Jump to content
HWBOT Community Forums

TaPaKaH

Members
  • Posts

    3657
  • Joined

  • Last visited

  • Days Won

    29

Posts posted by TaPaKaH

  1. Yes and no. There are some chips that keep scaling on FSB all the way up to their actual coldbug (as in, lowering FSB won't lower the coldbug) but most chips indeed peak out between -120 and -100 with reverse FSB scaling (similar to Athlon 64 where it is extreme) to be observed when going lower.

  2. core on air (LinX): 3600 1.40 for an acceptable chip, 1.35 for a good one, 1.30 for an awesome one
    core on air (spi 32M): 3900 1.40-1.45 is good for B2, 4000 1.40-1.45 is good for G0
    FSB on air: 550 acceptable, 575 good, 600 best
    CB is usually -140 to -120

    Note that low-multi chips (E6300) don't really need good core and high-multi chips (E6600+) have weaker requirements on FSB.

    • Like 3
    • Thanks 1
  3. I tried experimenting with this multiple times in the last decade, but I don't think you can do much better than Hypers on X48. Ultimately, it's all a matter of convenience - the only way to beat ambient Hypers is running cold on your memory. However, given how many more variables it adds to the troubleshooting procedure, how poorly-designed X48 boards are from reliability standpoint, and how difficult good boards are to replace in case of death, it's not really worth the effort.

    Still, if we want to have a theoretical discussion:
    D9GTR: single-sided sticks usually need high tRFC (60+) above 1025-ish and 1000+ capable dual-sided sticks are impossible to find (I wasn't able to do so in 11 years)
    BBSE: on ambient you're likely to hit the same IMC cap as Hypers while running looser tRCD, on cold you can probably do tRCD 6 at 1000-1050 which could be interesting (given that BBSE also like tight-ish tRAS and tRFC)
    BDBG / PSC: even on cold you're not likely to go below tRCD 7 at 1050, so you'll be losing to ambient Hypers, especially given that BDBG/PSC run looser tRAS and tRFC

    • Like 3
    • Thanks 1
  4. Then I would say that a separate category is useless since there is no way to verify which version was used.
    Also, I'm not sure if there is any actual practical (=overclocking) difference between the two versions - I had some EE versions of X2 chips and they seemed exactly the same.

    • Like 1
×
×
  • Create New...