3500+ s939

CrunchieCrunchie Mandurah. Western Australia. Member
edited October 2006 in Hardware
Hi. Wondering what the stock LDT multiplier is for the above CPU? I am led to believe that it is 5X but the one I have is set to 4X. So if I'm correct, 4 X 200 X 2 = 1600, not the 2000 HT speed as stated.
If I change the multiplier to 5X and increase the HTT to 220 how will running the HT @ 2200 affect it? Or is it better to keep the multiplier at 4X leaving the HT @ 1760?

Sorry if it's a little confusing. My first AMD & I haven't fully grasped the terminology yet :tongue:

Comments

  • muddocktormuddocktor
    edited October 2006
    You can try the 5X LDT out with a 220 HTT and see if it's still stable; most boards seem to be to that point. And AFAIK, all socket 939 cpu's use the 5X LDT multi. But you get no real life performance improvement using the 5X LDT multi at a 220 HTT, in my experience.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    Can you think of any reason why mine would be set to 4X by default? Will it be detrimental to run it at 4X? I basically just want to OC the cpu enough to have the RAM running at it's DDR433 speed, maybe a little higher later when I'm more confident :).
  • lemonlimelemonlime Canada Member
    edited October 2006
    I believe some boards support a maximum of 4X for 939 processors. I seem to recall some of the more vanilla Nforce (non-ultra/sli) supporting a maximum of 4X. Don't quote me on that though :)

    The 'official' default for all 939 processors is 5X (1000/2000MHz). Your DFI board does support 5X LDT as it is an 'ultra' chipset. I know with my NF4, if I leave it on 'Auto' the board will select an appropriate LDT multiplier based on the reference clock frequency I select. If you increased your reference clock to 220MHz, that is likely what your board did.

    To your second question, it is not detremental at all to run a 4X LDT multiplier. Once you start overclocking, your HTT frequency will begin to approach default clockspeeds anyhow. Ideally, you want to keep it as close to the default. Overclocking the HTT bus is not beneficial.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    Cheers for the info :). Gives me some peace of mind now.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    Have the RAM running @ 1:1 which atm is 220Mhz. Ran memtest86 and am getting errors in test #8.
    RAM is only mildly OC'd and running with 2.9volts. Timings are 2.2.2.5 1T
    Anybody know what the resolution for this is?
  • muddocktormuddocktor
    edited October 2006
    Try bumping the vdimm voltage up to 3.0v and see if it will pass memtest then. As long as you have decent airflow in your case around the ram slots you shouldn't have any heat problems with your BH5. BH5 doesn't get near as hot as ram based on TCCC or TCCD chips at elevated voltages. I have a couple of sticks of OCZ Value VX that have been running at 3.2v vdimm for well over a year with no problems.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    Just been in the BIOS again and it appears that the vdimm voltage is @ 3.0v even though I have it set to 2.9. I have read somewhere (possibly on dfi-street) that these boards can have a tendency to slightly over-volt.
    Will try giving it some more anyway :). BIOS allows up to 3.2v.
    Also, whilst I'm on, I cannot find in the BIOS where I can alter the RAM ratio. I can change the cpu multi and the LDT, but not the RAM. Would it be the version of BIOS that does that, or is there another option somewhere for more advanced options?
  • muddocktormuddocktor
    edited October 2006
    Since I don't have your board, I can't help you definitively on the ram divider question. But that would be strange as hell for a performance DFI board not to have ram dividers. I'm presently at the wellsite so I can't look to make sure, but I remember the ram dividers being set in the section that also adjusts the ram timings on my NF3-250 Gb Lanparty boards (socket 754). You might try looking in that section of the bios and see if the ram dividers are hiding there. They might just be hidden if you have some settings there set to "auto" instead of manual.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    I'm a dummy :). Was looking at it without seeing it. Can set the RAM to either 100, 133, 166 or 200. The info tells me that the RAM will not exceed the speed that you set, but I assume that is when the reference clock is set to 200? Once one starts upping the reference clock, the memory clock would rise in accordance with what you have set it, yes? So 166 would rise at 5/6 of the reference clock. Is that right?

    BTW. The person I bought the RAM from said that it was good for 250+ running ~ 3.3v. Thinking I should have no problem @ 225-230. What you reckon?
  • muddocktormuddocktor
    edited October 2006
    If that is old skool BH5 from back in the day, then I would say it would be good for at least that with your mobo. If it's modern UUT based stuff, it should still be good for at least 225 with your voltage limitations on your mobo, running 2-2-2-5-1T timings IMO. I've had my old skool BH5 Corsair stuff (2 X 256 sticks) at 260-270 MHz and 2-2-2-8 timings running 3.5v vdimm with active cooling.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    How can you tell the difference between older and newer?
  • muddocktormuddocktor
    edited October 2006
    I see you have your ram listed as UTT BH5. That sounds like the later version of the Winbond ram and not the original version like my Corsair XMS has.

    Just FYI, besides my value VX, which is Winbond UTT chips, I also has a kit of the OCZ VX Gold 3200 which had 1 stick that died and also the replacement for this was VX Gold PC4000, in which 1 stick also failed on me after several months. But OCZ made it good both times and the last time I RMA'ed the VX 4000, they sent me a kit of the Plat 4800 TCCD stuff, which is their highest end 1 GB kit. OCZ's warantee coverage abolutely rocks. :thumbsup:
  • lemonlimelemonlime Canada Member
    edited October 2006
    muddocktor wrote:
    I see you have your ram listed as UTT BH5. That sounds like the later version of the Winbond ram and not the original version like my Corsair XMS has.

    Just FYI, besides my value VX, which is Winbond UTT chips, I also has a kit of the OCZ VX Gold 3200 which had 1 stick that died and also the replacement for this was VX Gold PC4000, in which 1 stick also failed on me after several months. But OCZ made it good both times and the last time I RMA'ed the VX 4000, they sent me a kit of the Plat 4800 TCCD stuff, which is their highest end 1 GB kit. OCZ's warantee coverage abolutely rocks. :thumbsup:

    I'll second that.. OCZ was happy to replace my TCCD that was simply not overclocking well. They are the best company I have ever dealt with for product issues :)

    How is the overclocking coming Crunchie?
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    I haven't gone any higher as yet with the OC as I wanted to sort out the memtest errors. PC runs fine though. No crashes etc. I ran memtest again with the RAM at stock speed and voltage and still am getting errors in test #8.
    Here are the actual modules I purchased showing what the original owner had them running at http://forums.overclockers.com.au/showpost.php?p=5981328&postcount=2
  • muddocktormuddocktor
    edited October 2006
    Crunchie, try bumping the vdimm up to 3.1v and see if you can get rid of the memory errors. Your memory is waranted for 3.1v + 5% variation, so you won't be invalidating your warantee doing this. If the memtest errors still persist, then I would get in touch with The Lost Swede here and see about arranging an RMA for you Down Under.

    BTW, here is a direct link to your ram modules. They are an end of life product, so I'm sure that OCZ will send you a healthy upgrade, such as they did when they sent me my PC4800 Plat Elites for my Gold VX 4000 that went bad.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    Hi muddocktor. Been playing around with it again :). Hit them with 3.1 volts and got more errors than ever :(. Tried a couple of other things that made no difference.
    Have it back on 2.8 volts with 2.2.2.5 and 2T timings. It's on the 2nd pass and so far no errors :).
    Question I have now is, what sort of difference am I looking at in performance between 1T and 2T (and if you can explain what 1T and 2T are :D) on this PC? Should these modules be rated to run at 1T?

    Thanks a lot for your time too, I appreciate it :thumbup
  • muddocktormuddocktor
    edited October 2006
    Hey, it's no problem helping you out, Crunchie. After all, that's one reason why we come to Short Media and we all have our strengths and weaknesses in helping people out. :)

    As far as running stable with a 2T command rate with your memory, I'm tending to lean more towards a board problem than a memory problem now. I seem to remember that DFI had all sorts of problems with their socket 939 NF3 based mobo and quit manufacturing it shortly after it was introduced. And if I remember right, having to run at a 2T command rate with only 2 memory modules might have been one of the problems that board encountered. But in real life use, having to run a 2T command rate doesn't actually slow down your computer very much.

    Here is what TechARP has to say about the 2T command rate:
    TechARP wrote:
    2T Command

    Common Options : Enabled, Disabled, Auto

    Quick Review

    This BIOS feature allows you to select the delay between the assertion of the Chip Select signal till the time the memory controller starts sending commands to the memory bank. The lower the value, the sooner the memory controller can send commands out to the activated memory bank.

    When this feature is disabled, the memory controller will only insert a command delay of one clock cycle or 1T.

    When this feature is enabled, the memory controller will insert a command delay of two clock cycles or 2T.

    The Auto option allows the memory controller to use the memory module's SPD value for command delay.

    If the SDRAM command delay is too long, it can reduce performance by unnecessarily preventing the memory controller from issuing the commands sooner.

    However, if the SDRAM command delay is too short, the memory controller may not be able to translate the addresses in time and the "bad commands" that result will cause data loss and corruption.

    It is recommended that you try disabling 2T Command for better memory performance. But if you face stability issues, enable this BIOS feature.
    The quote is from this page here. TechARP's complete bios optimization guide can be found here. That is a good site to bookmark as they cover just about any option you might run into when you are adjusting settings in your bios.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    Thanks for the advice :). After I purchased the board I stumbled across posts here and there regarding problems it had/has. I should have seen them first :D.
    Anyways, it's a good starting point for me on the amd side of things. There's a bit more to play with than on my Intel Asus board.
    Having said that, I'm off to play :D.
  • CrunchieCrunchie Mandurah. Western Australia. Member
    edited October 2006
    No matter what I try, I cannot get it to run error free on 1T command rate. 2T it is then :).
    Running Everest, there is only a slight performance dip anyway, so me happy.
Sign In or Register to comment.