Jump to content


  • Content Count

  • Joined

  • Last visited

Posts posted by DemonEyez_NL

  1. Thanks ;)


    I got mine stable at 280MHz.. will post if you want to see them. I got the same stepping as you.. wish it weren't so damn picky on the memory dividers.

    280MHz!? :eek: Damn

    Yes, please post you're settings because whatever I do, my sticks will not run higher than 255MHz, even when i'm running one on one, without a divider.

    I'm at the point, that I'm pulling the hair out of my skull of frustration :sad:

  2. I was wondering if it were possible someone could mod the 28/7 bios so that the DRAM Response Time option (normal, fast and fastest) is available.

    In the current CFX3200-DR biosses it's missing. It's an option that always was present back on the DFI nF4 platform.


    Reason I'm asking, is that with my Crucial Ballistix i have serious trouble to get it over the 255MHz, even on 3-4-4-8 timings

    I had the same problem on my DFI nF4 Ultra-D, but after changing DRAM Response time from Fast (default) to Normal I was able to run 275MHz. So this option seems to have a lot of influence on stability when running with Crucial Ballistix PC4000 (Micron 5B -F chips).

    However, the option is not available on the CFX3200-DR

  3. Send 'em back to Crucial for an RMA.


    Demon, do you have your timing settings with you? We have nearly identical setups (CPU, RAM, board, nearly the GPU ;)), and I am having trouble running this RAM on a divider.

    I'm kind of having trouble with running dividers as well now I have an Opteron 165.


    Those timings I posted earlier were with my old X2 4600+ running 1:1 at 11x260HTT


    I got my Opteron 165 last week and it does 3000Mhz with 1.44v :)

    Right now I'm running with 150 (3:4) divider at 9x333HTT, my Crucials are running at 250MHz 3-3-3-8-1T

    This is all Memtest and Prime stable, so far so good :)


    The problems arise when going over the 255MHz on the ram (with or without a divider). I can't seem to get 260-266MHz stable

    One other thing I noticed, is that my bandwidth is rather low when running the sticks at 250MHz 3-3-3-8-1T

    Sandra reports 6400MB/s in the buffered test, in comparison with 6700-6800MB/s I find in reviews of other Micron 5B- F based 2GB kits


    Here is as you requested my stable settings :


    [B]Genie BIOS Settings:[/B]
    FSB Bus Frequency.............................  -  333
    LDT/FSB Frequency Ratio.......................  -  4
    CPU/FSB Frequency Ratio.......................  -  9
    RD580 HT PLL Speed............................  -  High
    HT Bus MMOS Drive Strength....................  -  01
    HT Bus PMOS Drive Strength....................  -  04
    HT Bus Receiever Impedance....................  -  10
    CPU HT Bus Drive Strength.....................  -  Strong
    CPU VID Control...............................  -  1.475v (gives 1.44 in WinXP)
    CPU VID Special Control.......................  -  auto
    DRAM Voltage Control..........................  -  2.74v
    SB PCIE Voltage...............................  -  1.84v
    NB Analog Voltage.............................  -  1.34v
    LDT Bus Voltage...............................  -  1.34v 
    NB Core Voltage...............................  -  1.34v
    [B]DRAM Configuration Settings:[/B]
    DRAM Frequency Set............................  -  150=RAM/FSB:03/04
    Command Per Clock (CPC).......................  -  Enabled
    CAS Latency Control (Tcl).....................  -  3
    RAS# to CAS# delay (Trcd).....................  -  3
    Min RAS# active time (Tras)...................  -  8
    Row precharge time (Trp)......................  -  3
    Row Cycle time (Trc)..........................  -  7
    Row refresh cyc time (Trfc)...................  -  14 
    Row to Row delay (Trrd).......................  -  2 
    Write recovery time (Twr).....................  -  2 
    Write to Read delay (Twtr)....................  -  1 
    Read to Write delay (Trwt)....................  -  2 
    Refresh Period (Tref).........................  -  3072 Cycles
    DRAM Bank Interleave..........................  -  Enabled
    Errata 94 Enhancement.........................  -  Disable
    Errata 123 Enhancement........................  -  auto
    Odd Divisor Correct...........................  -  Disable
    DQS Skew Control..............................  -  Off
    DQS Skew Value................................  -  0
    DRAM Drive Strength...........................  -  level 5 (always weak)
    DRAM Data Drive Strength......................  -  Level 2 (Reduce 30%)
    DIMM 1/3 Clock Timing Skew....................  -  auto
    DIMM 2/4 Clock Timing Skew....................  -  450 delay
    Max Async Latency.............................  -  8
    Read Preamble Time............................  -  6
    IdleCycle Limit...............................  -  16 Cycles
    Dynamic Counter...............................  -  Disable
    R/W Queue Bypass..............................  -  16 x
    Bypass Max....................................  -  07 x
    Burst Length..................................  -  Disable (4 Bursts)

  4. Are the 5B-F Microns as fast as the 5B-D?

    No, they are not. Crucial is playing on safe so it stopped using Micron 5b-D and is now using Micron 5b-F, which are more stable but can't take as much.

    For running Micron 5b-F 275MHz @ 3-3-3-7-1T you'll need 2.80-2.85v in comparison to 2.7v for the Micron 5b-D


    There are some reports of failing TD1. I think that's because although it's the new revison of the PCB, those sticks still use Micron 5b-D chips and not the new Micron 5b-F chips.


    I myself bought two TD sticks last year in March. Around the beginning of december they finally started to give me problems. I did a RMA and am now running with TF2Y, which has a new revision of the PCB and has Micron 5b-F chips.


    Running 260MHz @ 3-3-3-7-1T with 2.78v Drive Strength 5 and data Strength 2


    (it has been hell to work out the correct settings in my CFX3200-DR to get it stable)

  5. To me it seems that messing with Pmos, Nmos and HTTBus Reciever only effects the stability of the CPU and the LDT speed, and has really no effect on the Uli errors.


    I've been running 15-15-25 but those settings are not stable when Priming at 11x257HTT @ 1.488v and 5x257 LDT


    I've read on Xtreme that keeping Pmos higher than Nmos can help a lot when running a high HTT en high CPU clcok.

    It was suggested that 4-1-10 is a good setting when running high HTT with TCCD based sticks. It works for me, but only if do 4x257LDT, 5x is not possible (then again, I don't have TCCD in my system but Micron 5B-F)


    At the moment I'm doing 11x257HTT @ 1.488v and 5x257LDT with 5-5-10 and it works like a charm :)


    I'm still searching for a good guide (or any info for that matter) that explains how Pmos, Nmos and HTTBus Reciever have to be adjusted/are related to different CPU and Ram combo's

  6. Just wanted to let you people know that it IS possible run with both a Raid controller add-on card and with the ULi (non-raid) SATA ports enabled with bios 28/7


    To work around the m5288 freezing problem with two SATA2 disks, I decided to buy a PCIe 1x Raid SATA2 card for 40 euro's to connect my two SATA2 Seagate Barracuda's in Raid0.


    But I've also got an Samsung SATA burner, which is connected to port 4 of the Uli controller. I disabled the Raid function of the Uli in the bios and changed the SATA-mode to "AHCI".


    During boot, the sub-bios of my add-on Raid card is loaded, but NOT the Uli sub-bios, so I don't get the "Option ROM can not invoke error" :)


    Tried it again, but now with the SATA-burner connected to the Silicon Image controller. But that didn't work, even when you're not using the Raid function of this controller and the Silicon Image chip is set to non-raid mode, the sub-bios is still loaded, and you'll end up with the invoke error.


    I ended up with the following:

    Raid Add-on card + Uli non-raid AHCI mode -> Works

    Raid Add-on card + Uli non-raid PATA mode -> Works

    Raid Add-on card + Silicon Image non-raid mode -> Invoke error

    Raid Add-on card + Uli Raid enabled -> Invoke error


    Silicon Image non-raid mode + Uli non-raid AHCI mode -> Works

    Silicon Image Raid enabled + Uli non-raid PATA mode -> Works

    Silicon Image non-raid mode + Uli Raid enabled -> Invoke error

    Silicon Image Raid enabled + Uli Raid enabled -> Invoke error


    I noticed one other thing, even with the Uli SATA functions totally disabled in the bios, my SATA is still recognised during boot and works fine in Windows

  7. Mailed DFI tech support last friday about the option ROM can not be invoke message this what is wrote and what i got back:

    Dear DFI,


    When will there be a new beta bios release for the DFI lanparty UT CFX3200?


    In the latest beta bios (28/7), main problems with the Uli South Bridge seems to be resolved for some users. :)


    However, with this latest beta bios it’s not possible to enable both the Silicion Image controler and ULi SB Raid controler!


    Doing this results in a “Option Rom can not be invoked” message during boot.


    That’s basically a warning telling that there’s not enough room in the main bios to load both sub-bioses for the Silicon Image controller and the Uli SB controller, resulting in a not-working Raid array en dissapearing SATA controller in Windows


    When will this issue be resolved? I want to use the SATA ports from BOTH chips, and I was also planning to use a PCIe 1x add-on Raid controller card to use for my two SATA2 disks instead of the Uli Raid controller.


    But again, when using one of the two onboard Raid controllers together with an add-on Raid card, I still get the “Option Rom can not be invoked” message.


    Kind Regards,




    They replied as follows:

    Dear User: Thanks for your enquiry.


    "Option ROM cannot be invoked" was caused by the physical limitation on CMOS, therefore no matter users try to use both of RAID controllers or use one on-board RAID controller and one add-on RAID card, both situations will try to load two RAID controllers into CMOS during boot up stage and when CMOS cannot fit in with second RAID controller, the "Option ROM cannot be invoked" message will appear.

    This issue is related to the physical limitation on CMOS, update BIOS will not help users to eliminate this issue.


    Very strange, because in beta Bios 5/19 the issue was resolved, only to come back in the later bios releases.

    I find it very strange to say the least that they not willing to resolve the issue again. :(

    Because what does the release notes say with this bios:


    Major Reasons of Change:

    1. Fixed Silicon image ( SATA) + Uli (Raid ) combination.

    2. Turn on ULI and silicon image raid at same time but Silicon image just support Raid5 only at the combination .

  8. Dear Rgone thanx for sharing the info, but what about the "Option ROM can not invoke" error?

    In the latest beta bios (28/7) it's still not possible to run the Silicon chip together with the Uli chip enabled.

    When will that issue be resolved?

    I mean, the only thing that the error is telling you, is that there's not enough room in the bios to load both chip biosses, surely that can be resolved

  9. Can DFI HQ's get both controllers (ULI + SIL) to work together, that would be a good check..........
    Yeah, I was wonbering about that myself. In the latest beta (28/7) there's still the Option ROM can not invoke issue. When will that be resolved???

    Now I'm not using the Sil right now, but to avoid conflicts in the futute with SATA2 drives I was planning to buy a PCIe 1x Raid0 Controller add-on card.

    It's my understanding that you also cannot run with the Uli raid bios enabled and with an Add-on card Raid bios at the same time, it produces the same error as with Uli + Sil.

    I mean, the only thing that the error is telling you, is that there's not enough room in the bios to load both chip biosses, surely that can be resolved

  10. It would be great if you gave people a step by step guide on doing what you did.



    I thinks it's very simple. Before you do a install of Windows (Clean or Repair), first disable Marvell LAN port 2 in the bios and have the Uli Raid function enabled. Put you're graphic card in PCIe 16x slot one (upper slot).

    Now, let Windows install and you'll end up with The Marvell LAN port 1 on IRQ 18 and Uli Raid Controller on IRQ 19.


    Btw, where can I change the PCIe bus speed, I can't seem to find it anywhere? I'm using bios 28/7

  11. It SHOULD be possible. As long as your DVD drive is recognised when the ULI is in RAID mode.


    The RAID BIOS allows you to pick which drives are in the RAID array.

    You were right! :) In the Uli Raid Bios config, my Samsung SATA burner is recognized and works perfectly.


    I'm running the following combo:


    Bios 28/7 + Beta

    Uli Raid0 (Stripe 64K) 2x 120Gb SATA1 Maxtor

    Silicon Image Disabled


    No freezes at all :D

  12. I wonder which revision of mobo's you people have, perhaps it has to do something with that.


    I want to buy on since I saw 'em for about 69 euro's (:eek:) at a trustworthy store .. Is this problem with the SB very big (affecting the performance in general)? I'm going to take a closer look at this board because someone stated that this board doesn't OC that good, which'd suck since I've wanted a DFI-board over past the 2 years..



    How much does it cost in other countries btw,? Here the latest boards left are sold for about €200 (except for the board that I: saw), which is also the price they where introduced at..

    I just bought this motherboard last weekend, also for 69 Euro's. Comming week most webshops in Holland will drop their prices on the CFX3200.


    As far as the Uli SB is concerned. Just disable the Sata and Raid control functions and you're fine. If you really want to use SATA2, then you're better of buying an PCIe 1x controller card. Overall performance of the Uli Sb isn't that bad at all, it just has issues with SATA2 and Raid.


    I don't know if you are familair with Tweakers.net and Gathering Of Tweakers.net , but in this Subforum there a dutch thread concerning the CFX3200.



    People are reporting in the Dutch thread that the Uli SB has no issues when using SATA1 (in Raid or not) with the new AB0 revision. Even some are reporting that SATA2 is also possible with no problems at all with this latest revision.

  13. @ jhyunl


    Did you play with Max Async Latency, Read Preamble , Drive/Data Strength Settings and DRAM response time?

    Because those need to be changed when you get past 250MHz. It can really make a big difference in getting no errors at all or getting ten-thousends of them. Leaving them on auto will generate errors in Mem Test for sure


    My settings @ 237MHz (2.5-2-2-6-1T) vdimm 2.57v

    Max Async 7

    Read preamble 5.5

    Drive Strength 5

    Data Strength 2

    DRAM response time fast


    My settings @ 273Mhz (3-3-3-8-1T) vdimm 2.71v

    Max Async 9

    Read preamble 6

    Drive Strength 6

    Datsa Strength 2

    Data response time normal


    I'v been using these settings since January this year when I first got my two sticks of 16TD's and there still going strong, haven't had any problems so far. (I run Mem Test every other week)

  14. Ok I just got some of these not the tracer kind. But anyhow I went into bios to change timings and wouldnt boot back up. Wouldnt get past 2 led light on the board. Put some older ocz mem in booted right up. I put the setting at 250 3,4,4,8 timings I only was really able to run it at 240 with 3338 o and both were at 2.6 volts. It was the first day I had it didnt get a chance to even run mem test that was what i was going to do when it locked up. Anyone have any ideas? Also when it was running it wasnt stable locked up twice in about 10 min. Any ideas?



    not sure of date code of bios will look at when i get home

    What worked for me was the following: make sure DRAM RESPONSE TIME is to Normal when you run those sticks at 240Hz and up. Fast and Fastest are to much for these sticks

  15. Did you stop to think about the fact that this card is PCI based and operates at 33/66Mhz?


    A PCI busspeed of 66MHz gives you 266MB/sec burst data transfer rate. That's just enough for two SATA2 disks in Raid0


    However..... the pci slots on you DFI mobo only operate at 33MHz.

    A PCI busspeed of 33MHz gives you only 133MB/sec burst data transfer rate.


    PCI 66Mhz is NOT supported on your motherboard


    So using two SATA2 3.0Gb disks with this combo is useless.


    You'd be better off looking for an PCIe x1 controller card.

    A PCIe 1x busspeed gives you a 312MB/sec burst data transfer rate.





    HighPoint RocketRAID 2300 (4x SATA300, RAID, PCI-e x1



    Dawi Conroll DC-300e RAID - PCI Express x1- Serial ATA II RAID


  16. Another thing you might consider is this:

    The DFI Expert has issues with TCCD based memory modules, if you have TCCD based modules I would leave the Expert alone and go for an Ultra-D (SLI/DR) or Venus board.

    On the other hand, if it's memory with UTT or BH5/BH6 chips you could go for the Expert.

    Hope this helps :)


    Btw, If I were you I'd use 2 sticks of 1024Mb and not 4x512Mb. With two sticks of 1024MB, changes are a lot higher you reach running them at 1T


    I'm following a topic on Xtreme forums where a well known user, Windwithme, is testing the board. He succesfully ran his two sticks of 1024MB at 1T (doing 620DDR :) )

    He has yet to succeed in running 4x512 at 1T. At the moment it's still 2T

  17. what makes you think that "overclockability has been enhanced"? there is no data yet to make such a statement.

    Perhaps you should do some reading before you come to a judgement:




    In Nov/Dec 2005, DFI launched ATI RDX200 and nF4 SLI EXPERT 2 great OC boards.

    In 2006Chinese New Year, DFI is going to announce this 100% Aluminum Capacitors board.

    As the pioneer of 100% Japanese capacitors, Venus, this new model, is the other great work!


    Lets look some pics. Basically, the design is similar. The components is greater and enhanced OC ability.








    Just got it, I am still trying. First impression is the OC ability is better.

    Ultra-d brought the TCCD limitation. EXPERT unleash the BH5/UTT.

    What Venus Brings? Let?s wait for it.


    DFI strongest items: BIOS part

    CPU and DRAM voltage parts are the same as EXPERT.


    DRAM configuration part: The most items in all brands. This is why DFI?s OC is always the best.


    Recommend TCCD using LEVEL3+4 UCCC using LEVEL1+1


    Test configuration

    AMD 939 Opteron 144

    DFI LANParty UT nF4 SLI-DR Venus

    GSkill 1GBFF/2GBHZ

    ELSA 6600GT

    MAXTOR 30GB 5400rpm

    Enlight 420W


    EXPERT has weak compatibility with TCCD. Cannot reach DDR550~600 with some memories.

    In Venus, DFI re-layout it. And reappear the own glory of DDR600 1T!


    2GB is the mainstream. Usnig 2GBHZ DDR500 3 4-4-8 to combined test.


    2GB DDR600 1T 3 4-4-8 PI 32M Reached!

    OC is not in manufacturers guarantee. No matter CPU(200Mhz) / MB(200Mhz) / DRAM(DDR500). *


    Venus really solves some previous issues. Bringing the OC surprise again.


    Fine gift box and 1000 limited output globally. Maybe will be hard to buy it...



    1. 100% Aluminum Capacitors

    2. PCI-E wider slots, better for SLI cards heat

    3. Fine Box

    4. DFI traditional BIOS high voltage

    5. 6 USB in real panel

    6. Improve EXPERT weakness in TCCD and enhance UCCC



    1. High price

    2. Not dual PCI-E 16X(3D performance are similar at the moment with dual PCI-E 8X)

    3. RAID with SATAI 3114(nF4 built in SATAII RAID)


    OSKAR hits home run again. DFI Venus break the many motherboard records.

    Lets look forward to it.


    OMG! 2GB (2x1024) DDR610 3-4-4-8 @ 1T :9~


  18. Perhaps you try to increase vdimm to 2.8v (gives 2.9v in bios, but MBM5 shows 2.84- 2,87v under load in windows)


    I use OCZ P4800 Platinum Elite Edition which does 300MHz (600DDR) @ 2.5-4-4-8 and is rated 2.8v at those speeds. (heck , it can even do 310Mhz at those settings)


    First I could not do 320HTT (5:6) 255Mhz with your settings described above, PiFast and Prime kept on failing and MemTest gave me around 30 errors per pass in test 8. This was with 2.7v for the dimms


    After increasing the vdimm to 2.8v all the errors disapeared and Pifast and Prime were stable. Very strange indeed , but it seems to me that this board needs anough juice on your dimms to do a stable high HTT , evenwhen your ram is running slower with a divider (as is my case)


    I tested again running with 2.7v and the errors reappeared


    in addition you can try these settings:


    Max Async latency 10ns

    Read Preamble 8ns

    Dynamic Idle cycle counter 256

  19. I'm running with an Opteron 146 and 2 sticks of 512MB OCZ PC4800 Platinum Elite Edition (TCCD)

    Seems to me that the most influential settings for running a high HTT and 2 sticks of 512MB are the 1T/2T settings and Read Preamble and Max Async latency. For Opterons with high HTT i set Read Preamble to 7ns and Max Async to 9ns, and offcourse 2T for a HTT of 260 and above. Also i set vdimm to 2.8v (gives 2.9v) My sticks are rated 2.8v for 300MHz 2.5-4-4-8


    These are my setting for 10x300 @ 2T


    bios C19

    Orange slots

    vchipset 2.8

    vdimm 2.8v

    Max Memclock 200MHz

    1T/2T Memory Timing 2T

    Tcl 2.5

    Trcd 4

    Tras 8

    Trp 4

    Twr 2

    Trrd 2

    Trfc 16

    Trc 12

    Twtr 2

    Trwt 4

    Tref auto

    Twcl 1

    Max Async latency 8ns

    Read Preamble 6ns

    Dynamic Idle cycle counter 64

    DDR output driving weak

    DDR DQ Drive strength 50% reduction

    Bank Interleave enabled 2-way


    These are my setting for 9x325 @ 1T


    bios C19

    Orange slots

    vchipset 2.8

    vdimm 2.8v

    Max Memclock 166MHz using 6:5 divider

    1T/2T Memory Timing 2T

    Tcl 2.5

    Trcd 3

    Tras 7

    Trp 3

    Twr 2

    Trrd 2

    Trfc 16

    Trc 12

    Twtr 2

    Trwt 4

    Tref auto

    Twcl 1

    Max Async latency 9ns

    Read Preamble 7ns

    Dynamic Idle cycle counter 128

    DDR output driving weak

    DDR DQ Drive strength 50% reduction

    Bank Interleave enabled 2-way

  • Create New...