Page 2 of 2 FirstFirst 12
Results 26 to 42 of 42

Thread: What is the best memory for SB?

  1. #26
    I am Xtreme
    Join Date
    Aug 2008
    Posts
    5,586
    Quote Originally Posted by SteveRo View Post
    Mr. Honda was this to me?

    I have sampled seven 2600k's so far.

    Multi results - booting into windows xp32 -
    52x, 53, 55 (died), 52, 56, 55, 52.

    This was all at ud5 default bclock, HT off, mem at 1866, cpuv up to 1.61v.
    My remaining 56 and 55 might go higher with higher cpuv but I'm a little shy given that I killed one good one already (too much time at cpuv 1.66v?).
    that was for the misinformation....or as chew called it "hogwash" information


  2. #27
    Wpprriimmee crrraaazzzyyy
    Join Date
    May 2009
    Location
    Alabama
    Posts
    1,804
    Quote Originally Posted by SteveRo View Post
    What is the best testing approach for three sticks to find the best two.
    Test each one by itself or 1 with 2, 1 with 3 and 2 with 3?
    Also what is the best test -sp32m?
    Thats what I'd do and actually will be doing with three DomGT 1GB 2000C7 sticks this week.
    Test each stick individually starting at a known stable speed and see which ones can pass Superpi. You can test 32M or even less depending on the stability you want to see with each chip. 32M is tougher on the chips than to 3D and other 2D stuff.
    Heatware



    Custom SS unit by Gray Mole, Sdumper Cascade, K|ngp|n F1 Gemnni, TEK9 Fatboy, TEK9 Slim, NB-1, H2O stuff...
    Celeron 347, Celeron 356, Celeron 420, 2x P4 631, P4 670, P4 640, P4 651, P4 524, PD 915, P4 540J, 2x P4 530J, P4 520J Crucial Ballistix 2x1GB DDR2-8500 2x Asus REX X48, Gigabyte EP45-UD3P rev 1.0, Asus Commando P965,
    Stuff I used to have:
    Asus RIIIE X58, Evga Classified P55, Gigabyte EP45T-UD3LR, i5 670, E8600, Q9650, Q9550, , 3x2GB Mushkin Redline 2000C8,, G.Skill 2x2GB PC9600(Silver), Evga Evbot
    XFX 4770 modded, XFX 7950GT, XFX 6600GT, Evga 8600GTS 512MB, 3850 OC, 4x 3870s, and X600

    "In this world, Elwood, you must be oh so smart or oh so pleasant. Well, for years I was smart. I recommend pleasant." - James Stewart

  3. #28
    Xtreme Addict
    Join Date
    May 2007
    Location
    Romania
    Posts
    1,246
    Quote Originally Posted by Don_Dan View Post
    Actually I think the 2200 C7 are the higher bin, 2400 C8 is harder for the IMC, but for 2200 C7 you need better ICs imho. If you calculate true CAS for both kits, you get ~6.36ns for 2200 C7 and ~6.66ns for 2400 C8. Of course this is only a rough estimate, but since the RipjawsX are designed for SB and a lot cheaper than the PI I would go for them.



    Exactly...
    Alex, do you think these chips will miraculously get better if you test enough of them? I don't think so!
    G.Skill has already binned the best ICs for higher-end products, what you get in the lower specced products are only ICs that were not good enough for a higher bin. G.Skill is known for binning their stuff tightly.
    I have tested many high-end and lower-end D9, and although they use the same chips, there are huge differences between them. The same is still true for these PSC kits.

    Steve, just get a set of the Ripjaws X 2200 C7 first and try if those can help you achieve the settings you're hoping for, no need to buy many sets yet.

    Well you have a good point here,BUT.I had the 2200 C7 kit for SB,one friend had 2133 9-11-9 KIT for SB.It performed EXACTLY the same,scaled till 1.72V,did exactly the same main timings.So the two sets performed exactly the same.I don't know which one is being a better statement,that he got a very good kit or that i had an average 2200C7 kit,but neither of ones could do 6-9-6 32M.Too bad that i have one 2GB stick of some Exceleram that will do that,but the G.Skills were a no go.That's why i'm saying to go and bin yourself,and if you find a very good 2 sticks of 2133C9,they are fking cheap,so you actually save 200$ which you can use to buy something else,or get yourself a fine evening with a fine lady....Just saying

  4. #29
    Wpprriimmee crrraaazzzyyy
    Join Date
    May 2009
    Location
    Alabama
    Posts
    1,804
    Another thing Steve,
    Have you tried less voltage on your kit? For example 1.80v and less? Does it affect clocks and if so how much?
    Heatware



    Custom SS unit by Gray Mole, Sdumper Cascade, K|ngp|n F1 Gemnni, TEK9 Fatboy, TEK9 Slim, NB-1, H2O stuff...
    Celeron 347, Celeron 356, Celeron 420, 2x P4 631, P4 670, P4 640, P4 651, P4 524, PD 915, P4 540J, 2x P4 530J, P4 520J Crucial Ballistix 2x1GB DDR2-8500 2x Asus REX X48, Gigabyte EP45-UD3P rev 1.0, Asus Commando P965,
    Stuff I used to have:
    Asus RIIIE X58, Evga Classified P55, Gigabyte EP45T-UD3LR, i5 670, E8600, Q9650, Q9550, , 3x2GB Mushkin Redline 2000C8,, G.Skill 2x2GB PC9600(Silver), Evga Evbot
    XFX 4770 modded, XFX 7950GT, XFX 6600GT, Evga 8600GTS 512MB, 3850 OC, 4x 3870s, and X600

    "In this world, Elwood, you must be oh so smart or oh so pleasant. Well, for years I was smart. I recommend pleasant." - James Stewart

  5. #30
    I am Xtreme
    Join Date
    Nov 2008
    Location
    Warrenton, VA
    Posts
    3,029
    ^ I must confess - I've really neglected my memory
    I figured out that as a tripple on x58 this set liked high v.
    I just used my x58 dramv on the SB.
    Big assumption when going from tripple to double that it needed so much but I knew I didn't want to go higher (already at 1.88v).
    Might I be able to get higher performance at lower dramv?

  6. #31
    I am Xtreme
    Join Date
    Nov 2008
    Location
    Warrenton, VA
    Posts
    3,029
    Quote Originally Posted by Don_Dan View Post
    Actually I think the 2200 C7 are the higher bin, 2400 C8 is harder for the IMC, but for 2200 C7 you need better ICs imho. If you calculate true CAS for both kits, you get ~6.36ns for 2200 C7 and ~6.66ns for 2400 C8. Of course this is only a rough estimate, but since the RipjawsX are designed for SB and a lot cheaper than the PI I would go for them.



    Exactly...
    Alex, do you think these chips will miraculously get better if you test enough of them? I don't think so!
    G.Skill has already binned the best ICs for higher-end products, what you get in the lower specced products are only ICs that were not good enough for a higher bin. G.Skill is known for binning their stuff tightly.
    I have tested many high-end and lower-end D9, and although they use the same chips, there are huge differences between them. The same is still true for these PSC kits.

    Steve, just get a set of the Ripjaws X 2200 C7 first and try if those can help you achieve the settings you're hoping for, no need to buy many sets yet.
    What about getting two sets of the Ripjaws X 2200 C7 - for current SB - use best 2 sticks, if I am lucky and all four are good - keep all four for quad memory - what is the latest name? - maybe SB-E or LGA2011? -

    http://semiaccurate.com/2011/01/09/f...etails-appear/


    "Although the next picture might not look that different, if you look carefully you can make out some additional details here and more specifically the fact that the board has two DIMMs on each side of the CPU socket. The LGA-2011 platform is said to have quad channel memory support, so let's hope Intel has managed to take advantage of all the extra bandwidth on offer, but if the current LGA-1155 processors are anything to go by, it seems like Intel has managed to improve its memory bandwidth with Sandy Bridge."

  7. #32
    Wpprriimmee crrraaazzzyyy
    Join Date
    May 2009
    Location
    Alabama
    Posts
    1,804
    Quote Originally Posted by SteveRo View Post
    What about getting two sets of the Ripjaws X 2200 C7 - for current SB - use best 2 sticks, if I am lucky and all four are good - keep all four for quad memory - what is the latest name? - maybe SB-E or LGA2011? -

    http://semiaccurate.com/2011/01/09/f...etails-appear/


    "Although the next picture might not look that different, if you look carefully you can make out some additional details here and more specifically the fact that the board has two DIMMs on each side of the CPU socket. The LGA-2011 platform is said to have quad channel memory support, so let's hope Intel has managed to take advantage of all the extra bandwidth on offer, but if the current LGA-1155 processors are anything to go by, it seems like Intel has managed to improve its memory bandwidth with Sandy Bridge."
    By that time there may be better kits out but you could do that anyway to find the best chips.
    I'm just thinking that your kit may not need 1.88v since i have never heard of a 1.65v rated hyper kit scaling on air that high.
    Heatware



    Custom SS unit by Gray Mole, Sdumper Cascade, K|ngp|n F1 Gemnni, TEK9 Fatboy, TEK9 Slim, NB-1, H2O stuff...
    Celeron 347, Celeron 356, Celeron 420, 2x P4 631, P4 670, P4 640, P4 651, P4 524, PD 915, P4 540J, 2x P4 530J, P4 520J Crucial Ballistix 2x1GB DDR2-8500 2x Asus REX X48, Gigabyte EP45-UD3P rev 1.0, Asus Commando P965,
    Stuff I used to have:
    Asus RIIIE X58, Evga Classified P55, Gigabyte EP45T-UD3LR, i5 670, E8600, Q9650, Q9550, , 3x2GB Mushkin Redline 2000C8,, G.Skill 2x2GB PC9600(Silver), Evga Evbot
    XFX 4770 modded, XFX 7950GT, XFX 6600GT, Evga 8600GTS 512MB, 3850 OC, 4x 3870s, and X600

    "In this world, Elwood, you must be oh so smart or oh so pleasant. Well, for years I was smart. I recommend pleasant." - James Stewart

  8. #33
    Brilliant Idiot
    Join Date
    Jan 2005
    Location
    Hell on Earth
    Posts
    11,015
    Quote Originally Posted by SteveRo View Post
    Mr. Chew, sir - what memory did you use - here -

    http://www.xtremesystems.org/forums/...&postcount=428

    and here -

    http://www.xtremesystems.org/forums/...&postcount=332

    You seem to be getting very nice results!
    First is hypers binned from 40 sticks.

    Second is the gskill Flare I reccomended.

    Both had quite a bit of time put in.

    Basically unlike Alex's findings. I tested a crap load of ram first, once i found known good sets to work 1100+ at reasonable timings I started testing multiple cpu's.

    One was a fail for memory clocks, oddly enough it was also blck challenged and could barely do 100 bclk regardless of multiplier. C7 hyper and or C6 with 2133 PSC was a non issue, higher was.....

    11? I lost count were fine for 2200+, none of those were bclk challenged although some varied from 105-108 bclk.

    To do accurate IMC analysis you must first start off with known good ram

    Also when running higher or agressive ram speeds I found high vcore just makes things unstable.

    Msimax's chip we ran 32m at only 1.48v........

    Mine we ran around 1.536v
    Last edited by chew*; 02-21-2011 at 12:57 PM.
    heatware chew*
    I've got no strings to hold me down.
    To make me fret, or make me frown.
    I had strings but now I'm free.
    There are no strings on me

  9. #34
    Xtreme Memory Hoarder
    Join Date
    Mar 2010
    Posts
    2,661
    Quote Originally Posted by chew* View Post
    I know the 8-9-8 2133 gskills can do 7-9-7 but 6-9-6 2133 is a pipe dream for that model.

    I thought that the 8-9-8 2133 were a higher bin than the 7-10-7 2133 kit

  10. #35
    3D Team Captain Don_Dan's Avatar
    Join Date
    May 2007
    Location
    Munich, Germany
    Posts
    4,199
    Quote Originally Posted by l0ud_sil3nc3 View Post
    I thought that the 8-9-8 2133 were a higher bin than the 7-10-7 2133 kit
    These kits use different chips anyway, so comparing bins is not that easy.

    Quote Originally Posted by chew* View Post
    You can never have enough D9's.

  11. #36
    Xtreme Memory Hoarder
    Join Date
    Mar 2010
    Posts
    2,661
    Quote Originally Posted by Don_Dan View Post
    These kits use different chips anyway, so comparing bins is not that easy.


    really I assumed they were both psc ic's under there just different bins, but I guess not.

    I know my set can run 2200 7-9-7-24 with 1.68v, but there is no way 6-9-6 is happening with these

  12. #37
    3D Team Captain Don_Dan's Avatar
    Join Date
    May 2007
    Location
    Munich, Germany
    Posts
    4,199
    The first kits were BBSE ( the 8-9-8 one ), it's possible they changed chips to PSC by now. Your kit sounds like PSC indeed, can you check it?
    No need to pull the heatspreader, just tell me, what do the chip look like? If they have sharp edges, it's PSC.

    And what can your kit do at 800MHz?

    Quote Originally Posted by chew* View Post
    You can never have enough D9's.

  13. #38
    Xtreme Memory Hoarder
    Join Date
    Mar 2010
    Posts
    2,661
    Quote Originally Posted by Don_Dan View Post
    The first kits were BBSE ( the 8-9-8 one ), it's possible they changed chips to PSC by now. Your kit sounds like PSC indeed, can you check it?
    No need to pull the heatspreader, just tell me, what do the chip look like? If they have sharp edges, it's PSC.

    And what can your kit do at 800MHz?
    I will check when I get back to the house

  14. #39
    Xtreme Memory Hoarder
    Join Date
    Mar 2010
    Posts
    2,661
    the ic's most definitely have sharp edges, no rounding whatsoever

  15. #40
    3D Team Captain Don_Dan's Avatar
    Join Date
    May 2007
    Location
    Munich, Germany
    Posts
    4,199
    Yeah, sounds like PSC then.

    Quote Originally Posted by chew* View Post
    You can never have enough D9's.

  16. #41
    Xtreme Addict
    Join Date
    Apr 2005
    Location
    Calgary, AB
    Posts
    2,219
    PSC is cheap and works well on SB. You can get a good kit for barely over $100 that will do 2200 @ 7-9-7. Tough to find that 2200 kit of 7-7-6 hyper on SB. I would look for a proven kit from someone if that is what you need.
    MB Reviewer for HWC
    Team OCX Bench Team

  17. #42
    Xtreme Member
    Join Date
    Apr 2011
    Posts
    128
    Quote Originally Posted by chew* View Post
    Thats hogwash, tested 10 cpu's or around there.......unless they are bclk challenged they can clock ram fine.

    It's the boards that can or can not.

    Steve, you need to play the same game you played with your cpu's. Want good ram you have to bin.......ufortunately thats not gona happen with hypers anymore.

    They needed to be stocked up on quite a while ago........however your killing yourself over nothing.

    PSC is fast as well when tuned right, cheaper, requires less volts, not as painfull when they die, less stress on IMC.......And hyper is slow when tuned wrong.

    Tighter is not always faster....


    WTF! Does all those Dominators are Hypers?

Page 2 of 2 FirstFirst 12

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
  •