Announcement

Collapse
No announcement yet.

Another Look At Intel's Lynnfield Linux Performance

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    I put together an i7 860 system with a gigabyte p55m-ud4 board and 4x2GB dimms.

    Ran it in parallel with a dual 1.6 clovertown setup with 8GB.

    Programs are multi threaded, scaling linearly. Threads are multiple instances of the same instructions run on data packets.

    Automatch: 860: 7x 1.6x faster
    Least Squares Sparse System: 860: 2.6x faster
    OrthoRectification: 860: 1.6x faster

    Other tests resulted in the 860 system being IO limited.
    I have a dual hapertown 2.33 system I'd also like to run tests against.

    I would hopefully expect a dual core i7 system to really rock.

    Comment


    • #42
      Originally posted by bnolsen View Post
      I put together an i7 860 system with a gigabyte p55m-ud4 board and 4x2GB dimms.

      Ran it in parallel with a dual 1.6 clovertown setup with 8GB.

      Programs are multi threaded, scaling linearly. Threads are multiple instances of the same instructions run on data packets.

      Automatch: 860: 7x 1.6x faster
      Least Squares Sparse System: 860: 2.6x faster
      OrthoRectification: 860: 1.6x faster

      Other tests resulted in the 860 system being IO limited.
      I have a dual hapertown 2.33 system I'd also like to run tests against.

      I would hopefully expect a dual core i7 system to really rock.
      Tests against a dual harpertown 2.33 on a 5400 chipset shows the i860 6-10% faster depending on task. Pretty interesting, it's a hugely worthwhile upgrade from the core2 quad series.

      Comment


      • #43
        Is there anymore problems with p55? I havent found answer for that question. Does Turbo Mode etc. work well under linux nowdays?

        Comment


        • #44
          Originally posted by Immo View Post
          Is there anymore problems with p55? I havent found answer for that question. Does Turbo Mode etc. work well under linux nowdays?
          Frequency scaling doesn't seem to work. Turbo mode may...the machine is idle and reports 2.93GHz, spec is 2.80. I haven't worked with any core2 quad single socket, I can only compare to dual socket c2d's.

          Monday I'll get my hands on a pair of dual socket i7's and will do some testing on those.

          Comment


          • #45
            The Turbo mode can be shown only as a +1 mhz increase of default speed with /proc/cpuinfo while 2 other cores use a lower speed state.

            Comment


            • #46
              Thanks for answers.
              So Turbo doesnt work right then?
              I understood that if single core is under load, it should be 3.2Ghz and if all cores are under load they rise at 2.8Ghz.

              I am thinking of assembling my own computer and these Lynnfields look nice, but if they dont work properly under Linux i'll pass. Should I consider i7 920 or Phenom II instead?

              And sory for my bad English....

              Comment


              • #47
                Originally posted by crat3rs
                One of it's shortcomings, Hyper-Threading Technology is not present, which is needed for working with heavy multi-threaded applications.
                The i7 has hyper-threading, but the i5's don't. I'm no expert but I'd say if you've got 4 cores, that should be enough for most multi-threaded applications.

                Edit: In case anyone is wondering why I replied to a non-existent post, it turns out the post was spam.
                Last edited by ad_267; 26 December 2009, 04:38 PM.

                Comment


                • #48
                  If you compare AMD's Phenom II 965 and i5 which one wins when looking Linux performance in desktop use?

                  Comment


                  • #49
                    Is it safe to overclock Lynnfields despite the fact lm_sensors doesn't work?

                    Comment


                    • #50
                      I wonder when we'll see Michael re-run the tests on his hardware, see if his numbers match up with Intel's ...

                      I'm assuming the Intel supplied numbers are absolute best case scenario, you wont do any better without overclocking kind of thing, and you might not be able to match the numbers at stock if you live in a place where ambient temps hang around 32C.

                      Comment

                      Working...
                      X