Re:Python is part of the answer
on
Open Source Math
·
· Score: 1
Well, you're only really responsible for the correctness of your own code. As to the compiler and CPU, so long as you use a combination that have been verified as correct by other mathematicians you should be fine.
Well, it seems pointless to go to such extremes to get as much performance as possible from your hardware (and potentially shortening its lifespan or voiding its warrantee), only to then waste it running inefficient software. Surely a better idea would be more efficient software coupled with more reliable hardware.
As to your suggestion of the software performance being the programmer's concern, it could similarly be said that hardware performance is the concern of the engineers who designed it.
It's more fun, tho just changing the DNS is pretty stupid because someone can always access servers via direct IP, or simply hard set their dns servers. My laptop runs it's own dns cache that directly queries the root servers itself, because of ISPs with unreliable nameservers. On the other hand, I had an intentionally unencrypted wireless network that dropped you into a honeypot network with a few funny sites, the only way out was through a vpn server for which you needed an appropriate cert.
Netstumbler is awfull, all it does is send out probe requests which make it blatantly obvious what your doing. It also won't detect cloaked access points, give you any idea how many clients are connected to each ap or log any unencrypted traffic thats receivable by your card. Try Kismet or KisMac (mac gui version), it works a _LOT_ better.
Did you just keep hitting the "scan for networks" option on the 770? It's not always the most reliable... Especially if the signals are fairly weak. Also if an AP is inside a building, depending what the walls are made of you might not see it. Try installing kismet on your 770, and get a bluetooth GPS... Then drive around and see what you find, you can even plot a nice map showing all the points. I drove around a few residential streets here and found several hundred by driving around the outskirts of a housing estate. Of those, about 30% were open.
But how about squeezing every last bit of performance out of the software that runs on those chips?
Incidentally, i want quiet and power efficient on the systems i keep running 24/7 (my laptop, tho its usually suspended, and a media server) at home... Tho for a system that boots up to play games and then gets turned off i'm not so concerned. In a datacenter i want power efficiency and performance, noise is irrelevant.
There are still companies selling snakeoil encryption, and many of these products are easily pulled apart by a skilled cryptographer/reverse engineer...
As to the NSA's secrets... Really secret data is also kept physically secure, so you'd have to actually get hold of the encrypted data before you could even think about cracking it.
Not to mention the fact that a medication that merely subdues an illness, such that you need too keep taking the medication indefinitely is far more profitable than one that actually cures the illness.
Pharmaceutical companies would rather have AIDS sufferers on expensive combination therapies for 30+ years until they die, than a 1 month course that cures them. If you ask me, all medical research should be done by non profits or government, all their research goes public domain and pharmaceutical companies should be reduced to simple free market manufacturing of published formulae. Many countries already have nationalised medical services, or taxed medical insurance. Fund the research with this money.
That's not copyright laws your talking about, it's trademark laws. As the previous post says, preventing derivative works hinders our culture... And as you said, half-assed knock offs under the same brand would hinder people's enjoyment of the originals... But what about a derivative work which is clearly differentiated from the original? The "Harry Potter Lexicon" is clearly not a harry potter story, it is quite clearly a lexicon about harry potter, that is a descriptive guide. The only suggestion i would make, is that the name is perhaps changed to "The Unofficial Harry Potter Lexicon"... Or how about a derivative work that didn't use the brand, let's say a book called "Barry Rotter" or something, but used very similar names and stories... You wouldn't mistake this for the original brand.
No, you simply prove my point, that anyone not towing the most widely publicised line will be branded as insane.
Are you seriously trying to tell me that the governments of the US and UK haven't done unpleasant things and then covered them up? They've even been caught out a few times, how many more cases are there which haven't been exposed, or where the exposing was simply branded as insanity?
Ehm, when windows 95 came out 486s were common and early pentiums were around... a 386 was the absolute minimum spec, and was considered slow even in those days.
A little 40GB HD for swap probably won't be very quick. Newer drives are a lot faster because of the higher data density, a single revolution of the platter can read a lot more data on a 500GB drive than a 40GB.
Had you run windows 3.1 on that P100 with 8mb it would have flown... Win95 was bloated for it's time.
I agree that KDE and Gnome are bloated, but they are also far more fully featured than the default windows desktop environment. Windows provides you with a basic window manager and a rudimentary file manager and application launcher by default, i believe xfce also provides this and more. KDE provides you with things like kioslaves, among others. You cannot compare the default windows environment to a full blown environment like KDE.
Linux can still be mean and lean, if you accept a reduced feature set. An equivalent desktop feature set on linux is still faster than windows on the same hardware.
Why not try setting up machines with comparable environments and features and see which is faster.
The x86 compatibles succeeded because they were a comparably open architecture, while the competition despite being almost universally superior was not. You had a choice of vendors and plenty of competition bringing prices down and new features up. Windows and Dos got dragged along because they were considered a very small part of the overall cost, and it was better to have open hardware and closed software than closed hardware with closed software. Eventually the same thing should happen to software aswell, microsoft are trying to use their current position to hold on to that as long as possible just like ibm tried to do with the mainframe.
If one of the other architectures had been as open as the ibm compatible pc, the industry could be very different today. Imagine if Amiga clones had been available from multiple vendors in 1985 at half the price of what commodore was charging.
With monopolies the threshold for shit is much higher... Using proprietary formats/protocols/etc also raises the threshold for shit. Your product only has to be decent enough that the cost of replacing it outweighs the hassle it brings. And often even this is only considered short term (where the costs of replacement are higher, rather than long term where the benefits of removing the proprietary lock in are much higher). So the threshold for shitness of windows is much higher than the threshold of say suse linux (which is easily replaced with redhat if it becomes too shit).
Spend less on hardware.. Spend less on power.. Run more apps on less servers (including virtualization etc).. Extend the useful life of hardware.
Optimization is a feature very desirable to customers, especially large ones where the costs of inefficient code are multiplied.
Open source can help very much with the points you bring up too... Instead of different vendors writing their own implementations of every feature, everything needs only be written once and reused and improved (optimized etc) going forward.
Firstly your not considering the extra power consumed by the extra ram.
Assuming that your code will only ever run on a single machine, then hardware is cheaper short term... But what if your code will run on 500 machines? Your individual app may not, but the improvements you made to ruby probably will, so you have potentially saved money for millions of people.
What is the MAXIMUM ram your server can support?, by lowering your memory usage you leave more headroom before you hit that maximum and have to buy a whole new server.
The difference is that Linux and OSX are relatively clean designs, which was a half decent design in the first place. Windows on the other hand has gone through many half-assed decisions, which were deemed to be design flaws and changed in later versions (while keeping the old code around too for compatibility reasons)... resulting in an ever increasing and less manageable mess of code. As an example, password hashing on windows is done twice using 2 algorithms, neither are great but one is significantly weaker. The weaker one is kept for compatibility with older apps. Unix on the other hand, has a standard crypt() call, that can use any algorithm. There is also the PAM system which further abstracts the authentication process.
Linux/OSX on the other hand, stick to the original principles of unix, and although some backend things may change, the abstraction presented to user mode apps remains the same.
And do they only run one app on each machine? It all adds up, and all that extra bloat leaves you less space for your data files... If you need lots of apps in one machine, especially a laptop, you need to buy higher capacity drives which tend to be more expensive per capacity. If you want to run lots of apps at once, bloated apps make that harder. If you have a multiuser system where lots of people will be running apps, bloated apps significantly increase the resources you need. If you have a laptop, it will get hotter and drain the battery quicker (a desktop will also get hotter and draw more power). If you have lots of machines, the increased power usage multiplied by the number of machines can be quite significant.
Energy is expensive and getting more so by the day.
If software were more efficient we could get by with much older machines for day to day use, or better yet modern very power efficient designs. Laptops could be smaller, lighter, and have far more battery life. Data centers and large offices could save a significant amount of money on energy costs.
The ever increasing bloat of software is a terrible thing, and is causing problems far and wide.
Obligatory car analogy...
If you make a really heavy car, you need a big engine to drive it and it still won't perform very well. If you use a lightweight body, you can use a much smaller engine to achieve the same performance but with significantly less fuel. Or you can use the same large engine, and have a high performance vehicle.
I want the choice between a small lightweight laptop that lasts for hours, or a bigger heavier laptop that is insanely fast. Instead, because of bloat, i'm stuck with a bigger heavier laptop with so-so performance because it's bogged down by heavyweight software. Computers today are massively faster than they were 10 years ago, but the average user experience is roughly the same because the software has got slower to compensate for the extra performance.
Try running a really old OS on a modern computer, if you can get it to work properly just look at the performance difference. I used to use windows 3.1 on a windows 98 era machine (p233, 64mb ram) and it was insanely quick.
fat32 is a really quite crappy filesystem... No support for files over 4gb (most common use: dvd images), not case sensitive, no support for permissions, no journaling, no symlinks etc... I tend to use EXT3, linux/bsd support it natively, and third party drivers are available for osx and windows.
The eyecandy isn't that great, if you want great eyecandy try ubuntu with compiz/beryl... If you want things to "just work" your best off with Apple, linux/windows are always dogged with the problems of having to support so many different types of hardware.
Not that old, windows 3.1 was MASSIVELY faster if you turned off swap... You could safely do it with 16mb ram, or even 8mb at a push. We went and turned swap off on all the machines in school, they all had 16mb but were using runtime compression on the HD, including the swap file resulting in terrible performance.
Slashdot Mirror
Well, you're only really responsible for the correctness of your own code.
As to the compiler and CPU, so long as you use a combination that have been verified as correct by other mathematicians you should be fine.
Well, it seems pointless to go to such extremes to get as much performance as possible from your hardware (and potentially shortening its lifespan or voiding its warrantee), only to then waste it running inefficient software. Surely a better idea would be more efficient software coupled with more reliable hardware.
As to your suggestion of the software performance being the programmer's concern, it could similarly be said that hardware performance is the concern of the engineers who designed it.
If you want encrypted mail, run the encryption yourself... GPG is freely available. Then it doesn't matter via which service you transmit the mail.
It's more fun, tho just changing the DNS is pretty stupid because someone can always access servers via direct IP, or simply hard set their dns servers. My laptop runs it's own dns cache that directly queries the root servers itself, because of ISPs with unreliable nameservers.
On the other hand, I had an intentionally unencrypted wireless network that dropped you into a honeypot network with a few funny sites, the only way out was through a vpn server for which you needed an appropriate cert.
Netstumbler is awfull, all it does is send out probe requests which make it blatantly obvious what your doing. It also won't detect cloaked access points, give you any idea how many clients are connected to each ap or log any unencrypted traffic thats receivable by your card. Try Kismet or KisMac (mac gui version), it works a _LOT_ better.
Did you just keep hitting the "scan for networks" option on the 770? It's not always the most reliable... Especially if the signals are fairly weak.
Also if an AP is inside a building, depending what the walls are made of you might not see it.
Try installing kismet on your 770, and get a bluetooth GPS... Then drive around and see what you find, you can even plot a nice map showing all the points. I drove around a few residential streets here and found several hundred by driving around the outskirts of a housing estate. Of those, about 30% were open.
Intel's videocards have always been very much budget cards... Fine for general office computers but useless for gaming or heavy video related work.
But how about squeezing every last bit of performance out of the software that runs on those chips?
Incidentally, i want quiet and power efficient on the systems i keep running 24/7 (my laptop, tho its usually suspended, and a media server) at home... Tho for a system that boots up to play games and then gets turned off i'm not so concerned.
In a datacenter i want power efficiency and performance, noise is irrelevant.
There are still companies selling snakeoil encryption, and many of these products are easily pulled apart by a skilled cryptographer/reverse engineer...
As to the NSA's secrets... Really secret data is also kept physically secure, so you'd have to actually get hold of the encrypted data before you could even think about cracking it.
Not to mention the fact that a medication that merely subdues an illness, such that you need too keep taking the medication indefinitely is far more profitable than one that actually cures the illness.
Pharmaceutical companies would rather have AIDS sufferers on expensive combination therapies for 30+ years until they die, than a 1 month course that cures them.
If you ask me, all medical research should be done by non profits or government, all their research goes public domain and pharmaceutical companies should be reduced to simple free market manufacturing of published formulae. Many countries already have nationalised medical services, or taxed medical insurance. Fund the research with this money.
That's not copyright laws your talking about, it's trademark laws.
As the previous post says, preventing derivative works hinders our culture... And as you said, half-assed knock offs under the same brand would hinder people's enjoyment of the originals...
But what about a derivative work which is clearly differentiated from the original? The "Harry Potter Lexicon" is clearly not a harry potter story, it is quite clearly a lexicon about harry potter, that is a descriptive guide. The only suggestion i would make, is that the name is perhaps changed to "The Unofficial Harry Potter Lexicon"...
Or how about a derivative work that didn't use the brand, let's say a book called "Barry Rotter" or something, but used very similar names and stories... You wouldn't mistake this for the original brand.
Unfortunately true...
But they could have designed it properly in the first place (the unix hierarchy hasn't changed)...
Alternatively they could do what apple did with the transition from OS9 to OSX.
Otherwise people are just stuck with a poorly designed crufty system.
No, you simply prove my point, that anyone not towing the most widely publicised line will be branded as insane.
Are you seriously trying to tell me that the governments of the US and UK haven't done unpleasant things and then covered them up? They've even been caught out a few times, how many more cases are there which haven't been exposed, or where the exposing was simply branded as insanity?
Ehm, when windows 95 came out 486s were common and early pentiums were around... a 386 was the absolute minimum spec, and was considered slow even in those days.
A little 40GB HD for swap probably won't be very quick.
Newer drives are a lot faster because of the higher data density, a single revolution of the platter can read a lot more data on a 500GB drive than a 40GB.
Had you run windows 3.1 on that P100 with 8mb it would have flown... Win95 was bloated for it's time.
I agree that KDE and Gnome are bloated, but they are also far more fully featured than the default windows desktop environment.
Windows provides you with a basic window manager and a rudimentary file manager and application launcher by default, i believe xfce also provides this and more. KDE provides you with things like kioslaves, among others. You cannot compare the default windows environment to a full blown environment like KDE.
Linux can still be mean and lean, if you accept a reduced feature set. An equivalent desktop feature set on linux is still faster than windows on the same hardware.
Why not try setting up machines with comparable environments and features and see which is faster.
The x86 compatibles succeeded because they were a comparably open architecture, while the competition despite being almost universally superior was not. You had a choice of vendors and plenty of competition bringing prices down and new features up.
Windows and Dos got dragged along because they were considered a very small part of the overall cost, and it was better to have open hardware and closed software than closed hardware with closed software. Eventually the same thing should happen to software aswell, microsoft are trying to use their current position to hold on to that as long as possible just like ibm tried to do with the mainframe.
If one of the other architectures had been as open as the ibm compatible pc, the industry could be very different today.
Imagine if Amiga clones had been available from multiple vendors in 1985 at half the price of what commodore was charging.
With monopolies the threshold for shit is much higher...
Using proprietary formats/protocols/etc also raises the threshold for shit.
Your product only has to be decent enough that the cost of replacing it outweighs the hassle it brings. And often even this is only considered short term (where the costs of replacement are higher, rather than long term where the benefits of removing the proprietary lock in are much higher).
So the threshold for shitness of windows is much higher than the threshold of say suse linux (which is easily replaced with redhat if it becomes too shit).
But your customers may also want to:
Spend less on hardware..
Spend less on power..
Run more apps on less servers (including virtualization etc)..
Extend the useful life of hardware.
Optimization is a feature very desirable to customers, especially large ones where the costs of inefficient code are multiplied.
Open source can help very much with the points you bring up too... Instead of different vendors writing their own implementations of every feature, everything needs only be written once and reused and improved (optimized etc) going forward.
Firstly your not considering the extra power consumed by the extra ram.
Assuming that your code will only ever run on a single machine, then hardware is cheaper short term...
But what if your code will run on 500 machines? Your individual app may not, but the improvements you made to ruby probably will, so you have potentially saved money for millions of people.
What is the MAXIMUM ram your server can support?, by lowering your memory usage you leave more headroom before you hit that maximum and have to buy a whole new server.
The difference is that Linux and OSX are relatively clean designs, which was a half decent design in the first place.
Windows on the other hand has gone through many half-assed decisions, which were deemed to be design flaws and changed in later versions (while keeping the old code around too for compatibility reasons)... resulting in an ever increasing and less manageable mess of code.
As an example, password hashing on windows is done twice using 2 algorithms, neither are great but one is significantly weaker. The weaker one is kept for compatibility with older apps.
Unix on the other hand, has a standard crypt() call, that can use any algorithm. There is also the PAM system which further abstracts the authentication process.
Linux/OSX on the other hand, stick to the original principles of unix, and although some backend things may change, the abstraction presented to user mode apps remains the same.
And do they only run one app on each machine?
It all adds up, and all that extra bloat leaves you less space for your data files...
If you need lots of apps in one machine, especially a laptop, you need to buy higher capacity drives which tend to be more expensive per capacity.
If you want to run lots of apps at once, bloated apps make that harder.
If you have a multiuser system where lots of people will be running apps, bloated apps significantly increase the resources you need.
If you have a laptop, it will get hotter and drain the battery quicker (a desktop will also get hotter and draw more power).
If you have lots of machines, the increased power usage multiplied by the number of machines can be quite significant.
Energy is expensive and getting more so by the day.
If software were more efficient we could get by with much older machines for day to day use, or better yet modern very power efficient designs.
Laptops could be smaller, lighter, and have far more battery life.
Data centers and large offices could save a significant amount of money on energy costs.
The ever increasing bloat of software is a terrible thing, and is causing problems far and wide.
Obligatory car analogy...
If you make a really heavy car, you need a big engine to drive it and it still won't perform very well.
If you use a lightweight body, you can use a much smaller engine to achieve the same performance but with significantly less fuel. Or you can use the same large engine, and have a high performance vehicle.
I want the choice between a small lightweight laptop that lasts for hours, or a bigger heavier laptop that is insanely fast. Instead, because of bloat, i'm stuck with a bigger heavier laptop with so-so performance because it's bogged down by heavyweight software. Computers today are massively faster than they were 10 years ago, but the average user experience is roughly the same because the software has got slower to compensate for the extra performance.
Try running a really old OS on a modern computer, if you can get it to work properly just look at the performance difference. I used to use windows 3.1 on a windows 98 era machine (p233, 64mb ram) and it was insanely quick.
fat32 is a really quite crappy filesystem...
No support for files over 4gb (most common use: dvd images), not case sensitive, no support for permissions, no journaling, no symlinks etc...
I tend to use EXT3, linux/bsd support it natively, and third party drivers are available for osx and windows.
The eyecandy isn't that great, if you want great eyecandy try ubuntu with compiz/beryl...
If you want things to "just work" your best off with Apple, linux/windows are always dogged with the problems of having to support so many different types of hardware.
Not that old, windows 3.1 was MASSIVELY faster if you turned off swap...
You could safely do it with 16mb ram, or even 8mb at a push.
We went and turned swap off on all the machines in school, they all had 16mb but were using runtime compression on the HD, including the swap file resulting in terrible performance.