MMM
Results 1 to 5 of 5

Thread: can someone clarify this ? "Higher multi == higher mem bandwith" ??? (A64)

  1. #1
    Xtreme Member
    Join Date
    Dec 2004
    Posts
    455

    can someone clarify this ? "Higher multi == higher mem bandwith" ??? (A64)

    from:
    http://www.insanetek.com/index.php?page=overclocka64

    ***
    This means that there [for the older XP CPUs] was no difference in memory bandwidth when running 11x200FSB or 10x200FSB.
    The CPU horsepower was different, but the memory bandwidth stayed the same. This is no longer the case with Athlon64 CPUs. Multipliers affect the memory bandwidth significantly, which means that a higher multiplier results in higher bandwidth.
    ***

    I need to have clarification....this guy does some benches, eg. he has (WInnie 3000) at 9x200 and then at 8x200....and since the 8x200 is SLOWER (DUH, DUH !!!!) he states "a higher multi also equals higher memory bandwidth".

    Theoretical scenario:

    winchester running at
    200 HTT X 10, and then (for comparison)
    400 HT x 5 (ASSUMING i could hit 400FSB/HTT). The last one has the "lower" multi).

    Now i just heard (somewhere) Oscar Wu quoted, he made a comment about how a really low multi (eg. 5x) would run the memory "slower than you might think"

    So...what is it about "a higher multi is better than a lower multi" ? Eg. lets say i find my CPU hits a max of 2700...eg. i could do 9x300 or 10x270...what would REALLY be better ?

    (BTW,. i always thought that the multi itself is not important AS LONG as i am able to hit the CPU's limit as well as the memory limit ! But now it looks like that setting down the multi is something you better avoid ?)

    thanks

  2. #2
    Xtreme Member
    Join Date
    Jan 2005
    Location
    Manchester, UK
    Posts
    457
    The memory is faster at higher FSB and/or tighter timings. The memory controller will be faster at higher overall CPU frequency since it's on chip in A64s. So you might see an increase in memory bandwidth between 8x200 and 9x200. However 6x300 should still be way faster than 9x200.
    I think the guy in question is drawing the wrong conclusions - the chip is faster overall at 9x200 which is the difference he's seeing, it's not to do with the multiplier per se.

  3. #3
    Banned
    Join Date
    Oct 2004
    Posts
    36
    Indeed.
    There are marginal bandwidth increases as the multi goes up, for any given FSB.

  4. #4
    Xtreme Member
    Join Date
    Jan 2005
    Posts
    443
    ive read the same somewhere else a loong time ago about a64

    he did alot of testing and benching to proove it too...

    but...

    i forgot where...

  5. #5
    Xtreme Enthusiast
    Join Date
    Jun 2004
    Location
    Florida, USA
    Posts
    934
    A higher multi means a higher clock speed, assuming the same FSB. So of course, a 8x300 buffered bandwidth would of course be lower than a 9x300 buffered bandwidth.

    But in the case of comparing 10x270 to 9x300, which would be the same CPU clockspeed, the 9x300 would offer more bandwidth (of course), because of the higher FSB.

    Seems like an exercise in common sense to me.
    DFI LP LT X38-T2R
    Intel Core 2 Quad Q9450
    Thermalright Ultra-120 eXtreme
    2x2GB OCZ Reaper DDR2 PC8500
    eVGA 9800 GX2
    WD Raptor X 150GB
    PCP&P 750W Silencer
    CM Stacker 830 SE

Bookmarks

Bookmarks

Posting Permissions

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