Re: [Cooker] Optimizing for Athlon/Duron?
civileme wrote: > If you can find a 386 to run > KDE/Netscape/StarOffice, I'll eat it without mayonnaise. Are you prepared to eat a 486SLC40? That's really only a souped-up '386, not a '486 (no floating point or anything). I can plug a 32MB SIMM (not DIMM) into my gateway machine, and it will indeed run KDE, Netscape and StarOffice. I don't have one going, but I'd be delighted to fire up a 386DX40 motherboard which I have in storage, to prove the point... and I'll even allow you the mayonnaise... (-: -- Police in Radnor, Pennsylvania, interrogated a suspect by placing a metal colander on his head and connecting it with wires to a photocopy machine. The message "He's lying" was placed in the copier, and police pressed the copy button each time they thought the suspect wasn't telling the truth. Believing the "lie detector" was working, the suspect confessed.
Re: [Cooker] Optimizing for Athlon/Duron?
6 ??? 2000 16:48, civileme ???: | On Tuesday 05 December 2000 19:44, you wrote: | Moreover, some experiments showed optimizing for the 686-class machines | SLOWED the product. We probably need better optimizing compilers, | designed for optimizing AMDs, than are available today to make any real it's a pity... Good processors and bad compilers... Almost MS story :-) | gains. Finally, 686-class code tightens the timing requirements on IDE | beyond what we have now, and there's way too much sloppy hardware out | there. We have people who could boot 7.0 and 7.1 who can't touch 7.2 cause | the kernel was changed to accommodate ATA/100 drives (and works fine with Do you have any specific recommendations for selecting motherboard (with IDE controller built-in) or IDE RAID controller - to get it working in ATA/100 mode? Are IBM ATA/100 disks ok for 7.2 kernel? (may be, you can share with us what hardware you tested and how fast it was working :-) | 386 code) but the leeway or slop allowed from stated specification | tolerances for IDE hardware are very tiny with 586 code... Meaning that | bad hardware doesn't boot _because_ we optimize, and many users say | mandrake sucks because it won't boot on their hardware. yes, it's really a problem. Sometimes "optimizations" break code (KDE 2.0.1 was not compiling for i686 for me; at least kdebase) What about to have 2 versions of key packages in distribution, i586 and i686 (glibc, XFree, Apache, KDE2 )? You don't need to optimize Emacs or Traceroute for i686, but KDE and Apache will benefit, IMHO. As well as XFree86. | Of course, you have been added to the list of potential volunteer | testers, for when testing becomes practical. .-) | | Civileme Add me to these volunteers as well ;-)) -- Vadim Plessky http://kde2.newmail.ru (English) http://kde2.newmail.ru/index_rus.html (Russian) Do you have Arial font installed? Just test it! http://kde2.newmail.ru/font_test_arial.html
Re: [Cooker] Optimizing for Athlon/Duron?
On Tuesday 05 December 2000 19:44, you wrote: > Hello, > > One of the things that I like most about Linux-Mandrake is that it is > optimized for Pentium-class machines. I find that it gives a real speedup. > However, after watching the latest Pentium4/FlasK Mpeg debacle on Tom's > Hardware, it occurred to me that perhaps an Athlon/Duron optimized build > would be nice also. There seems to be some precedent in that a > 386-compiled version is usually released a little while after the > pentium-compiled release. Are there any plans to do this? Ummm, that's a 486 compiled version. If you can find a 386 to run KDE/Netscape/StarOffice, I'll eat it without mayonnaise. > > VanL Umm, almost exactly the same instruction set. Moreover, some experiments showed optimizing for the 686-class machines SLOWED the product. We probably need better optimizing compilers, designed for optimizing AMDs, than are available today to make any real gains. Finally, 686-class code tightens the timing requirements on IDE beyond what we have now, and there's way too much sloppy hardware out there. We have people who could boot 7.0 and 7.1 who can't touch 7.2 cause the kernel was changed to accommodate ATA/100 drives (and works fine with 386 code) but the leeway or slop allowed from stated specification tolerances for IDE hardware are very tiny with 586 code... Meaning that bad hardware doesn't boot _because_ we optimize, and many users say mandrake sucks because it won't boot on their hardware. Of course, you have been added to the list of potential volunteer testers, for when testing becomes practical. .-) Civileme
[Cooker] Optimizing for Athlon/Duron?
Hello, One of the things that I like most about Linux-Mandrake is that it is optimized for Pentium-class machines. I find that it gives a real speedup. However, after watching the latest Pentium4/FlasK Mpeg debacle on Tom's Hardware, it occurred to me that perhaps an Athlon/Duron optimized build would be nice also. There seems to be some precedent in that a 386-compiled version is usually released a little while after the pentium-compiled release. Are there any plans to do this? VanL