Page 1 of 2 12 LastLast
Results 1 to 10 of 13

Thread: Corsair DHX 4GB DDR2-800MHz

  1. #1
    Join Date
    Jan 2007
    Posts
    14,599

    Default Corsair DHX 4GB DDR2-800MHz

    Phoronix: Corsair DHX 4GB DDR2-800MHz

    It's been a while since last looking at any Corsair memory at Phoronix, but up for review this afternoon is their latest TWIN2X4096-6400C4DHX memory. This DDR2 memory features Corsair's DHX technology for cooling the memory ICs with EPP latencies of 4-4-4-12 and run at 800MHz. Like many other Corsair products, the TWIN2X4096-6400C4DHX is also backed by a lifetime warranty.

    http://www.phoronix.com/vr.php?view=12886

  2. #2
    Join Date
    Jan 2008
    Posts
    17

    Default

    I'm wondering why the quicker ( 4 4 4 12 ) timings seem to be slower then the standard timing. Both seem to have this behavior, In nearly every test they seem to be slightly slower , and in the ramspeed test they drop from nearly 3 gb/s to 2.2 .

  3. #3
    Join Date
    Mar 2007
    Location
    Savannah, GA
    Posts
    10

    Default

    I would think that lower latency at a given memory clock would result in higher frame rates for the Nexuiz, but for both OCZ and Corsair, lower frame rates are the result.

  4. #4
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,583

    Default

    Looks like the OCing (correction timing adjustment) of the modules was forcing the memory to run in unganged mode.
    Last edited by deanjo; 09-27-2008 at 11:04 PM.

  5. #5
    Join Date
    Mar 2007
    Location
    Savannah, GA
    Posts
    10

    Default

    Looks like the OCing (correction timing adjustment) of the modules was forcing the memory to run in unganged mode.
    I wonder if there is a benchmark that shines with improved threading where you could see the lower latency unganged memory outperform the ganged higher latency settings.

  6. #6
    Join Date
    Jun 2008
    Posts
    26

    Default

    These tests show we need a better memory benchmark.

    1. The tests confirm that for most desktop applications, memory doesn't matter. There was very little difference between timings or speed.
    2. The Corsair memory was reported as 2T. The OCZ was command rate was not reported. Back when I used optimize memory, I found command rate to have significant effect on benchmark results.
    3. Memory is only part of the equation. The memory speed and latency must not just be maximized, but must be matched to the requirements of the memory controller, cache, and CPU. Running odd or fractional speeds and multipliers may look good on paper, but may not produce real world benefit.
    4. With the integrated memory controller, my Athlon64 is much more sensitive to CPU speed than RAM speed. I believe most benchmarks will reflect this. Don't know if multicores are as insensitive to memory speed. RAM speed only matters if there are significant cache misses (on the last cache). Cache speed it tied to CPU speed.
    5. As deanjo mentioned there are other memory considerations not touched upon in the article.
    6. No memory temperature measurements were taken. Minimal voltage information was reported. No power consumption figures were listed. Of course, power only matters for heat generation, since an extra watt shouldn't tax the power supply or VRM section. Voltage only matters if your motherboard isn't adjustable. But temperature might be significant. Certainly, the heat sinks got plenty of editorial space. Was temperature related to stability? Did the heat sinks work?

    I'm sure there are (server) situations where memory speed matters, but I don't know any specifically. Maybe a web server where the cgi scripts are stored in memory for quick access?

    The review whets the appetite, but doesn't quite satisfy.

  7. #7
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,583

    Default

    Here is a article I ran across, ganged vs unganged. Very comprehensive.

    http://www.digit-life.com/articles3/...ganged-p1.html

  8. #8
    Join Date
    Aug 2008
    Posts
    30

    Default

    I have these memories for my own computer bought about 5 months back.

    I've been running them in 800Mhz 4.4.4.12.22.2T 2.1volts unganged mode.

    So should a after looking at these tests use the 5.5.5.18.23.2T ones instead? And try to OC them to 1066Mhz?
    I find it really odd the 4.4.4.12 are beaten by the 5.5.5.18 timings.
    I have a Phenom X4 9750 aswell

  9. #9
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,583

    Default

    Quote Originally Posted by Nighthog View Post
    I have these memories for my own computer bought about 5 months back.

    I've been running them in 800Mhz 4.4.4.12.22.2T 2.1volts unganged mode.

    So should a after looking at these tests use the 5.5.5.18.23.2T ones instead? And try to OC them to 1066Mhz?
    I find it really odd the 4.4.4.12 are beaten by the 5.5.5.18 timings.
    I have a Phenom X4 9750 aswell
    What motherboard are you running? Best way to to check what timings are suited to your machine is to run the Phoronix Test Suite and compare results.

  10. #10
    Join Date
    Jul 2008
    Location
    Greece
    Posts
    3,788

    Default

    And of course Phoronix doesn't even mention the paradox of 4-4-4-12 being slower.

    Bah, why am I still looking at this site, lol.

Posting Permissions

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