Re: GCC 4.1 now the default GCC version for etch
On Wed, Jun 21, 2006 at 01:51:46AM +0200, Adam Borowski wrote: The lack of x87 wouldn't be a problem. At least, i386 still has the code left from the old days and will emulate it transparently; when such an amd64 CPU pops out the x87 emulation can be ported. In fact, it has a separate hardware path from all other types of unknown opcodes (INT7 instead of INT6); x87 was _designed_ to be emulated if not present from the very start. x87 was a seperate chip. No wonder it has always been treated as rather seperate. :) That is probably also the cause of its lousy design. They probably had no choice when it was a seperate chip, the stack interface was probably the simplest given sharing the memory bus was probably not something they wanted to do at the time along with putting memory access hardware into the x87 chips. Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Lennart Sorensen [EMAIL PROTECTED] writes: On Sun, Jun 18, 2006 at 11:40:03PM +0200, Wouter Verhelst wrote: On Sun, Jun 18, 2006 at 02:24:35PM -0700, Blars Blarson wrote: (amd64 is only faster in 64-bit mode because of all the poorly designed x86 32-bit instruction set.) x86 32-bit instruction set and designed in one sentence? Hah. How about the fact it has more registers available in 64bit mode. People always said the x86 didn't have enough registers after all. Using sse for floating point rather than the awful stack based x87 probably helps too. Len Sorensen Exactly. Isn't the x86_64 instruction set basicaly the same as ia32 just with a few extra opcodes and more registers? Any general fault in the instruction set should still remain. MfG Goswin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
On Tue, Jun 20, 2006 at 12:32:18PM +0200, Goswin von Brederlow wrote: Exactly. Isn't the x86_64 instruction set basicaly the same as ia32 just with a few extra opcodes and more registers? Any general fault in the instruction set should still remain. x86 processors have multiple modes with different instructions and registers in each mode. AMD decided to make long mode remove some old features and in some cases replace them with new features. If there was a problem with some instruction in previous modes doesn't mean that they had to keep that fault in the new mode, since the new mode only applies to any OS that switches the cpu into that mode and hence knows about the new behaviour. So goodbye to mmx, and various other crap thigns. x87 is supported on current CPUs but there are feature flags that will permit the removal of x87 support in the future. Both linux and windows encourage the use of sse instead of x87 for floating point, and I believe all compilers by default do so in long mode for all current OSs that support amd64, so the removal shouldn't be a problem sometime in the future (although doing so would I suppose prevent running 32bit floating point applications anymore without recompiling those for sse.) Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Lennart Sorensen [EMAIL PROTECTED] writes: On Tue, Jun 20, 2006 at 12:32:18PM +0200, Goswin von Brederlow wrote: Exactly. Isn't the x86_64 instruction set basicaly the same as ia32 just with a few extra opcodes and more registers? Any general fault in the instruction set should still remain. x86 processors have multiple modes with different instructions and registers in each mode. AMD decided to make long mode remove some old features and in some cases replace them with new features. But the changes in the instruction set are minimal. With a bit of care you can use the same inline asm code for ia32 and x86_64 for example. They didn't fix any fundamental flaws in the 386 instruction set. Just droped some later addons like mmx. That is what I ment. MfG Goswin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
On Tue, Jun 20, 2006 at 09:17:08AM -0400, Lennart Sorensen wrote: So goodbye to mmx, and various other crap thigns. x87 is supported on current CPUs but there are feature flags that will permit the removal of x87 support in the future. Both linux and windows encourage the use of sse instead of x87 for floating point, and I believe all compilers by default do so in long mode for all current OSs that support amd64, so the removal shouldn't be a problem sometime in the future (although doing so would I suppose prevent running 32bit floating point applications anymore without recompiling those for sse.) The lack of x87 wouldn't be a problem. At least, i386 still has the code left from the old days and will emulate it transparently; when such an amd64 CPU pops out the x87 emulation can be ported. In fact, it has a separate hardware path from all other types of unknown opcodes (INT7 instead of INT6); x87 was _designed_ to be emulated if not present from the very start. Cheers, -- 1KB // Microsoft corollary to Hanlon's razor: // Never attribute to stupidity what can be // adequately explained by malice. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
On Sun, Jun 18, 2006 at 11:40:03PM +0200, Wouter Verhelst wrote: On Sun, Jun 18, 2006 at 02:24:35PM -0700, Blars Blarson wrote: (amd64 is only faster in 64-bit mode because of all the poorly designed x86 32-bit instruction set.) x86 32-bit instruction set and designed in one sentence? Hah. How about the fact it has more registers available in 64bit mode. People always said the x86 didn't have enough registers after all. Using sse for floating point rather than the awful stack based x87 probably helps too. Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
On Sun, Jun 18, 2006 at 02:24:35PM -0700, Blars Blarson wrote: (amd64 is only faster in 64-bit mode because of all the poorly designed x86 32-bit instruction set.) x86 32-bit instruction set and designed in one sentence? Hah. -- Fun will now commence -- Seven Of Nine, Ashes to Ashes, stardate 53679.4 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Thomas Bushnell BSG wrote: For those keeping score at home, notice that doing the much larger task of upgrading GCC's version is apparently easier than python's. Of course, if you read a little further in the message to -announce, you'll see the part about oh, and maybe we broke 1600 packages, too. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
* Anthony DeRobertis [EMAIL PROTECTED] [2006-06-07 00:07]: For those keeping score at home, notice that doing the much larger task of upgrading GCC's version is apparently easier than python's. Of course, if you read a little further in the message to -announce, you'll see the part about oh, and maybe we broke 1600 packages, too. No, those warnings were also there with gcc 4.0 and indicate potentially buggy *packages*. -- Martin Michlmayr http://www.cyrius.com/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
This one time, at band camp, Anthony DeRobertis said: Thomas Bushnell BSG wrote: For those keeping score at home, notice that doing the much larger task of upgrading GCC's version is apparently easier than python's. Of course, if you read a little further in the message to -announce, you'll see the part about oh, and maybe we broke 1600 packages, too. Full ACK. And it's also worth noting that the people interested in the transition filed bugs with patches for all but a few of them, which involved a massive amount of work. I suppose it's possible people interested in the python transition could do the same instead of complaining, but that seems a bit much to ask, sadly. -- - | ,''`.Stephen Gran | | : :' :[EMAIL PROTECTED] | | `. `'Debian user, admin, and developer | |`- http://www.debian.org | - signature.asc Description: Digital signature
Re: GCC 4.1 now the default GCC version for etch
* Thomas Bushnell BSG [Tue, 06 Jun 2006 19:39:57 -0700]: For those keeping score at home, notice that doing the much larger task of upgrading GCC's version is apparently easier than python's. And shutting the^W^Wbitting one's tongue apparently more difficult than not! -- Adeodato Simó dato at net.com.org.es Debian Developer adeodato at debian.org Listening to: Rosa León - A mi burro -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Stephen Gran [EMAIL PROTECTED] writes: And it's also worth noting that the people interested in the transition filed bugs with patches for all but a few of them, which involved a massive amount of work. I suppose it's possible people interested in the python transition could do the same instead of complaining, but that seems a bit much to ask, sadly. Actually, since the very same person sent the GCC 4.1 announcement, and is the maintainer of python-defaults... -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Adeodato Simó [EMAIL PROTECTED] writes: * Thomas Bushnell BSG [Tue, 06 Jun 2006 19:39:57 -0700]: For those keeping score at home, notice that doing the much larger task of upgrading GCC's version is apparently easier than python's. And shutting the^W^Wbitting one's tongue apparently more difficult than not! Since the python team have decided to leave the rest of us in the dark, all I can do is guess that the job of upgrading python is impossibly difficult, and the people involved are embarassed at that fact.
Re: GCC 4.1 now the default GCC version for etch
This one time, at band camp, Thomas Bushnell BSG said: Actually, since the very same person sent the GCC 4.1 announcement, and is the maintainer of python-defaults... Since the python team have decided to leave the rest of us in the dark, all I can do is guess that the job of upgrading python is impossibly difficult, and the people involved are embarassed at that fact. I don't remember seeing any posts from you about the progress of the transition testing. Have you gotten very far with it? How much of the python using archive have you rebuilt and/or tested? I am sure python people and the general devel readers would be interested in the results of your efforts. Thanks for your hard work on this. After all, just rolling up your sleeves and getting on with it instead of complaing is what makes the free software community great, isn't it? -- - | ,''`.Stephen Gran | | : :' :[EMAIL PROTECTED] | | `. `'Debian user, admin, and developer | |`- http://www.debian.org | - signature.asc Description: Digital signature
Re: GCC 4.1 now the default GCC version for etch
Stephen Gran [EMAIL PROTECTED] writes: This one time, at band camp, Thomas Bushnell BSG said: Actually, since the very same person sent the GCC 4.1 announcement, and is the maintainer of python-defaults... Since the python team have decided to leave the rest of us in the dark, all I can do is guess that the job of upgrading python is impossibly difficult, and the people involved are embarassed at that fact. I don't remember seeing any posts from you about the progress of the transition testing. Have you gotten very far with it? How much of the python using archive have you rebuilt and/or tested? I am sure python people and the general devel readers would be interested in the results of your efforts. I have asked for the status; I have been told nothing. I'm not making any insistence about when the work is done or on what schedule; I have been extremely annoyed that the maintainer has time for everything else but no time to write a two-line answer to a simple question. I'm doing lots of work of my own towards Debian release, but then, when my work is stymied because other developers can't even write two lines and say this is the status... something bad is happening. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Stephen Gran [EMAIL PROTECTED] writes: I don't remember seeing any posts from you about the progress of the transition testing. Have you gotten very far with it? How much of the python using archive have you rebuilt and/or tested? I am sure python people and the general devel readers would be interested in the results of your efforts. How am I to know what work would be helpful when the maintainer won't answer any email? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GCC 4.1 now the default GCC version for etch
Matthias Klose [EMAIL PROTECTED] writes: The compilers from GCC 4.1 provide now the default compiler for etch for Ada, C, C++, Objc, ObjC++, Fortran95 and for the Java language. The packages are currently in the incoming queue and will hit the archive on Wed. June 7. Compilers for Fortran77 and Pascal still default to the compilers built from GCC 3.4. Although we did not get feedback from all port maintainers, we did switch the compiler versions to 4.1 for all architectures [2]. For those keeping score at home, notice that doing the much larger task of upgrading GCC's version is apparently easier than python's. Thomas -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
