Which also means that, well optimized floating point code (which is more likely, since compilers have had many years to improve their floating point code generation) would run faster on an athlon than poorly written sse code.. Whereas the p4 would run the sse code faster simply because it runs fpu code so poorly. I'm sure well written sse code would run faster on both platforms, atleast in cases where vectorization makes sense.. Intel implemented the weak fpu unit in the p4 to try and steer users onto sse code even when it's not really appropriate.
Will it work based on known file TYPES or known file NAMES ? There is a very important difference, not that most windows users i speak to understand this.
Because the way IE renders things is not documented, changes with each version (even the mac versions render things very differently to the windows version) and has not seen any new features since 2001. Rendering in an ie-compatible way would mean taking out support for modern standards like png and css2, and replacing them with broken half-assed implementations. Also the only way to do this would be to reverse engineer ie, since there are no published standards to implement against. Reverse engineering like this could be considered illegal and nodoubt microsoft would stamp down on it. Aside from the fact that ie is NOT designed to be a decent browser, it's NOT designed to advance the web or to benefit the users. It's sole purpose in existance was to bankrupt netscape and increase dependance on microsoft products.
You bring up an interesting point tho.. Word processors are just designed for quick notes and always have been, the fact that modern word processors have lots of features that should belong in dtp/typesetting apps is a travesty. Years ago, you had word processors - simple tools for writing letters, and dtp/typesetting apps - large complex applications for writing complex professional documents, like publishing magazines. Even microsoft make a product called "publisher" and yet they insist on pushing word into areas where it's not suitable atall.
Gentoo supports binary packages too, and they're just tarfiles so you can install them on any system without needing the gentoo package management tools. The reason you don't hear much about it, is because most people use gentoo *because* of the tweakability of compiling a program.. On the other hand, when you have 6 identically configured gentoo boxes it makes sense to distcc build the packages on one and binary install them on the other 5.
You actually have the driver problem in reverse... Windows users have to find the manufacturer website and download the drivers, or else use the old ones that came with the hardware (if any). Linux distributions typically come with a good set of drivers which will support all the hardware in most modern machines, infact a modern linux distribution like mandrake actually supports MORE hardware out of the box than windows does, such as SATA.. If you have SATA drives then windows won't install without third party drivers, many video cards aren't detected byu default and are driven by generic (slow) drivers. A lot of users don't even realise they need to update the drivers either. Also, linux distributions present a single place to update everything, not just the os but also the drivers, windows update does provide a facility to update your drivers, but most hardware manufacturer's don't take advantage of this so it's rendered useless.
Sure they could, with enough hardware.. But almost any of the commonly used unix based os's when running on the same hardware would outperform it.. The idea of supercomputing is to push the envelope, not running top end hardware at a fraction of it's potential.
But it's not competitive, it's 71 places behind a cluster with exactly the same processors and built from the same dell boxes, only the one 71 places higher has 40 less cpus and runs redhat.. Quite how you can say a windows cluster that's 71 places that's places behind a unix cluster with a lesser number of the same kind of nodes is totally beyond me. That's not competitive, that's an incredibly poorly performing cluster.
Exactly, noone other than ms would build a cluster based on this.. Also consider that this system is a showcase built by ms to try and promote their system, surely they would optimize it as best they can to get the best performance from it, and into the top500 it goes at position 194.. Now scroll up to position 123, same processors, same dell poweredge systems, only this cluster has 40 processors LESS, and it runs redhat 7.3. No-one building a supercomputer would EVER choose ms, precisely because it costs way more and massively underperforms compared to the unix options. The fact that a redhat cluster can beat the ms cluster by 71 places while having 40 less cpus proves that. If this machine hadn't been built and donated by ms, it would be running a unix on the same hardware and be atleast 100 places higher in the top500.
What a waste.. That MS cluster comes in at position 194 with 640 2.4ghz xeon dell poweredge 2650's, but if you look up to position 123 you'l see a cluster of 600 of the same machines and processors.. Ohh, and the one in position 123 runs redhat 7.3. So 40 less machines, 20 places higher in the top500 list. You'd be absoloutely stupid to run windows on a supercomputing cluster, the performance gap between those 2 different machines is HUGE, not to mention the price, 40 more processors plus windows licenses, how many extra processors could you add to the linux based cluster for that money? Noone who was serious about performance would run windows, the only people who will deploy this are those being sponsored by ms.
Runtime optimized code can ve faster than poorly optimized code, since it knows the system it's running on and can profile the behavior of the app.. Anyone wanting supercomputer performance from their app will optimize it for the target system and profile it thoroughly.. they would be absoloutely stupid not to.
The biggest problem has to be the gui and mandatory web browser/email client.. It is a waste of memory and processing time, i say this a lot but people respond with "ohh well it doesnt use much and ram/cpus are cheap today".. So? if you have a cluster of 5000 machines then it all adds up to a lot, and the people who use supercomputers try to squeeze every last mip out of them.
That was exactly my point, ctrl+C atleast (not sure about other key combo's) were used for years before windows existed, even microsoft used ctrl+c to terminate applications under dos and xenix... The Amiga did the same as the mac, there were special "Amiga" keys on the keyboard which were used for keyboard shortcuts. When i first saw a windows keyboard, i assumed they were planning to do the same thing, but no such luck.. On a Mac or Amiga, the cut+paste combo is ALWAYS the same, tho other options may also be available, also on X11 it's always the same... On windows, the cut+paste method changes depending on the application and that is NOT good UI design atall!
Only that in linux, it's located in the same place as all other system configuration files, which also happens to be the same place as on every other major unix based os.. windows does things in a nonstandard way and uses nonstandard places to put files.
Manually lowering a window is very usefull if the window is so big as to completely hide those behind, you cant click on something you can't see, on windows the only solution is to minimise or resize the window infront. The unix system of multiple workspaces is also incredibly usefull, windows doesn't include anything like this by default and the implementations i have seen were basically kludges to mass minimize apps and unminimize another set, it wasn't well integrated into the window manager and didn't work very well.
Have you tried expose on OSX? hit F9 and it will zoom out of all your apps, not quite as flexible as the multiple workspaces most X11 window managers have.. Also, the cut+paste on OSX atleast uses apple specific keys, so you dont lose your ctrl+C combinations etc, which are often used in the commandline... This brings up another point, cut+paste is inconsistent in windows.. Many terminal apps implement it with different keyboard shortcuts than the rest of the os.
Well, that would require downtime, Solaris is very stable and it's not unusual to see machines which have been running for years still running.. Also, newer versions introduce new features which means theres more places for potential security holes to be found.
People will often install trojans on a store computer without booting another os, all you need is a cd with a windows autorun on it and a few seconds to insert it.. Not very difficult and unlikely to get noticed.
Most of the ie-only features aren't innovative atall, and are just nonstandard ways of doing things that can also be accomplished by following standard methods too. Also the only way to support them, is for opensource developers to reverse engineer microsoft code which could be illegal in some places. No websites should use anything until it has been adopted as a formal standard and supported by atleast 2 reference implementations.
Slashdot Mirror
Which also means that, well optimized floating point code (which is more likely, since compilers have had many years to improve their floating point code generation) would run faster on an athlon than poorly written sse code.. Whereas the p4 would run the sse code faster simply because it runs fpu code so poorly.
I'm sure well written sse code would run faster on both platforms, atleast in cases where vectorization makes sense.. Intel implemented the weak fpu unit in the p4 to try and steer users onto sse code even when it's not really appropriate.
Will it work based on known file TYPES or known file NAMES ? There is a very important difference, not that most windows users i speak to understand this.
Netscape 2.0 gold supports automatically installing spyware too? How?
Because the way IE renders things is not documented, changes with each version (even the mac versions render things very differently to the windows version) and has not seen any new features since 2001.
Rendering in an ie-compatible way would mean taking out support for modern standards like png and css2, and replacing them with broken half-assed implementations. Also the only way to do this would be to reverse engineer ie, since there are no published standards to implement against. Reverse engineering like this could be considered illegal and nodoubt microsoft would stamp down on it.
Aside from the fact that ie is NOT designed to be a decent browser, it's NOT designed to advance the web or to benefit the users. It's sole purpose in existance was to bankrupt netscape and increase dependance on microsoft products.
You bring up an interesting point tho..
Word processors are just designed for quick notes and always have been, the fact that modern word processors have lots of features that should belong in dtp/typesetting apps is a travesty. Years ago, you had word processors - simple tools for writing letters, and dtp/typesetting apps - large complex applications for writing complex professional documents, like publishing magazines. Even microsoft make a product called "publisher" and yet they insist on pushing word into areas where it's not suitable atall.
That's the american name, they're called Coco Pops in the UK.
Gentoo supports binary packages too, and they're just tarfiles so you can install them on any system without needing the gentoo package management tools.
The reason you don't hear much about it, is because most people use gentoo *because* of the tweakability of compiling a program..
On the other hand, when you have 6 identically configured gentoo boxes it makes sense to distcc build the packages on one and binary install them on the other 5.
It could if you ran the software from NFS and used your 250mb disk entirely for swap, ofcourse it would be horrendously slow.
You actually have the driver problem in reverse...
Windows users have to find the manufacturer website and download the drivers, or else use the old ones that came with the hardware (if any).
Linux distributions typically come with a good set of drivers which will support all the hardware in most modern machines, infact a modern linux distribution like mandrake actually supports MORE hardware out of the box than windows does, such as SATA.. If you have SATA drives then windows won't install without third party drivers, many video cards aren't detected byu default and are driven by generic (slow) drivers. A lot of users don't even realise they need to update the drivers either.
Also, linux distributions present a single place to update everything, not just the os but also the drivers, windows update does provide a facility to update your drivers, but most hardware manufacturer's don't take advantage of this so it's rendered useless.
Sure they could, with enough hardware.. But almost any of the commonly used unix based os's when running on the same hardware would outperform it..
The idea of supercomputing is to push the envelope, not running top end hardware at a fraction of it's potential.
Well the system in position 123 - 71 places higher, runs redhat 7.3 on the same type of cluster, only the redhat cluster has 40 less processors.
But it's not competitive, it's 71 places behind a cluster with exactly the same processors and built from the same dell boxes, only the one 71 places higher has 40 less cpus and runs redhat..
Quite how you can say a windows cluster that's 71 places that's places behind a unix cluster with a lesser number of the same kind of nodes is totally beyond me. That's not competitive, that's an incredibly poorly performing cluster.
Exactly, noone other than ms would build a cluster based on this..
Also consider that this system is a showcase built by ms to try and promote their system, surely they would optimize it as best they can to get the best performance from it, and into the top500 it goes at position 194..
Now scroll up to position 123, same processors, same dell poweredge systems, only this cluster has 40 processors LESS, and it runs redhat 7.3. No-one building a supercomputer would EVER choose ms, precisely because it costs way more and massively underperforms compared to the unix options. The fact that a redhat cluster can beat the ms cluster by 71 places while having 40 less cpus proves that. If this machine hadn't been built and donated by ms, it would be running a unix on the same hardware and be atleast 100 places higher in the top500.
What a waste..
That MS cluster comes in at position 194 with 640 2.4ghz xeon dell poweredge 2650's, but if you look up to position 123 you'l see a cluster of 600 of the same machines and processors.. Ohh, and the one in position 123 runs redhat 7.3.
So 40 less machines, 20 places higher in the top500 list. You'd be absoloutely stupid to run windows on a supercomputing cluster, the performance gap between those 2 different machines is HUGE, not to mention the price, 40 more processors plus windows licenses, how many extra processors could you add to the linux based cluster for that money?
Noone who was serious about performance would run windows, the only people who will deploy this are those being sponsored by ms.
Runtime optimized code can ve faster than poorly optimized code, since it knows the system it's running on and can profile the behavior of the app..
Anyone wanting supercomputer performance from their app will optimize it for the target system and profile it thoroughly.. they would be absoloutely stupid not to.
The biggest problem has to be the gui and mandatory web browser/email client..
It is a waste of memory and processing time, i say this a lot but people respond with "ohh well it doesnt use much and ram/cpus are cheap today"..
So? if you have a cluster of 5000 machines then it all adds up to a lot, and the people who use supercomputers try to squeeze every last mip out of them.
That was exactly my point, ctrl+C atleast (not sure about other key combo's) were used for years before windows existed, even microsoft used ctrl+c to terminate applications under dos and xenix...
The Amiga did the same as the mac, there were special "Amiga" keys on the keyboard which were used for keyboard shortcuts. When i first saw a windows keyboard, i assumed they were planning to do the same thing, but no such luck..
On a Mac or Amiga, the cut+paste combo is ALWAYS the same, tho other options may also be available, also on X11 it's always the same... On windows, the cut+paste method changes depending on the application and that is NOT good UI design atall!
Only that in linux, it's located in the same place as all other system configuration files, which also happens to be the same place as on every other major unix based os.. windows does things in a nonstandard way and uses nonstandard places to put files.
Use X11 and xterm, or whatever X terminal of your choice.. cut+paste seems to work fine in X11 apps running under the default OSX Xserver.
Manually lowering a window is very usefull if the window is so big as to completely hide those behind, you cant click on something you can't see, on windows the only solution is to minimise or resize the window infront. The unix system of multiple workspaces is also incredibly usefull, windows doesn't include anything like this by default and the implementations i have seen were basically kludges to mass minimize apps and unminimize another set, it wasn't well integrated into the window manager and didn't work very well.
Have you tried expose on OSX? hit F9 and it will zoom out of all your apps, not quite as flexible as the multiple workspaces most X11 window managers have..
Also, the cut+paste on OSX atleast uses apple specific keys, so you dont lose your ctrl+C combinations etc, which are often used in the commandline... This brings up another point, cut+paste is inconsistent in windows.. Many terminal apps implement it with different keyboard shortcuts than the rest of the os.
Well, that would require downtime, Solaris is very stable and it's not unusual to see machines which have been running for years still running..
Also, newer versions introduce new features which means theres more places for potential security holes to be found.
I used to do that with the Amiga, you could create a reset resident ramdisk so you could copy your OS into it and reboot from it..
People will often install trojans on a store computer without booting another os, all you need is a cd with a windows autorun on it and a few seconds to insert it.. Not very difficult and unlikely to get noticed.
Most of the ie-only features aren't innovative atall, and are just nonstandard ways of doing things that can also be accomplished by following standard methods too.
Also the only way to support them, is for opensource developers to reverse engineer microsoft code which could be illegal in some places.
No websites should use anything until it has been adopted as a formal standard and supported by atleast 2 reference implementations.