Re: -O2 considered harmful
On Fri, Feb 28, 2003 at 05:52:59PM -0800, Terry Lambert wrote: > > > > I have seen this as well, using -O2 -march=athlon-xp. > > > > The generated assembler tried to stuff -129 into a single byte. > > > > > > What about just trying -march=athlon? The only difference is the SSE > > > support, which is quite new and may have latent bugs anyway. > > > > I didn't try that, but i did not get that error using: > > -O2 -march=athlon-xp -mmmx -msse -m3dnow -mfpmath=sse > > to compile the xc/ tree > > > Is it just me, or does this whole discussion boil down to something > like "How can I overclock my compiler?"? > I'd say it's more something like "How can I get my money's worth out of my processor?" With modern processors having multiple functional units, capable of queuing and executing multiple instructions simultaneously, we need some way to take advantage of this - so that we use 100% of the CPU instead of 75%. So, for example using the new SSE functional unit instead of the aged hacked 387 makes sense, hence the -mfpmath=sse flag. I have just recompiled the world and all my ports using -O2 -march=athlon-xp -mmmx -msse -m3dnow -mfpmath=sse and everything except ezm3 compiled cleanly (adding -fno-schedule-insns2 to the flags made it compile). Mozilla doesn't crash any more. Bruce Cran To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
> > > I have seen this as well, using -O2 -march=athlon-xp. > > > The generated assembler tried to stuff -129 into a single byte. > > > > What about just trying -march=athlon? The only difference is the SSE > > support, which is quite new and may have latent bugs anyway. > > I didn't try that, but i did not get that error using: > -O2 -march=athlon-xp -mmmx -msse -m3dnow -mfpmath=sse > to compile the xc/ tree Is it just me, or does this whole discussion boil down to something like "How can I overclock my compiler?"? -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Fri, 28 Feb 2003, David O'Brien wrote: > On Wed, Feb 26, 2003 at 05:31:20PM -0500, Bryan Liesner wrote: > > I have seen this as well, using -O2 -march=athlon-xp. > > The generated assembler tried to stuff -129 into a single byte. > > What about just trying -march=athlon? The only difference is the SSE > support, which is quite new and may have latent bugs anyway. I didn't try that, but i did not get that error using: -O2 -march=athlon-xp -mmmx -msse -m3dnow -mfpmath=sse to compile the xc/ tree I know that some of the flags above are redundant, but this worked and X is running without any crashes or wierdness. (so far). I also applied all of the above compile flags when building xscreensaver and saw a very noticeable speed improvement in some of the hacks. -- == = Bryan D. Liesner LeezSoft Communications, Inc. = = A subsidiary of LeezSoft Inc. = = [EMAIL PROTECTED]Home of the Gipper= == To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, Feb 26, 2003 at 05:31:20PM -0500, Bryan Liesner wrote: > I have seen this as well, using -O2 -march=athlon-xp. > The generated assembler tried to stuff -129 into a single byte. What about just trying -march=athlon? The only difference is the SSE support, which is quite new and may have latent bugs anyway. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
Bruce Cran wrote: > I'm afraid you're wrong - the V2SI datatype and MMX functions > automatically become available after -march=pentium2, while > with other processor types you've got to explicitly add -mmmx. > -msse is presumed with -march=pentium3 and up. I'm afraid I'm not; I was talking strictly about -march=pentiumpro, not -march=pentium2 et cetera. > ... I've seen code which runs 40x faster when compiled for > athlon-xp than for i386, and I would guess that a lot of that > is because of clever use of sse and mmx. That wasn't an > audio/video program, it was the libgmp arbitrary precision maths > package. MMX won't help you even a bit in precise mathematical calculations, unlike SSE. Also, if you're familiar with the way how MMX works, you won't enable it when compiling any application that requires heavy floating-point computations, because constant switching between MMX unit and FPU results in significant latency. > Also, I'm sure most people wouldn't say no to 50% more processing > speed for free! I didn't mean to stop using MMX at all; I only intended to note that there is no need to enable it for _all_ applications. > So, if you've got a pentium, k6 or pentiumpro which supports MMX, > you _do_ need to explicitly add -mmmx, but for other processors > it's implied. You didn't pay enough attention to my original post; PPro doesn't support MMX. Though, you can specify -mmmx anyway, if you feel comfortable with SIGILLs. --- Regards, Rhett __ Do you Yahoo!? Yahoo! Tax Center - forms, calculators, tips, more http://taxes.yahoo.com/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Thu, Feb 27, 2003 at 09:49:13PM +, Nuno Teixeira wrote:
> On Thu, Feb 27, 2003 at 08:38:00AM +, Bruce Cran wrote:
> > I'm afraid you're wrong - the V2SI datatype and MMX functions automatically
> > become available after -march=pentium2, while with other processor types
> > you've got to explicitly add -mmmx. -msse is presumed with -march=pentium3
> > and up. It's far from absurd to use mmx for everyday applications - sure,
> > only a few applications may take advantage of it, but I've seen code which
> > runs 40x faster when compiled for athlon-xp than for i386, and I would guess
> > that a lot of that is because of clever use of sse and mmx. That wasn't
> > an audio/video program, it was the libgmp arbitrary precision maths
> > package. Also, I'm sure
> > most people wouldn't say no to 50% more processing speed for free!
> > So, if you've got a pentium, k6 or pentiumpro which supports MMX, you _do_
> > need to explicitly add -mmmx, but for other processors it's implied.
> >
> I searched gcc docs and didn't found info for what you say here. I'm
> seeing a lots of people using e.g. athlon-xp with -mmmx and -m3dnow
> included. I'm confused about if this optimizations are implied or not by
> processores that supports it.
>
I, too, use -mmmx -msse -m3dnow with -march=athlon-xp. I do it simply
because I don't trust gcc enough to do it for me - people have shown in
the past that -O2 and -O3 don't activate all the optimizations which
the docs claim they should, which is why you see people adding crazy stuff
like -funroll-loops -fomit-frame-pointer -fschedule-insns2 -fgcse ...
The surest way to find out about at which point gcc enables vector
extensions is, if you've got access to a suitable computer, compile the
following:
typedef int v4sf __attribute__ ((mode(V4SF)));
int main()
{
v4sf a = {1,2,3,4};
v4sf b = {5,6,7,8};
v4sf c = __builtin_ia32_addps(a,b);
return 0;
}
This will only compile when gcc has enabled sse instruction support. I've
found that this happens when you use -msse on it's own, even with
-march=pentium, and when you use -march=pentium3, -march=pentium4,
-march=athlon-xp etc without any extra -msse. In addition, when compiling
mmx code, -m3dnow implies -mmmx, which makes sense since 3dnow is just an
extension of mmx.
Of course what many people don't realise is that gcc, unlike icc, will not
produce any vector instructions unless either the -mfpunit=sse is enabled to
use sse for all floating point math, or vector instructions are
explicitly coded for, as above. So for most software, adding the extra flags
shouldn't affect it in the slightest, but for a few applications, it will
detect the vector unit and use it, resulting in a sometimes significant
performance gain. You should notice a fairly large increase in performance
when using the sse unit, because unlike mmx and 3dnow it is a seperate
functional unit and so was designed to be fast, instead of being crippled
with backward compatability with the 387. Indeed, with sse2, Intel seem
to have finally gained a very decent vector processing unit which can
compete with similar processors such as the G4 with its AltiVec.
Bruce Cran
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Thu, Feb 27, 2003 at 08:38:00AM +, Bruce Cran wrote: > On Wed, Feb 26, 2003 at 06:51:55PM -0800, Rhett Monteg Hollander wrote: > > Nuno Teixeira wrote: > > > > > > Hi, > > > > > > Just a little question: > > > > > > Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? > > Pentium Pro doesn't support MMX; -march=pentiumpro (aka -march=i686) enables > > compiling with main i686 instruction set, no MMX\SSE or whatever. > > > > > > > > I always thought that when I use -march it will enable other > > > porcessor specific optimizations like mmx and 3dnow (if available). > > No way. Besides, only a dozen of open-source UNIX applications can really > > profit from MMX, so it's absurd to use -mmmx for day-by-day use. To be correct, > > MMX is only useful for real-time video\audio processing. > > > > A little example: consider some 16-bit pixel set that you need to increase > > brightness, i.e. to add some value to every pixel. You can load a word into > > integer register, process, and store. You can also load a quadword of four > > 16-bit values into MMX register which is mapped onto FPU register, process them > > at once, and store. Indeed to be faster. > > > > Enabling MMX usually results in about 50% performance increase for true CISC > > processors like P-MMX, and about 20% for PII and up. > > > > I'm afraid you're wrong - the V2SI datatype and MMX functions automatically > become available after -march=pentium2, while with other processor types > you've got to explicitly add -mmmx. -msse is presumed with -march=pentium3 > and up. It's far from absurd to use mmx for everyday applications - sure, > only a few applications may take advantage of it, but I've seen code which > runs 40x faster when compiled for athlon-xp than for i386, and I would guess > that a lot of that is because of clever use of sse and mmx. That wasn't > an audio/video program, it was the libgmp arbitrary precision maths > package. Also, I'm sure > most people wouldn't say no to 50% more processing speed for free! > So, if you've got a pentium, k6 or pentiumpro which supports MMX, you _do_ > need to explicitly add -mmmx, but for other processors it's implied. > > Bruce Cran > > To Unsubscribe: send mail to [EMAIL PROTECTED] > with "unsubscribe freebsd-current" in the body of the message Hi, I searched gcc docs and didn't found info for what you say here. I'm seeing a lots of people using e.g. athlon-xp with -mmmx and -m3dnow included. I'm confused about if this optimizations are implied or not by processores that supports it. Thanks very much, Nuno Teixeira -- /* PGP fingerprint: C6D1 06ED EB54 A99C 6B14 6732 0A5D 810D 727D F6C6 */ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Thu, 27 Feb 2003, Nuno Teixeira wrote: > On Wed, Feb 26, 2003 at 06:51:55PM -0800, Rhett Monteg Hollander wrote: > > I understand it. Anyway, there is known problems by using -mmmx or > -m3dnow on builworld/buildkernel? Well, I used allways march=pentiumpro > on stable and now pentium2 on current without any problem. I've been using -march=athlon-xp -O2 -mmmx -m3dnow and -msse on my desktop for close to a month without any noticeable quirks. I rebuild world twice daily and kernel every other day... > I use some applications like grip, lame, mplayer, etc. Maybe I get some > extra performance running this apps. With -mmmx, -m3dnow or -msse, you're going to get a noticeable improvement, especially with lame and mplayer. I recommend compiling programs that use sound or graphics with high optimization cflags and keeping moderately conservative options for the base system and kernel. Regards, > Andre Guibert de Bruet | Enterprise Software Consultant > > Silicon Landmark, LLC. | http://siliconlandmark.com/> To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, Feb 26, 2003 at 06:51:55PM -0800, Rhett Monteg Hollander wrote: > Nuno Teixeira wrote: > > > > Hi, > > > > Just a little question: > > > > Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? > Pentium Pro doesn't support MMX; -march=pentiumpro (aka -march=i686) enables > compiling with main i686 instruction set, no MMX\SSE or whatever. > > > > > I always thought that when I use -march it will enable other > > porcessor specific optimizations like mmx and 3dnow (if available). > No way. Besides, only a dozen of open-source UNIX applications can really > profit from MMX, so it's absurd to use -mmmx for day-by-day use. To be correct, > MMX is only useful for real-time video\audio processing. > > A little example: consider some 16-bit pixel set that you need to increase > brightness, i.e. to add some value to every pixel. You can load a word into > integer register, process, and store. You can also load a quadword of four > 16-bit values into MMX register which is mapped onto FPU register, process them > at once, and store. Indeed to be faster. > > Enabling MMX usually results in about 50% performance increase for true CISC > processors like P-MMX, and about 20% for PII and up. > > --- > Regards, > Rhett > Hi, I understand it. Anyway, there is known problems by using -mmmx or -m3dnow on builworld/buildkernel? Well, I used allways march=pentiumpro on stable and now pentium2 on current without any problem. I use some applications like grip, lame, mplayer, etc. Maybe I get some extra performance running this apps. Thanks, Nuno Teixeira -- /* PGP fingerprint: C6D1 06ED EB54 A99C 6B14 6732 0A5D 810D 727D F6C6 */ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Thu, Feb 27, 2003 at 05:49:47AM -0500, Geoffrey wrote: > This is fine and good, but make.conf appears to be hiding in 5.0 > (or at least in the various installs/cvsups I've encountered to date). > What flags are accepted is a bit of a guessing game without a template in > /etc/defaults (yes, I could wade through man pages but with the rate of > change on release, those recommendations aren't a given either). > Have you looked at the make.conf in /usr/share/examples/etc ? -- Morten Rodal pgp0.pgp Description: PGP signature
Re: -O2 considered harmful
On Thu, 27 Feb 2003, Bruce Cran wrote: > On Wed, Feb 26, 2003 at 06:51:55PM -0800, Rhett Monteg Hollander wrote: > > I'm afraid you're wrong - the V2SI datatype and MMX functions automatically > become available after -march=pentium2, while with other processor types > you've got to explicitly add -mmmx. -msse is presumed with -march=pentium3 > and up. It's far from absurd to use mmx for everyday applications - sure, > only a few applications may take advantage of it, but I've seen code which > runs 40x faster when compiled for athlon-xp than for i386, and I would guess > that a lot of that is because of clever use of sse and mmx. That wasn't > an audio/video program, it was the libgmp arbitrary precision maths > package. Also, I'm sure > most people wouldn't say no to 50% more processing speed for free! > So, if you've got a pentium, k6 or pentiumpro which supports MMX, you _do_ > need to explicitly add -mmmx, but for other processors it's implied. > This is fine and good, but make.conf appears to be hiding in 5.0 (or at least in the various installs/cvsups I've encountered to date). What flags are accepted is a bit of a guessing game without a template in /etc/defaults (yes, I could wade through man pages but with the rate of change on release, those recommendations aren't a given either). Cheers. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, Feb 26, 2003 at 06:51:55PM -0800, Rhett Monteg Hollander wrote: > Nuno Teixeira wrote: > > > > Hi, > > > > Just a little question: > > > > Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? > Pentium Pro doesn't support MMX; -march=pentiumpro (aka -march=i686) enables > compiling with main i686 instruction set, no MMX\SSE or whatever. > > > > > I always thought that when I use -march it will enable other > > porcessor specific optimizations like mmx and 3dnow (if available). > No way. Besides, only a dozen of open-source UNIX applications can really > profit from MMX, so it's absurd to use -mmmx for day-by-day use. To be correct, > MMX is only useful for real-time video\audio processing. > > A little example: consider some 16-bit pixel set that you need to increase > brightness, i.e. to add some value to every pixel. You can load a word into > integer register, process, and store. You can also load a quadword of four > 16-bit values into MMX register which is mapped onto FPU register, process them > at once, and store. Indeed to be faster. > > Enabling MMX usually results in about 50% performance increase for true CISC > processors like P-MMX, and about 20% for PII and up. > I'm afraid you're wrong - the V2SI datatype and MMX functions automatically become available after -march=pentium2, while with other processor types you've got to explicitly add -mmmx. -msse is presumed with -march=pentium3 and up. It's far from absurd to use mmx for everyday applications - sure, only a few applications may take advantage of it, but I've seen code which runs 40x faster when compiled for athlon-xp than for i386, and I would guess that a lot of that is because of clever use of sse and mmx. That wasn't an audio/video program, it was the libgmp arbitrary precision maths package. Also, I'm sure most people wouldn't say no to 50% more processing speed for free! So, if you've got a pentium, k6 or pentiumpro which supports MMX, you _do_ need to explicitly add -mmmx, but for other processors it's implied. Bruce Cran To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
Nuno Teixeira wrote: > > Hi, > > Just a little question: > > Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? Pentium Pro doesn't support MMX; -march=pentiumpro (aka -march=i686) enables compiling with main i686 instruction set, no MMX\SSE or whatever. > > I always thought that when I use -march it will enable other > porcessor specific optimizations like mmx and 3dnow (if available). No way. Besides, only a dozen of open-source UNIX applications can really profit from MMX, so it's absurd to use -mmmx for day-by-day use. To be correct, MMX is only useful for real-time video\audio processing. A little example: consider some 16-bit pixel set that you need to increase brightness, i.e. to add some value to every pixel. You can load a word into integer register, process, and store. You can also load a quadword of four 16-bit values into MMX register which is mapped onto FPU register, process them at once, and store. Indeed to be faster. Enabling MMX usually results in about 50% performance increase for true CISC processors like P-MMX, and about 20% for PII and up. --- Regards, Rhett __ Do you Yahoo!? Yahoo! Tax Center - forms, calculators, tips, more http://taxes.yahoo.com/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
Andre Guibert de Bruet wrote: On Wed, 26 Feb 2003, Nuno Teixeira wrote: Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? I always thought that when I use -march it will enable other porcessor specific optimizations like mmx and 3dnow (if available). IIRC, 166 and 180 Mhz PPros don't support MMX. Neither do PPro's 200: CPU: Pentium Pro (199.43-MHz 686-class CPU) Origin = "GenuineIntel" Id = 0x619 Stepping = 9 Features=0xf9ff I'm seeing differences in compiled code between simple -march=athlon-xp and -march=athlon-xp -mmmx -m3dnow -msse. The documentation says the same: those options have to be enabled to use extended instruction sets. -march/-mcpu just optmizes boundaries of jump labels, variables, etc... - and sometimes the kind of instructions to use, eg. 'loop label' or 'dec bcx, jne label', but that's all, AFAIK. So long, Jens -- L i W W W i Jens Rehsack LW W W L i W W W W i nnnLiWing IT-Services L iW W W Wi n n g g i W W i n n g gFriesenstraße 2 06112 Halle g g g Tel.: +49 - 3 45 - 5 17 05 91ggg e-Mail: <[EMAIL PROTECTED]> Fax: +49 - 3 45 - 5 17 05 92http://www.liwing.de/ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, 26 Feb 2003, Nuno Teixeira wrote: > Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? > > I always thought that when I use -march it will enable other > porcessor specific optimizations like mmx and 3dnow (if available). IIRC, 166 and 180 Mhz PPros don't support MMX. I'm seeing differences in compiled code between simple -march=athlon-xp and -march=athlon-xp -mmmx -m3dnow -msse. Regards, > Andre Guibert de Bruet | Enterprise Software Consultant > > Silicon Landmark, LLC. | http://siliconlandmark.com/> To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, 26 Feb 2003 13:06:54 +0100 Jens Rehsack <[EMAIL PROTECTED]> wrote: > Dag-Erling Smorgrav wrote: > > It seems that with -O2 on ia32 (-march=k6-2 in my case), gcc will in > > some cases generate short jumps to targets too far away for the offset > > to fit in a single byte. A surefire way to reproduce this is to build > > Mesa (or XFree86-4-libraries, which includes parts of Mesa). > > > > Has anybody else run into this? > > > > I build world and ports on desktop machines usually using -O2 and never > got any problems, except with lang/ezm3. If you tell me, where you've > find the problems, I'll tell my gcc to keep the temporaries and check > them after a make build. you need to add "-fno-schedule-insns2" to your CFLAGS to compile ezm3 with -O2... To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, 26 Feb 2003, Dag-Erling Smorgrav wrote: > It seems that with -O2 on ia32 (-march=k6-2 in my case), gcc will in > some cases generate short jumps to targets too far away for the offset > to fit in a single byte. A surefire way to reproduce this is to build > Mesa (or XFree86-4-libraries, which includes parts of Mesa). > > Has anybody else run into this? > > DES > I have seen this as well, using -O2 -march=athlon-xp. The generated assembler tried to stuff -129 into a single byte. -- == = Bryan D. Liesner LeezSoft Communications, Inc. = = A subsidiary of LeezSoft Inc. = = [EMAIL PROTECTED]Home of the Gipper= == To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
Hi, Just a little question: Does -march=k6-2 implies -m3dnow? Or -march=pentiumpro implies -mmmx? I always thought that when I use -march it will enable other porcessor specific optimizations like mmx and 3dnow (if available). Thanks, Nuno Teixeira On Wed, Feb 26, 2003 at 05:06:35PM -0300, Fred Souza wrote: > > I was recently a participant in a thread in another forum where all sorts > > of people, including a well respected gcc developer, said categorically > > that the latest (stock) gcc produces correct code with -O2 in all cases on > > ia32. If it doesn't, the gcc folks would like a bug report. > > I use -O3 -mmmx -m3dnow -march=k6-2 on nearly everything (things that > break with that many optimization, such as XFree86-4-libraries, I > "downgrade" to the highest optimization possible - usually -O2), and > I'm not noticing any bad impacts because of it. > > And by "nearly everything" I even meant the kernel (yes, I can respect > if anyone screams "[EMAIL PROTECTED]" ;-) > > > Fred > > > -- > Insulting a person's appearance: > > That's what they mean by dark and handsome. When it's dark, you're > handsome... -- /* PGP fingerprint: C6D1 06ED EB54 A99C 6B14 6732 0A5D 810D 727D F6C6 */ To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, Feb 26, 2003 at 05:06:35PM -0300, Fred Souza wrote: > > I was recently a participant in a thread in another forum where all sorts > > of people, including a well respected gcc developer, said categorically > > that the latest (stock) gcc produces correct code with -O2 in all cases on > > ia32. If it doesn't, the gcc folks would like a bug report. > > I use -O3 -mmmx -m3dnow -march=k6-2 on nearly everything (things that > break with that many optimization, such as XFree86-4-libraries, I > "downgrade" to the highest optimization possible - usually -O2), and > I'm not noticing any bad impacts because of it. > > And by "nearly everything" I even meant the kernel (yes, I can respect > if anyone screams "[EMAIL PROTECTED]" ;-) > Is it just me, or does mozilla break when things are compiled with -march=athlon-xp? I'm not sure what happens, but mozilla starts to load, a window flashes up briefly before mozilla quits. Running dbg on it says that it segfaults in WindowGetHeight() or some such function. When I recompile _everything_ back using -O -march=pentiumpro it works again. I'm sure I've compiled it for athlon-xp before on 5.0-RELEASE, has something changed to break it? Bruce Cran To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
> I was recently a participant in a thread in another forum where all sorts > of people, including a well respected gcc developer, said categorically > that the latest (stock) gcc produces correct code with -O2 in all cases on > ia32. If it doesn't, the gcc folks would like a bug report. I use -O3 -mmmx -m3dnow -march=k6-2 on nearly everything (things that break with that many optimization, such as XFree86-4-libraries, I "downgrade" to the highest optimization possible - usually -O2), and I'm not noticing any bad impacts because of it. And by "nearly everything" I even meant the kernel (yes, I can respect if anyone screams "[EMAIL PROTECTED]" ;-) Fred -- Insulting a person's appearance: That's what they mean by dark and handsome. When it's dark, you're handsome... pgp0.pgp Description: PGP signature
Re: -O2 considered harmful
On Wed, 26 Feb 2003, Dag-Erling Smorgrav wrote: > It seems that with -O2 on ia32 (-march=k6-2 in my case), gcc will in > some cases generate short jumps to targets too far away for the offset > to fit in a single byte. A surefire way to reproduce this is to build > Mesa (or XFree86-4-libraries, which includes parts of Mesa). I was recently a participant in a thread in another forum where all sorts of people, including a well respected gcc developer, said categorically that the latest (stock) gcc produces correct code with -O2 in all cases on ia32. If it doesn't, the gcc folks would like a bug report. Doug To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
On Wed, 26 Feb 2003, Dag-Erling Smorgrav wrote: > It seems that with -O2 on ia32 (-march=k6-2 in my case), gcc will in > some cases generate short jumps to targets too far away for the offset > to fit in a single byte. A surefire way to reproduce this is to build > Mesa (or XFree86-4-libraries, which includes parts of Mesa). The assembler generates these. Bruce To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: -O2 considered harmful
Dag-Erling Smorgrav wrote: It seems that with -O2 on ia32 (-march=k6-2 in my case), gcc will in some cases generate short jumps to targets too far away for the offset to fit in a single byte. A surefire way to reproduce this is to build Mesa (or XFree86-4-libraries, which includes parts of Mesa). Has anybody else run into this? I build world and ports on desktop machines usually using -O2 and never got any problems, except with lang/ezm3. If you tell me, where you've find the problems, I'll tell my gcc to keep the temporaries and check them after a make build. Jens To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
-O2 considered harmful
It seems that with -O2 on ia32 (-march=k6-2 in my case), gcc will in some cases generate short jumps to targets too far away for the offset to fit in a single byte. A surefire way to reproduce this is to build Mesa (or XFree86-4-libraries, which includes parts of Mesa). Has anybody else run into this? DES -- Dag-Erling Smorgrav - [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
