RAM voltage too high, can't lower either

Discussion in 'General Hardware' started by 0rpheus, Oct 14, 2009.

  1. Flopster

    Flopster Banned

    Messages:
    3,026
    Likes Received:
    0
    GPU:
    Club3D 8600GTS 256MB
    Yeah, your probably all set:p

    Though later on, you could go for 3.6ghz:nerd: Though tampering with voltage is required for that, but anyhoo, it's your choice...
     
  2. 0rpheus

    0rpheus Active Member

    Messages:
    67
    Likes Received:
    0
    GPU:
    Gigabyte nV 9800GT 512MB
    So, I changed the timings to 5-5-5-15 (so it's basically CL5 now...?) and ran Prime95 for 2 and a half hours without any errors or warnings. This kit doesn't seem to be as bad as some of you made it out to be :) Of course I'll have to run the test a lot longer, probably tomorrow, but for now it looks fine.

    Oh, and while I'm at it, another question: is the 10.5x 320 FSB (SPD is 2.50) combination faster than 12.5x 266 (SPD is 3.00)?
     
    Last edited: Oct 15, 2009
  3. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,098
    Likes Received:
    50
    GPU:
    RTX 2080Ti
    You running small FFT's, Large or Blend? Two and a half hours isn't really very long for prime, most will run for around 8.

    For your other question the only way to find out is to try it :)
     
  4. 0rpheus

    0rpheus Active Member

    Messages:
    67
    Likes Received:
    0
    GPU:
    Gigabyte nV 9800GT 512MB
    Oh yeah. I automatically click "OK" and don't actually look, sorry, I should have mentioned this. I ran the "Blend" test. Which one would you recommend? I'd also like to test the CPU's OC stability while I'm at it, so it would be nice if I could stress them both, RAM and CPU.
     

  5. gampamu

    gampamu Ancient Guru

    Messages:
    1,628
    Likes Received:
    0
    GPU:
    MSI 8800GTS 320
    At identical CPU and DRAM clocks, higher FSB almost always produce better results, particularly the memory bandwidth.

    Prime blend is a good CPU/chipset/memory stress test. But a lengthy test is required to verify stability. A minimum of an overnight run is recommended.
     
  6. 0rpheus

    0rpheus Active Member

    Messages:
    67
    Likes Received:
    0
    GPU:
    Gigabyte nV 9800GT 512MB
    OK, I ran the Prime95 Blend test for 12 hours 20 minutes, which should be more than enough, and didn't have any errors/warnings, so I think the RAM is ok.

    Now, I have another question. When I used the 10.5x 320 FSB combination and ran the Prime95 "Large" test, after 9 minutes I got an error on one of the "workers" (I suppose it's a core), the other one continued the test- "FATAL ERROR: Rounding was 0.46875, expected less than 0.4" I did some research and found some topics in various forums about it, and in one of them some guy said that it's "overspeed". What does it mean? And is there a specific fix for this (like increasing the Vcore or something..?) or do I have to use the trial and error method here?
     

Share This Page