Well the processing power of Altivec or MMX/SSE/3DNow or whatever is nowhere near the power of you newest NVidia/ATI card you have surely bought for playing Doom III. Why not use it then? Get the brook compiler! Furthemore, I see they introduce classes like vec, etc. Such classes have been already designed successfuly for C++. Why not try porting Blitz to the Altivec and/or to the GPU?
As long as job is done. I choose the tools, I shoulder the blame when something goes wrong. And binary only distros have their own problems. I can remember a Debian package that would assume I have a Debian stock binary kernel package installed, not a custom compiled kernel. Finally I had to unpack the package, tweak a setting or two, pack it again and then do the install. No such problems when you are building from source (BSD's, Gentoo). However long compile times (esp. KDE) can be problematic on, say, a notebook.
I wrote 'user time'. Have you never heard of Unix or something? "User time" is time spent in CPU. It does not take into account the I/O. The I/O goes into "system time" or "real time" (aka 'wall clock time'). Therefore "user time" indicates relative CPU performance.
I have lately started thinking about buying MacMini, iBook or a PowerBook. Because I most often use Gentoo GNU/Linux or FreeBSD (but I also use Fedora, SuSE, M$ Windows and Solaris) therefore I have done some benchmarks to compare compile times on some different architectures. My software of choice is PostgreSQL database since it's size is just right (a few minutes of compile time anyway). Ok, here come the benchmarks (the most important is the 'user time' which is how long really the compile took, it discard other factors like filesystem, HDD and system load factors):
In the news.com.com.com.com article you read:
The white paper also suggests tightening the rules under which patents are issued to allow both proposed and issued patents to be challenged more easily.
This is very, very funny, indeed... emphasis mine.
Is of course the GNU/Darwin. Somehow it is what GNU always wanted to have - a GNU running on a microkernel (at least sort of). (I admit, I don't know what licence applies to Darwin).
Ok, Dothan is an excellent processor. The only question is whether Intel can beat its own marketing. When I recommend Athlon 64 over P4 Prescott to my friends they always say: P4 has HyperThreading! I will be able to run two apps a the same time! That's tremendous! I can't buy this Athlon thing, it has no HT!. And unfortunately for Intel there are some troubles with incorporating HyperThreading into Dothan's architecture due to short pipeline (HT was a hack to fix problems resulting from large wrong-branch-prediction penalty). So people not willing to buy Athlon will not want to buy Dothan (or any Pentium M) either. At least until it has HT or Intel convinces people that all their PR said was wrong.
There is an article in one of polish most important newspapers, the Gazeta Wyborcza.
This article claims, that one of the forces behind Mr Marcinski's mission was polish ex-Prime Minister Professor Jerzy Buzek, currently in European Parliament.
Sadly, the same article informs us that probably Poland will accept the directive in January, under the presidency of Luxembourg. Polish diplomats are afraid that voting against the directive at this late stage would endanger polish interest in other fields, where Poland fears that other countries would act similarly, withdrawing already voiced approvals. Example of such field is the REACH package. Nonetheless Mr Jerzy Buzek and Mr Adam Gierek will try to make some amendments to the directive that would clarify that software is not patentable.
So, we won a battle, but war is probably lost anyway - we can only try to limit the scale of our defeat. The link to the article [in Polish]
Unfortunately you don't have Polish->English option in babelfish.altavista.com
The problem is, as you can see, that patents are not beneficial to anyone involved. They hurt the Open Source, but they hurt Big Business also. The governments should not listen to what big companies say, because they don't know what they are doing. In best interest of all parties involved, the patents on software and business methods should never be allowed.
Google starts to suck
on
Google Suggest
·
· Score: 1, Flamebait
Well the new Groups really suck - try reading any group that deals with postfix or sendmail - all examples really screwed up.
And now that... - Google is going the same way that all went. Take for instance Microsoft. Ten years ago it was the icon of geekhood, they made a really good compiler, assembler and whatnot. And now? They even had a lapse of sobriety with Win 2k - quite a decent product - but it finally got fscked up by creeping featuritis disease when transformed into Win XP Proffesional.
And now Google. Bill Gates is ready to welcome you in the club.
What I see as a vital point in distributing code with source and license allowing for changes is the ability for the users to change the code. This might seem bit obvious, but it was the nerve behind GPL (you can read about it in the book about Stallman). For me Linux or BSD are much better than Solaris, because when something does not work in Solaris I have to find a klugde to go around. If I have the same problem in Linux or BSD I can always fix the broken code. This is of course tedious task, but sometimes you have no other choice. In closed source environment you don't have such an option. In my opinion releasing Solaris with source code would really help. It would also allow for writing better software for Solaris, since there is no better way to understand the inner workings of some software feature than to see the source code. The only problem is how much code would be open sourced. I would hope for the most important parts of kernel (memory, scheduler) and network stack (fire engine). Otherwise there is really not much sense in open sourcing (at least in my opinion).
Well, Intel is doing excellent job, unlike nVidia and ATI. And as to the docs, well I think that Intel will have nothing against looking at them, ehm, over the shoulder, you know.
I was speaking about Red Hat, perhaps generally about Linux, and I always thought FreeBSD is not a distribution of Linux. But if I am wrong, please do correct me.
OTOH if you think that FreeBSD is _very_ stable, I guess you should have tried 5.3 before the release. There were some really serious bugs. As far as I know the SCHED_ULE still is not ready for production and we have to get on with the old SCHED_4BSD.
However, if you only wanted to say that if I cannot choose between different flavours of Linux I should take a look at FreeBSD - well that is not always that easy. I prefer to use FreeBSD where I can, but there are things Linux has and FreeBSD does not. For instance native Java 1.5.
1) Fedora is a good distro, if you want the newest staff it will never be rock-solid, try for instance the debian-unstable, or Gentoo when it has a bad day
2) RHEL is something important - RH needs money to support itself. A good UNIX operating system (like Linux cannot be cheap). Also people from management want to pay because:
- they think if something is free cannot be good
- they think that you should have someone you can blame
3) RH didn't steal the Linux - it is free, what you have to pay for are two things: trademark and support - if you can support yourself and don't care about trademarks but have to use software that needs RHEL try a RHEL clone. On the other hand if you have enough money to afford such software (think Oracle) why not give some to the Linux community.
But it's only my 0.02 Euro...
The inside of the BTX case looks very similar to the workstation designs, especially those of SGI. I remember that the Indy workstation didn't have a single fan and was virtually noiseless due to correctly designed air ducts.
PCI Express which is somewhat like SGI's crossbar (PCI Express uses switch instead of bus), AMD's on-CPU memory controllers with NUMA, SATA almost like SCSI, etc. made PC's more and more like workstations. I think that correct thermal design is the last and final thing and BTX is a big step in this direction.
It is a matter of personal feelings, however I never liked ettercap - for me too shiny. I have preferred tcpdump end, especially, [t]ethereal. Ethereal has a decent Gnome GUI for those who do like such things.
I have worked with two programs for designing buildings (Finite Elements Method) - one was designed according to the Windows(tm) Interface Design Guidelines - working with this program was a nightmare, while the second was designed to naturally mirror the steps engineer takes - and it was real pleasure to work with it. However the second one could never qualify for a certificate of conformance to Windows GUI standards.
Slashdot Mirror
Well the processing power of Altivec or MMX/SSE/3DNow or whatever is nowhere near the power of you newest NVidia/ATI card you have surely bought for playing Doom III. Why not use it then? Get the brook compiler! Furthemore, I see they introduce classes like vec, etc. Such classes have been already designed successfuly for C++. Why not try porting Blitz to the Altivec and/or to the GPU?
As long as job is done. I choose the tools, I shoulder the blame when something goes wrong. And binary only distros have their own problems. I can remember a Debian package that would assume I have a Debian stock binary kernel package installed, not a custom compiled kernel. Finally I had to unpack the package, tweak a setting or two, pack it again and then do the install. No such problems when you are building from source (BSD's, Gentoo). However long compile times (esp. KDE) can be problematic on, say, a notebook.
I wrote 'user time'. Have you never heard of Unix or something? "User time" is time spent in CPU. It does not take into account the I/O. The I/O goes into "system time" or "real time" (aka 'wall clock time'). Therefore "user time" indicates relative CPU performance.
Well, not bad for 0.8 GHz, heh
In the news.com.com.com.com article you read:
The white paper also suggests tightening the rules under which patents are issued to allow both proposed and issued patents to be challenged more easily.
This is very, very funny, indeed... emphasis mine.
Please, excuse me Anonymous Cowardous Troll, but Darwin _is_ a microkernel based kernel, like Hurd.
Is of course the GNU/Darwin. Somehow it is what GNU always wanted to have - a GNU running on a microkernel (at least sort of). (I admit, I don't know what licence applies to Darwin).
Ok, Dothan is an excellent processor. The only question is whether Intel can beat its own marketing. When I recommend Athlon 64 over P4 Prescott to my friends they always say: P4 has HyperThreading! I will be able to run two apps a the same time! That's tremendous! I can't buy this Athlon thing, it has no HT!. And unfortunately for Intel there are some troubles with incorporating HyperThreading into Dothan's architecture due to short pipeline (HT was a hack to fix problems resulting from large wrong-branch-prediction penalty). So people not willing to buy Athlon will not want to buy Dothan (or any Pentium M) either. At least until it has HT or Intel convinces people that all their PR said was wrong.
There is also more coverage there. There are some very fine points explained in this article.
There is an article in one of polish most important newspapers, the Gazeta Wyborcza.
This article claims, that one of the forces behind Mr Marcinski's mission was polish ex-Prime Minister Professor Jerzy Buzek, currently in European Parliament.
Sadly, the same article informs us that probably Poland will accept the directive in January, under the presidency of Luxembourg. Polish diplomats are afraid that voting against the directive at this late stage would endanger polish interest in other fields, where Poland fears that other countries would act similarly, withdrawing already voiced approvals. Example of such field is the REACH package. Nonetheless Mr Jerzy Buzek and Mr Adam Gierek will try to make some amendments to the directive that would clarify that software is not patentable.
So, we won a battle, but war is probably lost anyway - we can only try to limit the scale of our defeat.
The link to the article [in Polish]
Unfortunately you don't have Polish->English option in babelfish.altavista.com
eProvisia Spam Eradicator. If it's good enough for lcamtuf then it's good enough for me.
You will need two such cards to play Doom4 in 640x480 at 25 fps :)
For my brand new Prescott based computer I need at least 390W :)
The problem is, as you can see, that patents are not beneficial to anyone involved. They hurt the Open Source, but they hurt Big Business also. The governments should not listen to what big companies say, because they don't know what they are doing. In best interest of all parties involved, the patents on software and business methods should never be allowed.
Well the new Groups really suck - try reading any group that deals with postfix or sendmail - all examples really screwed up.
And now that... - Google is going the same way that all went. Take for instance Microsoft. Ten years ago it was the icon of geekhood, they made a really good compiler, assembler and whatnot. And now? They even had a lapse of sobriety with Win 2k - quite a decent product - but it finally got fscked up by creeping featuritis disease when transformed into Win XP Proffesional.
And now Google. Bill Gates is ready to welcome you in the club.
What I see as a vital point in distributing code with source and license allowing for changes is the ability for the users to change the code. This might seem bit obvious, but it was the nerve behind GPL (you can read about it in the book about Stallman). For me Linux or BSD are much better than Solaris, because when something does not work in Solaris I have to find a klugde to go around. If I have the same problem in Linux or BSD I can always fix the broken code. This is of course tedious task, but sometimes you have no other choice. In closed source environment you don't have such an option. In my opinion releasing Solaris with source code would really help. It would also allow for writing better software for Solaris, since there is no better way to understand the inner workings of some software feature than to see the source code. The only problem is how much code would be open sourced. I would hope for the most important parts of kernel (memory, scheduler) and network stack (fire engine). Otherwise there is really not much sense in open sourcing (at least in my opinion).
He just has to buy the relevant patents.
Well, Intel is doing excellent job, unlike nVidia and ATI. And as to the docs, well I think that Intel will have nothing against looking at them, ehm, over the shoulder, you know.
I was speaking about Red Hat, perhaps generally about Linux, and I always thought FreeBSD is not a distribution of Linux. But if I am wrong, please do correct me.
OTOH if you think that FreeBSD is _very_ stable, I guess you should have tried 5.3 before the release. There were some really serious bugs. As far as I know the SCHED_ULE still is not ready for production and we have to get on with the old SCHED_4BSD.
However, if you only wanted to say that if I cannot choose between different flavours of Linux I should take a look at FreeBSD - well that is not always that easy. I prefer to use FreeBSD where I can, but there are things Linux has and FreeBSD does not. For instance native Java 1.5.
1) Fedora is a good distro, if you want the newest staff it will never be rock-solid, try for instance the debian-unstable, or Gentoo when it has a bad day
2) RHEL is something important - RH needs money to support itself. A good UNIX operating system (like Linux cannot be cheap). Also people from management want to pay because:
- they think if something is free cannot be good
- they think that you should have someone you can blame
3) RH didn't steal the Linux - it is free, what you have to pay for are two things: trademark and support - if you can support yourself and don't care about trademarks but have to use software that needs RHEL try a RHEL clone. On the other hand if you have enough money to afford such software (think Oracle) why not give some to the Linux community.
But it's only my 0.02 Euro...
The inside of the BTX case looks very similar to the workstation designs, especially those of SGI. I remember that the Indy workstation didn't have a single fan and was virtually noiseless due to correctly designed air ducts.
PCI Express which is somewhat like SGI's crossbar (PCI Express uses switch instead of bus), AMD's on-CPU memory controllers with NUMA, SATA almost like SCSI, etc. made PC's more and more like workstations. I think that correct thermal design is the last and final thing and BTX is a big step in this direction.
It is a matter of personal feelings, however I never liked ettercap - for me too shiny. I have preferred tcpdump end, especially, [t]ethereal. Ethereal has a decent Gnome GUI for those who do like such things.
I have worked with two programs for designing buildings (Finite Elements Method) - one was designed according to the Windows(tm) Interface Design Guidelines - working with this program was a nightmare, while the second was designed to naturally mirror the steps engineer takes - and it was real pleasure to work with it. However the second one could never qualify for a certificate of conformance to Windows GUI standards.
Give me back my gigahertz!!!
I want all my precioussss... gigahertz!!!