Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
AMD Software Hardware Linux

64-bit Linux On The Opteron 325

Posted by timothy from the unchecked-envy dept.
JigSaw writes "A few moths ago Robert Minvielle put to test AMD's Opteron regarding its 64-bit Linux compatibility. The results back then were not very positive but he is now back testing more 64-bit updated distros: Gentoo, SuSE, Mandrake, Red Hat and Fedora. And this time the results are more positive with Linux offering good Opteron support where Windows-64 doesn't seem to. FreeBSD also lists the AMD64 platform as a tier-1 architecture."
This discussion has been archived. No new comments can be posted.

64-bit Linux On The Opteron More

64-bit Linux On The Opteron

Comments Filter:

  • Market Share (Score:5, Interesting)

    by Hadur ( 636978 ) on Tuesday December 16, 2003 @08:39PM (#7740922)
    Interesting revelation in the tests: Linux, while not having a great share of the market now, will progressively gain user base simply because it is so capable of evolving with new technology.

    • Re:Market Share (Score:5, Insightful)

      by metallicagoaltender ( 187235 ) on Tuesday December 16, 2003 @08:41PM (#7740945) Homepage
      That conclusion is a bit of a leap of faith - what's to say people will take to Linux just because it adapts better? I'd guess that more people will wait for Windows to provide solid support for new technology than switch OSes just because they can be more cutting edge.

    • What kind of user base, though? A highly-specialised user base with a ton of cash to spare and who are interested in Linux? Linux has a relatively small market share as it is (compared with Windoze), although certainly early adaptors are included in this group. -1 Incoherent, sorry ;)

    • Re:Market Share (Score:5, Informative)

      by NanoGator ( 522640 ) on Tuesday December 16, 2003 @09:12PM (#7741150) Homepage Journal
      "Interesting revelation in the tests: Linux, while not having a great share of the market now, will progressively gain user base simply because it is so capable of evolving with new technology."

      I can see this for customers such as Hollywood. This isn't necesssarily true in the consumer world, however. Too many variables to make that a reliably true or false statement.

      Frankly, I find this statement a bit overrated. Nothing personal, but a little bit of clarification would have sounded less like 'pat-linux-on-the-back-karma-whoring' and more like something informational.
    • Linux, while not having a great share of the market now, will progressively gain user base simply because it is so capable of evolving with new technology.

      Linux will gain user base because it's cheaper, and because some forward-thinking organizations are finally starting to see the benefits of not being chained to Microsoft.

      Technology has nothing to do with it. Case in point: Intel shipped the 80386 in 1985, and it was ten years before Microsoft finally shipped a consumer-grade 32-bit operating system.

  • Well... (Score:5, Funny)

    by i_am_syco ( 694486 ) on Tuesday December 16, 2003 @08:39PM (#7740925)
    I was always preferable to Linux over Windows on 64-bit processors. Of course, I'm talking about the G5...

  • whats the deal (Score:3, Interesting)

    by nil5 ( 538942 ) on Tuesday December 16, 2003 @08:41PM (#7740938) Homepage
    don't really understand why this shouldbe any different than supporting any other architecture. Linux does run other 64-bit architectures, eg64bit sparc and the antiquated among others. it's really just a matter of time before it's perfected.

    big woop. so what else is new?

    • Re:whats the deal (Score:4, Interesting)

      by the_bahua ( 411625 ) on Tuesday December 16, 2003 @08:51PM (#7741006) Homepage Journal
      Yeah, that's kind of my feeling. It's nice that it's working, but it looks like it still isn't quite there yet. Until I hear some better things about 64-bit performance in GNU/Linux with the Opteron, I'm going to stick with my athlon for now.

      It'd be nice if some more practical benchmarks were posted, though, like I/O, database performance/stability, positive effects of the new memory access, etc, instead of, or at least in addition to telling us how well KDE works.

    • Re:whats the deal (Score:5, Informative)

      by wafflemonger ( 515122 ) on Tuesday December 16, 2003 @08:55PM (#7741036)
      It is the difference between porting the kernel and porting a distro. There may be some apps in a distro that do not work well on the new architecture. Also each of the apps has to interact with the others and those combinations can cause problems. There could also be issues in the libraries that cause dependent programs to crash in 64 bit mode. Yes in time it will be perfected, but if there are problems now they need to be smoked out and fixed.
      • The biggest issue for me when I last tried out some 64bit betas was that the 64bit NVidia drivers suck. They don't have an ioctl32 interface, so if you use them you *must* run 64bit 3d apps - which rules out pretty much anything worth running game wise.

        btw. what's the article on about 'new' NVidia drivers? They're dated September 23rd!

    • Re:whats the deal (Score:4, Insightful)

      by iggymanz ( 596061 ) on Wednesday December 17, 2003 @01:08AM (#7742542)
      Actually, if you read the fine print for, say, the UltraLinux port, you'll find that the kernel is 64 bit and userland is still 32 bit. Now the *BSD on the other hand are much cooler in that regard..... To be fair, don't even get me started on companies like Sun and SGI claiming they had 64 bit back in the mid 90's....their OS were 32/64 hybrids in many various and interesting ways also. Even in Solaris 8 there's issues if you run the 64 bit kernel, some commands don't work correctly when one hits the 32 bit limit!

  • Awesome (Score:5, Interesting)

    by bigjnsa500 ( 575392 ) <bigjnsa500@nOSpAM.yahoo.com> on Tuesday December 16, 2003 @08:41PM (#7740941) Homepage Journal
    We went to a Sun demonstration on campus and they showed the new to be released Opteron servers with 1-4 CPUs. Price and performance is very, very good. They come with a SuSE derivitve distro. I couldn't tell if its real SuSE or a SuSE Sun optimized. Anyway, we are going to order a few of them for a BLAST cluster to replace our existing cluster.

  • Windows 64 (Score:3, Interesting)

    by swordboy ( 472941 ) on Tuesday December 16, 2003 @08:41PM (#7740943) Journal
    Windows has a native 64-bit version but Intel have prompted MS to delay the release until they can come up with a competitive processor. AMD serves to steal much of Intel's marketshare otherwise. Useful or not, console wars has caused "64 bits to be better than 32".
    • Well, 64 is better than 32. Just not in the way of faster, nessesarily. I mean, more memory address space, the ability to process 64-bit IDs in one cycle... just is better for those apps written so as to take advantage of it.
      • Er, there are 32-bit processors out there that can address more than 4GB of RAM.

        • Re:Windows 64 (Score:4, Informative)

          by quantum bit ( 225091 ) on Tuesday December 16, 2003 @09:09PM (#7741140) Journal
          Er, there are 32-bit processors out there that can address more than 4GB of RAM.

          Usually via bank switching (e.g. PAE) which is slow and cumbersome.

          • Re:Windows 64 (Score:3, Informative)

            by gbnewby ( 74175 ) *
            64-bit address space means that a PROCESS (loosely equal to a PROGRAM) can access > 4GB. In Linux, processes are limited by 32bit (i.e., 4GB - though in practice a process can usually not get quite that much).

            So, the "big deal" about 64-bit is that (a) there will be *direct* access to the memory beyond 4GB, as the previous message mentioned; and that (b) individual processes will be able to access > 4GB.

            (a) provides a performance boost, by removing the need for "mapping" between 32-bit and 64-bit

    • Re:Windows 64 (Score:5, Funny)

      by zulux ( 112259 ) on Tuesday December 16, 2003 @08:54PM (#7741031) Homepage Journal
      Windows has a native 64-bit version but Intel have prompted MS to delay the release until they can come up with a competitive processor.

      Actually, the delay in Windows64 was trying to come up with a new prefix for all the hungarian-notaion in their code.

      I think they settled on the following for a 64 bit pinter to a string: pllpsexsfeString
      • My kingdom for mod points!

        'jfb

      • S mi a baj magyarral? Szerintem egy nagyon szep nyelv, de lehet hogy egy kicsit eloiteletes vagyok ebben az esetben...

        (What's wrong with Hungarian? I think it's a very beautiful language, but I may be a little prejudiced in this case...)

        MT

        BTW, there should be accented and tilded characters in the above sentence, but I don't know how to submit them using the Slashcode engine.

        • Re: Hungarian Notation (Score:3, Interesting)

          by bluGill ( 862 )

          The language is fine. The notation as used in programing (like MS does) is a pain. I have to use it, meaning I'm always making up prefexes for each class and structure I have. I have yet to see any benifit to it. I try to be kind and remember I've only worked on this code for a couple months, but I still hate it.

          • Re: Hungarian Notation (Score:4, Interesting)

            by rhinoX ( 7448 ) on Tuesday December 16, 2003 @09:28PM (#7741245)
            That's why you use a handful of well-defined and strictly used prefixes.

            s - struct
            a - array
            str - string (stl)
            sz - string (c)
            csz - CString

            etc. Making up new prefixes for everything is just about as useful as not using them at all.

    • Re:Windows 64 (Score:5, Informative)

      by PCM2 ( 4486 ) on Tuesday December 16, 2003 @09:08PM (#7741132) Homepage
      Windows has a native 64-bit version but Intel have prompted MS to delay the release until they can come up with a competitive processor.
      What you say may be true behind the scenes, but would you care to cite a source? [eweek.com] Last I heard, Microsoft's decision to withhold its Windows 2003 update would impact both AMD and Intel. At any rate, it's not like Microsoft isn't working with AMD. [pcworld.com]
    • Do you know if everything is recompiled, or is it bits and pieces, or is this another Win98 32 bit in places with 16 bit in others?

      (Everything includes drivers, libraries, through to applets, databases, and Office itself.)

      • Re:Windows 64 (Score:3, Interesting)

        by Nothinman ( 22765 )
        I read a while ago that it comes with both versions of alot of applications because right now it's more of a development platform than anything else.

        And frankly with how well Opteron handles 32-bit apps there's no reason to ship 64-bit versions of everything, a 32-bit userland with a 64-bit kernel and the option to run 64-bit userland apps as necessary is more than enough and infact that's how a number of Linux builds work, sparc64 comes to mind quickly since I have two of them.

        The big problem is shared l

    • Re:Windows 64 (Score:5, Insightful)

      by ctr2sprt ( 574731 ) on Tuesday December 16, 2003 @11:37PM (#7742018)
      Do you have any actual evidence to support this claim? Don't bother answering, I know very well you don't. For crying out loud, moderators, it's your job to weed out inflammatory, made-up nonsense like this! It's not being biased to mod down articles that make extraordinary claims with zero evidence - it's responsible moderation.

      I suppose I should spend some time demonstrating why this is stupid to avoid being flamebait. So first, Intel isn't working on a competitive processor, not in the sense you mean it. If they are working on one, it would almost certainly be a year or longer before they could roll it out, and there's no way MS would agree to wait that long. Second, there is no possible reason for MS to withhold Windows just because Intel asked them to. MS is allowing free operating systems to have a monopoly on AMD64 right now. Do you really think they'd do that voluntarily? Third, it seems clear that Intel is betting on there being no market for desktop 64-bit machines. I don't want to get into that particular flamewar, so let's just say that, right or not, that's what Intel believes, so it makes sense for their business decisions to reflect that belief.

  • moths (Score:5, Funny)

    by Anonymous Coward on Tuesday December 16, 2003 @08:42PM (#7740951)
    A few moths ago

    Even 64-bit Linux doesn't prevent spelling mistakes on Slashdot.

  • FYI No benchmarks (Score:3, Informative)

    by civilengineer ( 669209 ) on Tuesday December 16, 2003 @08:45PM (#7740963) Homepage Journal
    Benchmarking: Again, I do not have any benchmarks on CPU performance to show off. I did run the povray benchmark, but looking over the Povray benchmark page I see P4 3.0GHz machines getting beat out by Athlon 2800's and vice-versa, so I have little faith in the results (or at least in peoples postings to the site). Other sites on the net have run benchmarks on the Opteron, but without a 64 bit OS, and without 64 bit compiled benchmarks, I also have little faith in them. They should be used as a "touchy-feely-sort-of-may-bee" mark at best, in my opinion.

    • www.aceshardware.com for some benchmarks (Score:5, Informative)

      by slash-tard ( 689130 ) on Tuesday December 16, 2003 @08:53PM (#7741018)
      http://www.aceshardware.com/read.jsp?id=60000275

      They have 32 and 64 bit apache benchmarks along with some others compared against single and dual xeons.

    • Re:FYI No benchmarks (Score:5, Informative)

      by Serveert ( 102805 ) on Tuesday December 16, 2003 @08:53PM (#7741022)
      I've been benchmarking the opteron for the last week, it is at least 26% faster on high mysql load vs a comparably priced opteron system.

      Tom's Hardware, Anandtech and aceshardware have all benchmarked the opteron on linux. Tom's hardware's benchmarking isn't that great, aces hardware does the best job.. The Opteron kicked butt in all reviews.

      This is by far the best review so far IMO:

      http://aceshardware.com/read.jsp?id=60000275 [aceshardware.com]

      We're going to order a bunch of them by the end of this year so the government doesn't hit us with too many taxes, woo hoo!
    • I'm disappointed at the lack of any impressive benchmarks from the Opteron. I can only assume that the Opteron really isn't any faster than the P4. The extra address space might be nice though, for programmers at least.

      • The extra address space is nice. However most apps aren't programed to take advantage of 64 bits, and many couldn't benifit from it. Don't expect any improvent in times to do 2+2, because on either both numbers fit into your native data size. Expect massive improvements when adding 6 billion to 6 billion though.

  • Opteron and *BSD (Score:5, Informative)

    by BattleBlow ( 633941 ) on Tuesday December 16, 2003 @08:45PM (#7740966)
    I think you'll find that FreeBSD has only made amd64 a tier-1 architecture starting with FreeBSD 5.2 which isn't out yet [freebsd.org] and has been recently delayed until January.

    On the other hand, NetBSD has had amd64 support since 2001 [netbsd.org].

    OpenBSD is reportedly working on it [openbsd.org], but I haven't seen anything hit the tree as yet.

  • Opteron support could perform better (Score:5, Informative)

    by Suicyco ( 88284 ) on Tuesday December 16, 2003 @08:49PM (#7740994) Homepage
    The linux opterons we have run SuSE but since the opteron compiler support is still not up to par performance wise they have yet to make a big impact on run times. AMD needs to fund some good compiler development for this architecture, as it CAN perform incredibly, it just doesn't due to unoptimized compilers. Thats why IA64 still beats the pants off Opteron IMHO. The Madison chips from Intel are insanely fast, and their compiler is top notch. PG's compilers just aren't optimized as well as Intels, and it really shows. The numbers I've seen from AMD compared to the numbers I get, are two different things, obviously due to poor optimization at the compiler level.

    I suppose I dont even know the purpose of this post, just some observations :-)

  • Seti problems with x86-64 kernel (Score:5, Interesting)

    by Alizarin Erythrosin ( 457981 ) on Tuesday December 16, 2003 @08:50PM (#7741002)
    I don't know exactly what caused it, and it may not be much of a concern for other people, but the cpu time on my Seti@home units wouldn't increment using a Redhat beta for x86-64, with both the 64- and 32-bit clients. I liked the idea of using a 64-bit Linux distro but if I couldn't get Seti to run correctly on it, I'll just run a 32-bit version for now (Fedora Core 1 currently).

    As much as I'd love to support Linux by purchasing a distro, SuSE wants $130 for their AMD64 distribution, which I just can't afford right now. And I'm too much of a noob to build my own from scratch using pure source, so I'll hafta wait.

    But anyways, it's exciting to see more AMD64 distros, even if conspiracy theory says that Microsoft keeps delaying because of Intel pressure. I'm very happy with my dual opteron server, and will be even more-so when I can run pure 64-bit Linux.
    • This is why I'm glad I started out using slackware. 90% of people think it's actually hard to compile your own software.
      You have 2 options
      1) use a source based distro (gentoo, sourcerer would be best)
      2) make your own distro

      I GUARENTEE that if you follow the directions on the gentoo install page, that you won't have one problem. Any of the serious obsticles I've had with gentoo are user errors (I get bored during an emerge, start reading ahead, and skip a step).
      I never have been able to get sorcerer to inst
      • I currently run Gentoo, and while I'm quite happy with it, I think it's rather naive to say that the install guide accounts for all possible problems. I had problems installing Gentoo due to kernel problems, which are still problems even if Gentoo isn't responsible for them. As it turns out, I eventually discovered that the reason why the install process was repeatedly freezing (very odd for linux) was that an obscure bug existed in the driver for my sata hard drive controller. Switching controllers was
    • You're basing you decision on whether or not it can run Seti@home?? I don't see how something like that could possibly be a deal breaker...

  • hurdles (Score:5, Insightful)

    by potpie ( 706881 ) on Tuesday December 16, 2003 @08:55PM (#7741039) Journal
    I think there is a general benefit to Open Source that we haven't been able to observe until now. It is a fact that Open Source is more easily ported and adapted, but the major systems haven't changed much for the past many years (Mac, X86, etc.). Now that an entirely new system is out, proprietary software developers will be stumbling over themselves as they try feverishly to make something from scratch, while Open Source developers will benefit from working as a group.

    In a way, this has always been the way it worked, but now that there is a large jump in computing (32 to 64 bit processing is a pretty big jump, neh?) and the scale of development is made larger, the Open Source projects will show just how slow and inefficient proprietary software developing methods are.

    • Re:hurdles (Score:3, Insightful)

      by ergo98 ( 9391 )
      What a bunch of propaganda claptrap (you work for the Soviet communist party in a former life?) -- Lenin praises you comrade.

      Microsoft has developed all of their code to be cross-platform for years (NT used to run on several processors, but when the same software was available on multiple platforms it strangely led people to Intel), and upwardly bit-scalable, and has been demoing 64-bit editions of Windows for years. Having the technical ability to toss a basic operating system out the door, and wanting to

      • Re:hurdles (Score:3, Informative)

        by HalfFlat ( 121672 )

        I think you may have fallen for a little propaganda yourself ...

        Microsoft has developed all of their code to be cross-platform for

        years

        Not true I'm afraid. Windows NT 3.5 and 4.0 (and maybe 3.1 too?) were available for Alpha, MIPS and PowerPC. But other than these versions of the OS, and some associated server software (such as IIS and SQL server) everything was Intel only. In particular, the Microsoft Office software was only ever supported under emulation for Alpha.

        and upwardly bit-scalable

        Not r

    • Re:hurdles (Score:4, Insightful)

      by General Sherman ( 614373 ) on Tuesday December 16, 2003 @10:05PM (#7741496) Journal
      What? The Mac OS has just changed recently, the classic OS is gone and it's now based on a BSD core with the mach kernel. That's a pretty big change if you ask me.

      But If you're talking about 32-64 bit, then Apple has made the transition quite smoothly with the G5 if you ask me. While the OS itself is not true 64 bit, it supports 64 bit applications perfectly.

  • The conclusion... (Score:3, Informative)

    by joestar ( 225875 ) on Tuesday December 16, 2003 @08:58PM (#7741062) Homepage
    This lets you directly jump to the conclusion without having to read the 3-pages:


    Conclusion:

    The coming months (weeks?) should be interesting in that Mandrake is set to release the AMD64 version any time now, as they are taking pre orders for it in the Mandrake store. Recall, it was one of the best (if not the best) in my first review, and I blame the drive problems on the Asus BIOS update. Gentoo is nearing (from what I read) a really stable working system, and I have read repeatedly that others have it working fully (as a workstation with X windows) on other motherboards, so I again blame the Asus for my troubles with Gentoo. Red Hat is another story, having dropped the desktop edition, the "workstation" edition is well beyond my financial reach. A corporation may consider purchasing a copy for evaluation, but I would be tempted to wait on Mandrake or Suse.

    FYI: costs as of 12-16-2003 for AMD64 Linux distributions:

    Mandrake pre order $100
    Mandrake corporate server $750 (standard support) $1500 (unlimited support)
    Red Hat AMD64 workstation $792
    Red Hat Advanced Server $1992
    Suse Professional 9.0 $120 (distribution on DVDs, no CDs)
    Suse Enterprise Server $767 (2 cpu) $1450 (4 cpu)

    Looking at the above cost matrix and my experience, it is almost tempting to purchase SuSe just to have the DVDs (no CDs, strange). The enterprise/server editions seem to all be priced about the same, with no definitive mention of CPU capability from RedHat or Mandrake on the server editions. (I assume at least 2 CPU capability built into the kernel)

    Side note: the Mandrake pre-order in question is Mandrake 9.2 (pre-order is at http://www.mandrakestore.com [mandrakestore.com])

  • Speed vs Memory (Score:5, Interesting)

    by ArkiMage ( 578981 ) on Tuesday December 16, 2003 @08:59PM (#7741071)
    The whole idea of a CPU with more bits of addressability is memory... MORE memory... 4GB of addressable RAM on a 32-bit processor is simply not enough today. Speed is a side-issue, they're already fast, some of us just want more RAM.

    We have a couple of Opterons with 8GBM RAM each running as MySQL/INNODB backend database servers. With that much RAM databases that would crawl on IA32 are very fast since so much more of it can be cached in RAM.

    The only real problem is memory technology hasn't kept up. 1GB DIMMs can be had at almost reasonable prices but 2GB density ones are out of range of most everyone. 4GB are on the distant horizon.

    I'd have gladly stuffed 16 or 32GB of RAM in the boxes we have if it had been affordable. More for less!

    • Re:Speed vs Memory (Score:5, Interesting)

      by IvyKing ( 732111 ) on Tuesday December 16, 2003 @09:25PM (#7741228)
      The only real problem is memory technology hasn't kept up. 1GB DIMMs can be had at almost reasonable prices but 2GB density ones are out of range of most everyone. 4GB are on the distant horizon.

      Crucial is listing their CT51272Y265 DIMM's for a measly $6999 - these are 4GB PC2100 registered with ECC. The price (ahem) may be a bit high, but if you really need the memory...

      Hal Computers had an interesting "benchmark" back in the late 90's. Their Sparc box was capable of handling 3 GB (at close to 80 grand per GB), one chip simulation took 40 hours with 2 GB and 1.5 hours with 3 GB.

    • Re:Speed vs Memory (Score:3, Interesting)

      by Nothinman ( 22765 )
      If you're letting the OS do the caching you could get an IA32 system that handles >4G memory and the performance would probably be close, I'm not sure how much of a performance hit PAE in the kernel has but the price difference might make it worthwhile. And since you'd be relying on the normal Linux page cache the applications don't have to be 64-bit aware to have all that memory used for their caching.
    • I want to play Civ IV with 18 million terrabytes of addressable memory. Paris is is researching Creme Brulee, Liverpool is building the mod haircut, and there is a coppertone shortage in Pismo Beach.

      Is there a word for a mega-terrabyte?

  • Compatibility-performance fallacy (Score:3, Informative)

    by scd ( 541350 ) <scottdp.gmail@com> on Tuesday December 16, 2003 @09:00PM (#7741080)
    The Opteron may have compatibility for 32-bit programs, but it won't be as effective as its native 64-bit mode.

    I'm sick of people making the mistake that a '64-bit' processor will automatically perform poorly at apps compiled for 32-bits. In the case of the Opteron, a 64-bit app will probably run better due to more general purpose registers (32 vs. 8), but by his tone, the author of the article seems to think that 32-bit app performance will be unimpressive (like Itanium).

    This just ain't the case with Opteron or Athlon64.

    • Ummm... those 32-bit compiled programs aren't exactly going to be making use of those extra general purpose registers, you know... register allocation is (mostly) done at compile time.

  • Debian port infomation. (Score:4, Informative)

    by bjarvis354 ( 319402 ) on Tuesday December 16, 2003 @09:01PM (#7741088) Homepage
    Debian has not released its port yet, but it is coming. Here is the official Documentation (FAQ and HOWTO) [debian.org]

  • What about 4GB? (Score:5, Interesting)

    by localman ( 111171 ) on Tuesday December 16, 2003 @09:02PM (#7741100) Homepage
    Doesn't seem the article tests the system with >4GB. That seems odd since that is one of the most compelling reasons to go 64bit (other than pure bragging rights).

    My company [zappos.com] upgraded to SuSE on Opteron a few months back, and had some random memory corruption with our 8GB setup. Turned out it was some bad interaction between the Tyan motherboard, the BIOS, and the stepping 1 of the Opteron. What a pain.

    We're stable now with 4GB, but the memory was the only reason we upgraded in the first place. I'd like to see more tests with lots of memory.

    Cheers
    • The ability to flat-address more memory is nice, but not necessary to take advantage of A64 performance.

      Most 32/64 bit hybrid architectures lose speed when going from 32 to 64 bit software due to losing bandwidth to storing pointers.

      AMD64 actually has an advantage to offset that in the fact it doubles the number of registers in 64 bit mode, so even code that doesn't break the 4GB barrier has a chance of running better.

    • Re:What about 4GB? (Score:4, Interesting)

      by Junta ( 36770 ) on Wednesday December 17, 2003 @12:31AM (#7742359)
      I've been working with an Opteron platform as well and I've seen several issues when the memory gets above 4GB. A lot of it has to do with them recycling the 32-bit code used for BIOS operations before that breaks under those conditions. In our case, the problems are mostly solved. We have seen a couple of dirvers, however, that had issues (some really picky portions that really relied on addresses being 32 bit values). The 32-bit capability is a double-edged sword, gives greater compatiblity, but someimtes companies take shortcuts with what they have since it at least gives the initial impression of working, when the reality is that the shortcuts break things in some obscure ways.

  • heh? (Score:5, Interesting)

    by DashEvil ( 645963 ) on Tuesday December 16, 2003 @09:02PM (#7741103)
    "The only questionable aspect of the Suse distribution is the choice of kernel, which is 2.4.21. I know that 2.6.x is beta for now, but it does seem (from the Gentoo installs) that it is faster and able to play nice with the ACPI, unlike 2.4.x on this motherboard."

    Can someone tell me why using a stable kernel over a development kernel is a 'questionable' decision?

    I stopped reading the article there, that is just stupid.
    • I'm actually surprised he got gentoo amd64 to compile.. the last time I tried (just last week, actually) it couldn't even compile a stage 3 due to compile errors, let alone get as far as X.

      Debian is months away from a stable distro (waiting on dpkg and apt upgrades to support multi-arch installs) and I'm damned if I'm going back to RPM... it's 64bit hell out there :)

    • Re:heh? (Score:4, Interesting)

      by EMN13 ( 11493 ) on Tuesday December 16, 2003 @10:18PM (#7741585) Homepage
      That's not so stupid at all. Although i have no experience with 2.3; I've heard it said that the general stability of 2.6-test is quite beyond that of the early 2.4-test - so much so that it's quite useable.

      Furthermore; Before a kernel is "stable" it's going to have to be stable on most arches that it's to run on; support all the hardware correctly etc etc etc. For a distro specifically targetted at one arch, it can be much simpler to target a good stability because the problematic hardware interactions are far simpler.

      Finally, it seems entirely normal, and indeed the opposite rather "questionable", for a _beta_ distribution to include that software that they intend to ship with. 2.6 Is definitly nearly at that point; as such it's the obvious choice to use. By the time AMD64 under linux is ready for prime time, i bet that 2.6 will be too.

      --Eamon

  • Anyone know about the maturity of NUMA support? (Score:3, Interesting)

    by Serveert ( 102805 ) on Tuesday December 16, 2003 @09:05PM (#7741114)
    I believe the latest 2.4 Linux kernels have NUMA support but is it mature or will it get any better? Are other unix OS'es better at taking advantage of NUMA compared to Linux and will this change in a future Linux version?

  • .NET (Score:5, Interesting)

    by MrBlack ( 104657 ) on Tuesday December 16, 2003 @09:05PM (#7741118)
    That is one of the interesting benefits of virtual execution environments. AFAIK the JIT compilation process can take advantage of the target system's architecture, so .NET apps would not need to be recompiled to see a beneift. I don't have access to a 64 bit CPU so I haven't investigated but I did notice 64 bit versions of the .NET framework in the latest Whidbey betas. I'm not sure if there is a 64 bit version of framework 1.0 or 1.1. I did notice at least one Tech-Ed presentation from this year was on writing .NET apps to target 32 and 64 bit platforms.

    • Re:.NET (Score:3, Informative)

      by Keeper ( 56691 )
      Microsoft currently doesn't have a 64bit version of the .Net runtime for 1.0 or 1.1. Whidbey (.Net 2.0) is supposed to ship with a 64bit version.

  • Radeon drivers (Score:3, Informative)

    by Fnord ( 1756 ) <joe@sadusk.com> on Tuesday December 16, 2003 @09:18PM (#7741188) Homepage
    I've been very happy with Suse on my opteron system, but there's one thing that keeps a 32 bit installation on another partition. ATI, though they've made several press releases about how they "fully support the opteron", has not felt the need to release 64 bit versions of their drivers (either for linux or windows), and the open source radeon driver doesn't support the radeon 9700 pro that I own. I'm almost tempted to get an nVidia card simply because they have 64 bit drivers, even though this generation of cards just isn't as good as ati's.

  • 64 bit is more about memory (Score:4, Interesting)

    by stabiesoft ( 733417 ) on Tuesday December 16, 2003 @09:43PM (#7741362) Homepage
    I ported my software to the opteron a few months back. It was quick & easy (I used a beta red hat distro). The main reason I got the box was to provide customers an alternative to sun. I work in EDA (we do the software to make the chips) and 4GB is not enough for the big chips. I'd encourage other developers to give the opteron a try. I think it took all of 2 days to do the port. Performance has been good, but since I can't afford a fast sun box, I can't really compare.

  • What's needed is a Killer App (Score:5, Insightful)

    by Ridgelift ( 228977 ) on Tuesday December 16, 2003 @10:33PM (#7741654)
    ...my point is that Linux has 64-bit support and it has it now. Linux and AMD are a natural partnership.

    What's needed is a killer game that runs on Linux-64. The must-have toy will drive Linux faster and further than any business app could. It's the reason I know most people overspend on a PC, so they can play the latest and greatest games.

    Intel's known this for years. That's why they give early release processors to the top game manufacturers so that when the new processor hits the street, there's software that'll shine with it.

    • Re:What's needed is a Killer App (Score:5, Insightful)

      by NerveGas ( 168686 ) on Tuesday December 16, 2003 @11:19PM (#7741919)

      RDBMS systems are your killer app. Opterons are well-suited to RDBMS work, to the point of nearly seeming intended for it. Between the "big iron" memory architecture and the 64-bit address space, AMD has finally provided commodity hardware that can truly tackle real, heavy database environments.

      The only reason I didn't buy an Opteron for our main RDBMS server this year was because they weren't ready for our peak season. This coming year, I'll be getting a minimum of a dual-opteron, more likely a quad - and getting it for a fraction of what similar performance would cost from Sun.

      steve

    • Business Apps (Score:4, Insightful)

      by traskjd ( 580657 ) on Tuesday December 16, 2003 @11:45PM (#7742061) Homepage
      In my opinion the "killer app" that not only 64bits systems, but linux in general needs to make any serious headway is business applications. Another person who replied touched on this with mention of RDMS.

      What linux needs more of is actual business systems (Point of sale, finance tracking etc - for small to medium sized businesses). If you could run your point of sales system on linux the savings of several hundred dollars per system would be a major advantage. I mean it was the spreedsheet that really brought pushed PCs mainstream (you start using one at work, then you think you should probably have one at home... story goes on).

      It's just an opinion - but I think we have more than enough text editors and windowing environments.

      - traskjd

  • Should of benchmarked kernel 2.6 (Score:3, Informative)

    by Billly Gates ( 198444 ) on Wednesday December 17, 2003 @12:40AM (#7742414) Journal
    2.6 supports many more processors, more chipsets including the new 64 bit ones which 2.4 is buggy when apic or apci is turned on, Serial ATA hard drives and cdroms, etc.

    Of course alot of it has to do with the gnu C compiler being optimized but the newer kernel will better take advantage of the newer chipset and other features.

  • FreeBSD/amd64 (Score:5, Interesting)

    by DarkHelmet433 ( 467596 ) * on Wednesday December 17, 2003 @01:29AM (#7742629)
    I threw together a 30 second screenshot in case anybody is interested. http://people.freebsd.org/~peter/desktop.png [freebsd.org]

    FreeBSD/amd64 is a pure 64 bit OS. There is no 32 bit code at all. The kernel, userland, ports/packages etc are all 64 bit. None of this hybrid 64/32 stuff. :-)

    Actually, this is probably our greatest liability. While we can run 32 bit binary applications (can you say perforce?), it isn't perfect. Much more work is still going to be done in this regard.

    If anybody is interested in giving FreeBSD/amd64 a whirl on one of these machines, we'd appreciate folks trying out the 5.2-RC1 ISO images. See the ftp link on the story above. Since RC1, lots of bugs have been found and fixed. Most notably for support of KDE and gnome environments. If you do try it out, do be aware that its still a little green in this area.

    I personally, have been running a FreeBSD/amd64 desktop for about 2 months. I do subscribe to the 'eat my own dogfood' mantra. I do not have any x86 unix machines left except for my 486 firewall and a laptop. That goes for both home and work. My work desktop is FreeBSD/amd64 too.

    Anyway, it's nice to see a FreeBSD reference here for a change.

  • If X doesn't work on Gentoo/Opteron... (Score:3, Informative)

    by Brane2 ( 608748 ) on Wednesday December 17, 2003 @01:34AM (#7742653)
    How come I'm using it on my dual Opteron 240 (on Tyan S2885) ?
    It's true that many things doesn't work in 64-bit mode (loke OpenOffice, Abiword etc), but system WORKS ! By "system" I mean X, cups, samba, KDE, Gnome etc. It works so well that I have bought two dualCPU machines (update of two aged workstation machines-P4 2 GHz and Tualatin 1.4 Ghz @ 1.7 GHz) and I'm waiting for a third to arrive- that one will replace fileserver/printeserver/firewall/etc machine.... Gentoo on Opteron works, and unpolished details are getting its shine rapidly. I use: Motherboard TYAN Thunder K8W S2885 2x Opteron 240 2 Gb of PC2700 ECC Reg (512 Mb modules) GeForce 4 Ti4200 HDD EIDE 120 Gb WDC

  • 64 bit ...not nescessarily for performance (Score:5, Insightful)

    by Chanc_Gorkon ( 94133 ) <<moc.liamg> <ta> <nokrog>> on Wednesday December 17, 2003 @01:35AM (#7742656)
    I have had a 64 bit AIX machine running for a while with the 64 bit kernel. While I have not really had the load yet to test it, I and many others in the AIX realm don't necessarily think that 64 bit is going to increase performance. How do you test a performance increase when it only increases by a few nanoseconds??

    64 bit is all about memory addresability. You can directly address more memory on a 64 bit machine then you can a 32 bit machine. Period. When you would like to get the best performance you can out of your RDBMS, most shops like to load as much of the DB as they can into memory. DB's are getting larger then 4 GB now! :) So, the need for more memory is upon us.

    BillG said 640 KB out to be enough for anyone..ha ha Bill. Very funny.

    • Re:64 bit ...not nescessarily for performance (Score:4, Insightful)

      by NerveGas ( 168686 ) on Wednesday December 17, 2003 @02:14AM (#7742820)
      .... but a 64-bit address space and very fast memory controllers do make for good performance!

      Every year, as our business has grown, I've had to upgrade our DBMS server to keep up. We've gone from a 4xP3 Xeon to a 2x AthlonMP to a 2xP4 Xeon, and next year it will be a 2x or 4x Opteron.

      In every case, when the machine is finally hit it's max capacity, the CPU's were nearly *never* at full use. Even though the entire operation was running from memory and cache(the disk lights rarely blinked), the memory bandwidth has always been the limitation. Between the Opterons having VERY fast memory controllers (and each chip having it's own controller), and the ability to address vast amounts of memory, it's a recipe for letting those CPU's fulfill just a bit more of their true potential! : )

      steve

  • Windows applications (Score:4, Insightful)

    by Sheepdot ( 211478 ) on Wednesday December 17, 2003 @01:57AM (#7742752) Journal
    If 40% of windows applications are going to have 64-bit support in the next year, isn't that a high enough number to actually *justify* getting the machine for a Windows-based system?

    I mean, I understand that Linux applications will most likely have 64-bit support a lot sooner, but 40% of windows application support in the first year sure looks like enough of a reason to purchase the machines now.

    I guess I don't see a huge argument in justifying that only %40 of windows applications are going to have 64-bit support when there's virtually no drawback to buying a 64-bit processor from AMD vs. an equally priced 32-bit processor from Intel.

    Sure, you can argue that it's a "waste", but even if only three of the big players have 64-bit applications (Microsoft, Macromedia, Adobe) within the first year, that's still 90% of the applications that are used on Windows machines in a corporate or even personal environment for the average user.

    The driving force is going to be the gaming community, and AFAIK, the major game software companies plan on having 64-bit games available too, so I fail to see what the real issue regarding support is.

    If %40,%30,%20,%10 is a fair assessment of compatibility over the next five years, that means that in three years %90 of the Windows applications can be assumed to have 64-bit support, which is perfectly fine for the corporate or average 3-year life cycle of a computer.

    Or am I missing something?

  • C types (Score:4, Interesting)

    by WhiteDragon ( 4556 ) * on Wednesday December 17, 2003 @03:08AM (#7743057) Homepage Journal
    I suppose this is really more of a gcc question, but here goes. Does the amd 64 use a 64 bit pid_t, time_t, uid_t, etc? In my opinion, that is one of the more important reasons to switch to a 64 bit processor.

Slashdot Top Deals

Saliva causes cancer, but only if swallowed in small amounts over a long period of time. -- George Carlin

Close