Page 1 of 3 123 LastLast
Results 1 to 10 of 23

Thread: AMD vs. intel for Linux KVM machine

  1. #1
    Join Date
    Jun 2009
    Posts
    17

    Default AMD vs. intel for Linux KVM machine

    Hi guys.

    I'm building a machine to play with Linux KVM. Actually I have 4 options to go:

    Intel Core2 - support for: VT-x, VT-d / IOMMU (on Q35/45 chipsets)
    Intel Core i7 - support for VT-x, VT-d (chipset), EPT (nested pages)
    K8 AMD Athlon64 - support for AMD-V
    K10 AMD Phenom II / Athlon II - suport for AMD-V and Nested pages


    Second Option Intel Nehalem I expect to be most powerful, but it is too expensive for me. K8 Architecture I think is outdated. So I'm deciding mainly between Core2 and AMD K10.5 architecture, which is VT-d vs. Nested pages decision too.

    What is the best option to go? AMD 10.5? Intel Core2? Or Core i7 bcause the advantage si too big? I appreciate every idea. Thanks a lot.

    PS: I saw almost every benchmark here on Phoronix, but it usually does not compare main competitors AMD vs. Intel or ATI vs. nVidia. This time, comparison AMD vs. Intel would be useful.
    Last edited by Pepazdepa; 06-21-2009 at 03:45 AM.

  2. #2
    Join Date
    Jul 2008
    Location
    Berlin, Germany
    Posts
    790

    Default

    Anandtech has done some comparison between AMD and Intel Server CPUs, including virtualization. The results should be roughly transferable to a desktop CPU of the same architecture/clock speed/number of cores.

  3. #3
    Join Date
    Nov 2007
    Posts
    1,353

    Default

    Quote Originally Posted by chithanh View Post
    Anandtech has done some comparison between AMD and Intel Server CPUs, including virtualization. The results should be roughly transferable to a desktop CPU of the same architecture/clock speed/number of cores.

    I definitely have to disagree with you there...

    The types of software that is run on a desktop is a lot different then software for servers. I dont run apache on my desktop, but I do run Nexuiz... They arent exactly comparable loads now are they?

  4. #4
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    4,729

    Default

    If anyone is going to make a comparison, I'd make the wish of including the Via Nano in it. And not only compared to each other's virtualization performance, I'm mainly interested in a comparison between their native and virtualized performance.

  5. #5
    Join Date
    Jul 2008
    Location
    Berlin, Germany
    Posts
    790

    Default

    Quote Originally Posted by duby229 View Post
    The types of software that is run on a desktop is a lot different then software for servers. I dont run apache on my desktop, but I do run Nexuiz... They arent exactly comparable loads now are they?
    I was referring to desktop CPU vs. server CPU (ie. hardware), not desktop workload vs. server workload.

  6. #6
    Join Date
    Nov 2007
    Posts
    1,353

    Default

    Quote Originally Posted by chithanh View Post
    I was referring to desktop CPU vs. server CPU (ie. hardware), not desktop workload vs. server workload.
    In AMD platforms the server brands are essentially identical to the workstation and desktop brands. Really the only comparisons that make sense is the loads that you put on them. Examples being AMD's Barcelona is a server processor that is identical to AMD's Deneb Desktop processor. They come off the same line from the same factory and are even cut from the same wafers. The difference is in the binning process. According to quality control standards some are binned a Opterons, some as Phenom X4, Others with one bad die are X3, and still others are binned as Athlon X2. From the worst of the batch to the best of the batch.

    So since we know that the CPU's are essentially the same, your trying to say that a server load is indicative of a desktop load, and that simply is not the case. You can look at apache benches and say that this CPU will run apache at this level, and a Phenom X4 should roughly produce the same results. The problem is that we dont run apache on our desktops and so that benchmark has no value to a desktop user. I dont play mysql. I dont edit documents with apache. I dont browse the web with gcc.
    Last edited by duby229; 06-29-2009 at 11:25 AM.

  7. #7
    Join Date
    Oct 2007
    Location
    Under the bridge
    Posts
    2,099

    Default

    Fortunately, there are many benchmarks that compare Phenoms and Cores in typical usage *and* virtualization.

    Phenoms II CPUs are very competitive on both fronts and slightly cheaper. The big question is the rest of the platform (motherboard, video cards etc). AMD tends to have the advantage from a feature/price standpoint, but Linux changes the equation.

    Are you planning to use the built-in IGP or buy a separate GPU?

  8. #8
    Join Date
    Jul 2008
    Posts
    1,712

    Default

    http://techgage.com/article/amd_phen...lack_edition/4

    be carefull with core2. Intel loves to screw over customers.

    See this:
    http://ark.intel.com/Product.aspx?id=36500#specs

    as an example. Four cpus, three times the exact same name. But only one can do vt.

  9. #9
    Join Date
    May 2009
    Posts
    49

    Default

    If you want to guarantee that you actually get a cpu that supports virtualization, stick with AMD. These intel crapolas may or may not support virtualization in any particular model number at their whim and getting documentation from them to prove one way or another is quite nearly impossible.

    $ cat /proc/cpuinfo
    processor : 0
    vendor_id : GenuineIntel
    cpu family : 6
    model : 23
    model name : Intel(R) Core(TM)2 Quad CPU Q8200 @ 2.33GHz
    stepping : 7
    cpu MHz : 2003.000
    cache size : 2048 KB
    physical id : 0
    siblings : 4
    core id : 0
    cpu cores : 4
    apicid : 0
    initial apicid : 0
    fpu : yes
    fpu_exception : yes
    cpuid level : 10
    wp : yes
    flags : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good pni dtes64 monitor ds_cpl est tm2 ssse3 cx16 xtpr pdcm sse4_1 lahf_lm
    bogomips : 4682.26
    clflush size : 64
    cache_alignment : 64
    address sizes : 36 bits physical, 48 bits virtual
    power management:
    Quote Originally Posted by Pepazdepa View Post
    Hi guys.

    I'm building a machine to play with Linux KVM. Actually I have 4 options to go:

    Intel Core2 - support for: VT-x, VT-d / IOMMU (on Q35/45 chipsets)
    Intel Core i7 - support for VT-x, VT-d (chipset), EPT (nested pages)
    K8 AMD Athlon64 - support for AMD-V
    K10 AMD Phenom II / Athlon II - suport for AMD-V and Nested pages


    Second Option Intel Nehalem I expect to be most powerful, but it is too expensive for me. K8 Architecture I think is outdated. So I'm deciding mainly between Core2 and AMD K10.5 architecture, which is VT-d vs. Nested pages decision too.

    What is the best option to go? AMD 10.5? Intel Core2? Or Core i7 bcause the advantage si too big? I appreciate every idea. Thanks a lot.

    PS: I saw almost every benchmark here on Phoronix, but it usually does not compare main competitors AMD vs. Intel or ATI vs. nVidia. This time, comparison AMD vs. Intel would be useful.

  10. #10
    Join Date
    Jul 2008
    Posts
    1,712

    Default

    Ibcoder - looks like your cpu can't do vt in hardware - or I am blind

    well, my point stands - intel tries to screw users. There are SEVERAL cpus with the EXACT same naming - some of them can do virtualization in hardware, some can't and you can not check that until you have built them into your box.

    Screw intel, go amd. The 955 and 966 gives the 920i and qx9770 a run for their money - while still cheaper. Money you can put into quality mainboards and lots of fast ram.

    If you want virtualization, go AMD.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •