernel drivers ported to 64bit.
They have already kludged PAE support in...
All they need to do, is remove the arbitrary licensing restriction that stops people from using it. Did you not read the article? It also states that prior to xp/sp2 PAE support was available by default and it was possible to use more than 4gb in a 32bit system.
32bit machines were around for years before MS bothered supporting them...
Their first 32bit OS came out in 1993 or 1994, and wasn't targeted at consumers. The first 64bit cpus and associated software were already out by then.
It took MS another 10 years before they came out with a 64bit os... I was quite surprised to see there weren't any 128-bit cpus available yet.
Some of us have had 64bit Linux/Unix machines since the 90s..
Why then, did XP support more ram before SP2?
A lot of places are stuck with legacy applications and/or hardware that only works on 32bit versions of windows...
I know several places that buy machines with 4GB, and install 32bit xp resulting in 1gb of memory that is completely unusable, and they have several reasons for wanting to do that ranging from old apps, to a desire not to support multiple versions across a large network.
There is even lots of proprietary software that is still being supported, and yet it doesn't support 64bit... A lot of commercial vpn clients for example.
The market for people who have 64bit systems which can hold far more than 4gb of ram, but for whatever reason want or need to run 32bit versions of windows is much larger...
A lot of systems have broken ACPI implementations... Broken in that they don't comply with Intel's documented spec for ACPI, they follow microsoft's specs instead.
Depending on the hardware, linux can have better battery life than windows. My experiences have varied massively depending on the machine.
OSX achieves good battery life because the hardware and software are designed together, from the brief testing i've done linux on my macbook pro has better battery life than windows but not as good as osx.
Not to mention the hassle of having to carry so many physical dvds around, a flash drive or hard disk is massively more convenient.
Sometimes it's better to have a small set of features that work well, than a large set of buggy and broken features...
Just because something is google branded, doesn't mean the carriers won't supply it to their customers.
There's no reason Google couldn't get their own branded phone manufactured and shipped to multiple carriers, virtually every carrier around offers blackberry, nokia, sony-ericson branded phones already.
The source for the kernel and some parts of the OS are open, so there's no reason third parties couldn't port those parts to PPC...
If it's truly a shipping and handling fee, why can't you walk into an apple store and get it free or cheaper (since there will be no shipping to do)?
Browser history is only kept for a finite period of time, so it has no reason to slow down at all beyond that point.
On a unix based system, the configuration for each app is kept separately and only loaded if required, on windows it's all bundled in to the registry so it needs to be parsed wether you're using that application or not.
So no, simply having more applications and associated their configuration on your drive should NOT slow the system down.
Also, windows apps, especially games, like to load kernel drivers for drm and auto update processes etc... I have a windows machine that is used for nothing but games, and on startup it loads several drm drivers for the various games which have been installed. If you remove the drm crap, the associated game stops working, if you remove the auto updater then your apps will not get updated and could suffer from security holes, or you might find yourself unable to play online because all the other players have been updated to the latest version.
The fact that you consider it normal for windows to get gradually slower under typical usage is just absurd, why should people have to watch out for how many things they install incase the system will grind to a halt?
Also, there is the problem with uninstallers not removing stuff properly and leaving detritus littering the system... It comes from relying on the application vendor to write an uninstaller, a proper package manager that kept track of these things would work much better.
If the modem presents itself to the system in a standard way then sure, you could probably even install android on it and use it like a (very large) phone.
I doubt you could get 12 hours runtime, especially if you have an nvidia gpu that's capable of decoding hd video.. Not without a stupidly large battery anyway.
You could probably manage that with an ARM based laptop and a hardware video decoder, but it's not really worth putting such specialized hardware into a laptop like that.
Security is of little concern in this context...
The system will only be running while the game is being played, after that it gets turned off and everything is gone.
There would be no network services running except the game itself (assuming it had some kind of server mode) - having anything extra running would just waste resources that could be used for playing.
Local security wouldn't be an issue, since it would inherently be a single user system.
Linux users aren't buying games because there aren't many available...
There are plenty of people, myself included who use Linux for day to day tasks, and windows for games, or use windows for everything because they primarily play games, but would consider linux otherwise.
I find windows absolutely terrible, but often a necessary evil if you want to play games on your computer... Even doing nothing other than gaming, the system gradually breaks down... Once you have a bunch of games installed, along with a bunch of drm drivers shoehorned into the kernel at various points things get really messy. And the hassle of having to keep physical media around because games want you to physically insert the media to play...
And NoCD cracks aren't practical, some prevent you playing online, most prevent you from updating (and modern games are poorly tested, extremely buggy and require updating)..
How about producing gaming livecds? A bootable linux livecd with ati and nvidia drivers preinstalled, configured to go straight into a particular game and nothing else... You could bring some of the convenience of console gaming, while not bogging down someone's live install.
And what about people who downloaded the linux binary to play the demo version?
How about people who acquired the linux binary from mirror sites or third party packages?
There is already a 64bit version of flash:
http://labs.adobe.com/downloads/flashplayer10.html
Depends on your processor really...
On Sparc, MIPS or PPC, having a 64bit browser will just consume more memory (tho it can potentially use more) and possibly run slower...
On x86-64 a 64bit browser may run faster due to having access to a larger pool of registers, as well as being compiled for a more modern architecture which is closer to the cpu being used (most x86 software tends to be compiled for a 386 as the lowest common denominator).