Who wants to discuss CAS Latency in PC 8500 DDR2?

Discussion in 'OT Technology' started by KGB ate my bread, Nov 19, 2009.

  1. KGB ate my bread

    KGB ate my bread Made you look. OT Supporter

    Joined:
    Mar 15, 2005
    Messages:
    97,612
    Likes Received:
    111
    Location:
    Louisiana
    :hsd:

    My current ram is PC 6400? and has pretty slow timings for what they are;

    5-5-5-18 :o


    My Mobo supports PC8500 and I figured I'd get a 4gb (2 x 2gb) dual channel sticks since I've upgraded to a Phenom X4 940 II and figured the extra speed wouldn't hurt, and would eventually let me run 8gb instead of the 4gb I'm currently running (4 x 1gb sticks).

    Anyways, What's a good latency to look for in PC 8500? I've noticed the very a bit, and new egg likes to throw stuff out there like 'primary use: high performance and gaming' which to me is just marketing to make it look like that memory might be used better in those situations.


    Anyways, what sort of timings should I be looking for?

    I've seen them range from 5-5-5-15, to 5-6-6-18, to 6-6-6-18, to 7-7-7-20..


    Which would be the best timings to consider :dunno:
     
    Last edited: Nov 19, 2009
  2. KGB ate my bread

    KGB ate my bread Made you look. OT Supporter

    Joined:
    Mar 15, 2005
    Messages:
    97,612
    Likes Received:
    111
    Location:
    Louisiana
    I've read this article:
    http://en.wikipedia.org/wiki/Cas_latency

    "In general, the lower the CAS latency, the better. When comparing latencies at different clock speeds, they must be translated into times to make a sensible comparison; a higher numerical CAS latency can be less time if the clock is faster"

    It gives a DDR3 1066 example of 7 for the CAS LAtency.. Is it safe to assume that the modules with 5-5-5-15 are quicker?
     
  3. DIABLOS

    DIABLOS New Member

    Joined:
    Jan 18, 2004
    Messages:
    71
    Likes Received:
    0
    Location:
    U.K
    The lower the Latency the quicker the RAM is responds to a request so lower is better, but the difference is very minimal in the real world.
     
  4. CodeX

    CodeX Guest

    Depends on what you're doing.

    IO bound applications (word processors, web browsers, image editors (for the most part)) will not care, there aren't enough memory accesses for the difference to matter.

    Processor bound applications working with large data sets will see a significant improvement in performance with lower memory access delays.

    On my project at work, during a certain mode of operation (I'm not going to get into it, it is complicated), The processor is accessing our off-chip DRAM module 4.68 million times per second, by changing the external interface timings a small amount (as far as I could go before it failed), I was able to achieve a 25% performance increase and turn a 16 minute process into a 12 minute process...

    I wouldn't call that a minimal difference.
     
  5. OniMinion

    OniMinion ...recalls when this forum was actually about cars OT Supporter

    Joined:
    Jan 26, 2005
    Messages:
    4,894
    Likes Received:
    1
    Location:
    Indiana/Minnesota
    Usually less than .5% when looking at FPS in games.
     
  6. CodeX

    CodeX Guest

    If I had to guess I would say most of the intense stuff happens in video ram anyway in that case...

    Things like video compression/decompression, high definition rendering, (huge) database processing would all benefit more I think.
     

Share This Page