Re: [Fwd: Re: [rfc] improving 32bit user performance/experience...]
Mackenzie Morgan wrote: > On Tuesday 19 May 2009 9:35:28 pm Christopher Chan wrote: > >> Pentium the Original. Pentium II, III are all based on the Pentium Pro >> (i686) and they have a different architecture than Pentium the Original. >> Pentium II is basically the Pentium Pro + MMX. Pentium 4 and M use the >> Netburst architecture which is again different from the Original. >> > > Oh, wow, ancient then, yeah. I've got a Pentium 2 here. The most RAM it > can take is 384MB. That *might* be enough RAM to run Xubuntu (mine > has 192MB and technically GNOME can run, if it has about 1GB of swap, > but then no other apps can). A motherboard from the days of the > original Pentium would max out at...what? 128MB? > > Higher end ones can go up to 768MB. The Pentium uses Socket 7 and that is used by many others like the AMD K6-2, AMD K6-III, Cyrix..all these preferring i486 over i586 specific binaries and of course i686 over the rest but with gotchas with respect to MMX or 3DNow optimization/support. -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [Fwd: Re: [rfc] improving 32bit user performance/experience...]
On Tuesday 19 May 2009 9:35:28 pm Christopher Chan wrote: > Pentium the Original. Pentium II, III are all based on the Pentium Pro > (i686) and they have a different architecture than Pentium the Original. > Pentium II is basically the Pentium Pro + MMX. Pentium 4 and M use the > Netburst architecture which is again different from the Original. Oh, wow, ancient then, yeah. I've got a Pentium 2 here. The most RAM it can take is 384MB. That *might* be enough RAM to run Xubuntu (mine has 192MB and technically GNOME can run, if it has about 1GB of swap, but then no other apps can). A motherboard from the days of the original Pentium would max out at...what? 128MB? -- Mackenzie Morgan http://ubuntulinuxtipstricks.blogspot.com apt-get moo signature.asc Description: This is a digitally signed message part. -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [Fwd: Re: [rfc] improving 32bit user performance/experience...]
Mackenzie Morgan wrote: > I have no idea how Christopher sent his message, but KMail claims > there's nothing to quote O_o > :-D I did not hit reply all and then sent it to sounder by mistake. Then I forwarded that back here. My apologies. > Anyway... > He said, "i586 binaries should be only installed on actual Pentium > computers." > > Does that mean Pentium The Original, or does it include Pentium 2, 3, 4 > and M? > > Pentium the Original. Pentium II, III are all based on the Pentium Pro (i686) and they have a different architecture than Pentium the Original. Pentium II is basically the Pentium Pro + MMX. Pentium 4 and M use the Netburst architecture which is again different from the Original. -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [Fwd: Re: [rfc] improving 32bit user performance/experience...]
I have no idea how Christopher sent his message, but KMail claims there's nothing to quote O_o Anyway... He said, "i586 binaries should be only installed on actual Pentium computers." Does that mean Pentium The Original, or does it include Pentium 2, 3, 4 and M? -- Mackenzie Morgan http://ubuntulinuxtipstricks.blogspot.com apt-get moo signature.asc Description: This is a digitally signed message part. -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
[Fwd: Re: [rfc] improving 32bit user performance/experience...]
--- Begin Message --- > I challenge anyone to find someone using Ubuntu 8.10/9.04 on a > processor which doesn't support the full i586 instruction set (eg > i386/i486 or something with incomplete i586 support). > i586 binaries should be only installed on actual Pentium computers. Every other subsequent cpu's architecture all handle i486 or i686 and above better than they would handle i586 tuned binaries. The Mandrake business of releasing i586 is plain dumb and only relevant to actual Pentium processors. > All older VIA processors, AMD Geode procs and so on support the full > i586 instruction set, which including MMX instructions and registers, > which itself can provide a good win. > You might find that i586 specific optimization will actually be slower on those processors than i486 optimization. Have you done a test at all? > Also, even if 1% of the users use i586, we can allow instruction > scheduling for deeper pipelines (with eg -mtune=i686) for the other > 99%, generating fewer stalls on more modern processors - important > with far higher core vs memory latency. This gains more for i686 than > it loses for i586. > > Pentiums are so rare. I still have one but I sure am not going waste my time getting Pentium specific tuning. > > I think we have opportunity; Gentoo users tackled this problem in > their way 10 years ago. > I wonder how many had actual Pentiums and how many months it took to compile. -- sounder mailing list soun...@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/sounder --- End Message --- -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [rfc] improving 32bit user performance/experience...
Daniel J Blueman wrote: > All older VIA processors, AMD Geode procs and so on support the full > i586 instruction set, which including MMX instructions and registers, > which itself can provide a good win. > Geodes have partial implementation, particularly they only handle a few PREFETCH instructions IIRC. >>> (does any of this apply to x86-64, eg -mtune=core2 or k8?) >> Yes but this becomes a mess. Leave it as is. gcc is good at tuning >> to general-purpose in an instruction set. > > -mtune is instruction-set invariant. gcc will tune for for i386 > scheduling, ie fewer pipeline stages. It's later processors that have > had to optimise for short-pipeline-scheduled code, not the converse. > IA64 would have been crippled if the instruction scheduling wasn't > right from the outset. > Why the hell would you tune 64-bit instructions for a 32-bit processor? x86-64 has as much to do with IA-32 (i386, x86, i686, etc) as PPC has to do with SPARC. -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [rfc] improving 32bit user performance/experience...
On Tue, May 19, 2009 at 1:21 AM, John Moser wrote: > On Mon, May 18, 2009 at 7:24 PM, Daniel J Blueman > wrote: [snip] > > Even if we split up Ubuntu in i486 and i686, i686 gets its most major > gains from the CMOV instruction family-- a conditional MOV instruction > that acts as a branch-and-mov in one (compare, then either conditional > branch and mov or use cmov). See discussion here: > > http://ondioline.org/mail/cmov-a-bad-idea-on-out-of-order-cpus > > Its most major gain is a wash. Also there's no guarantee any CPU > supports CMOV (it's an option), and thus to guarantee 100% > compatibility we'd have to add a kernel level illegal instruction > handler that handles the CMOV instruction family rather than throwing > a SIGILL at the process (yes this is doable), which is very slow. > > Mind you I'm not against abandoning anything below i686 on desktops > eventually, but some embedded systems will need i586 and the like. > Cost-benefit analysis here. I was trying to raise a more general point about the minimum spec across the board, including the embedded and old-server hardware. I challenge anyone to find someone using Ubuntu 8.10/9.04 on a processor which doesn't support the full i586 instruction set (eg i386/i486 or something with incomplete i586 support). All older VIA processors, AMD Geode procs and so on support the full i586 instruction set, which including MMX instructions and registers, which itself can provide a good win. Also, even if 1% of the users use i586, we can allow instruction scheduling for deeper pipelines (with eg -mtune=i686) for the other 99%, generating fewer stalls on more modern processors - important with far higher core vs memory latency. This gains more for i686 than it loses for i586. >> (does any of this apply to x86-64, eg -mtune=core2 or k8?) > > Yes but this becomes a mess. Leave it as is. gcc is good at tuning > to general-purpose in an instruction set. -mtune is instruction-set invariant. gcc will tune for for i386 scheduling, ie fewer pipeline stages. It's later processors that have had to optimise for short-pipeline-scheduled code, not the converse. IA64 would have been crippled if the instruction scheduling wasn't right from the outset. I think we have opportunity; Gentoo users tackled this problem in their way 10 years ago. -- Daniel J Blueman -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [rfc] improving 32bit user performance/experience...
2009/5/19 Alexandre Strube : >> Even though I'm performance freak I will be staying on my 32bit >> dual-core laptop for quite a while still. >> >> (I have access to Sparc 64bit grid ;-) to run my simulations on) > > Hello Dima, > I guess the issue here was deprecating the -386 in favor of taking advantage > of -586(and beyond) instructions, even in 32 bit, which would give 32-bit > ubuntu a slight boost. > However, there are people who opposes this, as there are CPUs that > doesn't necessarily support those, besides of people who uses 386s... (Which > is hard to understand why would someone use ubuntu on it). > [] > Alexandre Strube > su...@ubuntu.com > I wasn't that well informed. i've read up on the issue. Lets now consider following when: On desktop should be gone by now, but formally Intel slashed production of 386 in 2007 (according to wikipedia). Dapper is still supported until 2011 and I see a lot of 386's still staying around as slow, small "home" servers. who: The cange will negarivly affect old PC users, old PC server's and embedded platforms. In addition did anyone confirm 586 instructions are fine with Via & AMD (i hope they are fine on modern Intel chips)? what: Rebuild the whole archieve.. Did anyone try that with i586? I would like to see a rebuild of the archive from scratch. I bet there will be quite a few FTBS. what else: This change will probably result in a new architecture in the archive because of Old PC's, Old servers and Embedded systems which are still i386. On the other hand we can be all hype and force Old PC's, Old servers and Embeddeb systems to use Debian = In conclusion I'm against the change until I see a full rebuild of the archive and some *significant* real-life figures why whould I switch from 386 to 586. -- With best regards Dmitrijs Ledkovs (for short Dima), Ледков Дмитрий Юрьевич -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [rfc] improving 32bit user performance/experience...
2009/5/19 Markus Hitter : > > Am 19.05.2009 um 01:24 schrieb Daniel J Blueman: > >> A number of benchmarks show a significant performance loss on 32bit >> ubuntu over 64bit [...] >> Just how much user experience do we trade away for i386/i486 legacy >> compatibility these days? > > IMHO, you draw an odd conclusion here. You recognize AMD64 to be > faster than i386 and take this argument to turn away some of the i386 > users? If you are so keen on performance, by all means, install AMD64. > > > Markus > > - - - - - - - - - - - - - - - - - - - > Dipl. Ing. Markus Hitter > http://www.jump-ing.de/ > Even though I'm performance freak I will be staying on my 32bit dual-core laptop for quite a while still. (I have access to Sparc 64bit grid ;-) to run my simulations on) -- With best regards Dmitrijs Ledkovs (for short Dima), Ледков Дмитрий Юрьевич -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
Re: [rfc] improving 32bit user performance/experience...
Am 19.05.2009 um 01:24 schrieb Daniel J Blueman: > A number of benchmarks show a significant performance loss on 32bit > ubuntu over 64bit [...] > Just how much user experience do we trade away for i386/i486 legacy > compatibility these days? IMHO, you draw an odd conclusion here. You recognize AMD64 to be faster than i386 and take this argument to turn away some of the i386 users? If you are so keen on performance, by all means, install AMD64. Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss