Jump to content


  • Content Count

  • Joined

  • Last visited

About NonDecepticon

  • Rank
    New Member
  1. I sometimes run into the "cannot find ntloader" I think it's called as well. The ULi controller is sketchy especially when running a RAID array. After switching my Hitachi's to SATA150 to no avail, next step is just reformatting on the SI3114 controller. I'll have to split up the RAID because the 3114 is on the archaic PCI bus which saturates easily. You could try an RMA and hope for the best, but all DFI did for me was send my board back without so much as an explanation. The timeout errors actually seem worse after getting my board back. I'm done with this POS controller. Soon the motherboard as well. Since you've only tried the 2.20 package so far, try these and setting PCI latency to 64 or 32. SB voltage to 1.85v, etc... something might work for you. My A00 board is hopeless.
  2. If it had just used an ICH7 or something, it would've been worth the $227 I spent on it...
  3. Cool heatsink is not always indicative of cool heat source. This sometimes means the thermal joint is not proper or the heatpipes are not working (I've only heard of Zalman pipes ceasing to function though). It could be a really concave/convex IHS, either on the internal or external surface, the mounting mechanism is failing to provide level pressure when the case is upright (laying the case down or pressing against the heatsink to counteract the lever force can reveal this). Or it could just be borked thermal diodes. My Tuniq gets quite warm when I am running Orthos and Coretemp/A64Info/Coretemeter all report the same 53-54 degrees at 2800/1.55v (crappy 0551 UPMW).
  4. uwackme discovered that heat "blankets" can actually hinder Winbond clocking. It's all about airflow over the PCB where the heat from the chips is dissipated. 3.3-3.4v is a good 24/7 range with some quiet airflow. Actually the chips can get fairly warm without airflow even at 3V JFYI. Also if you mess with the jumper, it's probably emphasized in any guide about it, but just remember to have the BIOS set to more than 3V when that jumper is set to 4V max (particularly an issue when flashing as the CMOS default of 2.6v could damage the VReg quickly, unless they removed options lower than 3V in later BIOS versions at the higher jumper setting). Any lower could fry the VReg as it draws from the 5V rail in that position and the larger differential is problematic for it.
  5. I can confirm it happens with x64 as well. I'm not sure if the Vista 64-bit drivers help as I didn't run them longer than a day because it seemed to cause some BSoD's and strange hard drive parking sounds at shutdown. It's quite sad that NVIDIA is all over fixing the same issue with its 680 chipset but leaves us in the dark just because it wants to hurt ATi.
  6. Man I know how you feel, sometimes it's like finding the zen point or just getting the best Jenga piece out and the thing crumbles :mad: This phenomena especially affects the K8 processors somehow, it could be the on-die memory controller, could be something screwy with the SOI process *shrug* Sometimes it can be something simple like the chipset fan crapping out or a barely visible coat of dust on heatsinks. It could even be a faulty cap, are any of those suckers bulging? One way to save some time, instead of booting into windows and going through the whole Prime thing, Memtest can be very useful. This board has it right in the bios (bottom of "Genie" page) so you don't have to mess with removable media (that alone has saved tons of time compared with my previous NF2 board). I like to run 50 loops of #5 (hit C, 1, 3, 5, enter, then 0 to loop #5 exclusively), then 1-4 hours of the full suite, and finally 5 loops of #8, all consecutively. #5 is pretty reliable enough though, even just 25 loops if you need to use the machine. If it produces errors, it can be stable in Windows, but I never risk it, especially for 24/7 operation. If it freezes or reboots, it may need more drive strength on the memory or more VCore (or anything, tweak that badboy!). Then after this, I like to run two instances of SuperPi 32M or in cooler weather, 1/2 hour of OCCT, and if it passes without issues, then I go for the Prime gold. Those two shorter tests save me a lot of time for finding settings that could possibly run Prime longer.
  7. My 165 does the same "random stability" bit if Max Async Latency and Read Preamble are too tight at >285HTT. Try MAL9/RP6 to start with. Doesn't really affect performance luckily. It might be something else with the RAM or the CPU suddenly deciding it wants more VCore though.
  8. Shoot, my memory was foggy on the max DS for the SLI/Ultra-D :tooth: I know Winbond chips traditionally like the even numbers and Micron/Samsung/Infineons like the odd numbers, but you could try 7 or 5 just to see how it works. Also try the other colored memory slots. Some boards just work best when either the orange or yellow slots are used with Winbond chips. MAL = Max Async Latency which is an interesting option that on my setup comes into required tweaking around the upper 280's to have at 8ns. Strangely if I use a divider to keep the mem around 200MHz while at those HTT's, there aren't any stability issues at 7ns, just when the memory is up above 216 at that HTT. Dyn. ctr. = Dynamic Counter, this one is also a head scratcher, some systems/memory prefer it disabled, others enabled. ICL = Idle Clock Limit which I believe at lower numbers means slightly more power consumption and possible stability issues, but a bit better performance and Winbond chips prefer these lower numbers.
  9. Perhaps it needs more drive strength, my KHX3000 BH-5 (2-2-2-5 at 2.62v, haven't visually confirmed by HS removal though ) is very sensitive to DS. It needed a setting of 10 to do 245 and 12 at 250. Just food for thought, hope that helps Also my BH-5 sticks like a tREF of 4708, 3120, 3684, or 1560. Try MAL 8, dyn. ctr. disable, and ICL 16-64.
  10. I'm also curious what is the best aircooler that uses the bolt-through retention for K8. Been looking at the Tuniq Tower but it's always OOS. Is the BT (non-VX) the best "bolt-er" ?
  11. Yeah, same here, BSoD on boot (XP SP2). LKGC didn't work for me, so I had to boot into my x64 install to switch the SYS file back, no biggie. I appreciate the effort, SuperBORG It's more help than the faceless chipset corporations show us.
  12. A00 here as well. Happy it works for ya though Silva Good news is the 32 byte SB payload seems to be helping, but I've been spending most of the time Memtesting some voltage curves for two extra BH-5 sticks so it could still be problematic. Thanks Nem. update- got the first m5288 error since changing the payload size. Interestingly, it happened while the machine was idle and I was AFK so I don't know if anything froze.. :confused:
  13. You would have to desolder the leg(s) that are attached to the PCB. Don't ask me why they're soldered on. It's been preventing me from probing VDimm on my CFX as there aren't any visible areas to probe on this board.
  14. The 15Meg file CFX3200_M1575_Ch_F6.zip for some strange reason is the 400KB m5288 drivers with the ULi integrated installers versions 2.13 and 2.14 (~7.6Meg each) *scratching head* No need for those to install the m5288 drivers. You can download the 2.20 drivers from ULi or NVIDIA.
  15. Aye, and I spent $215 (227 after shipping) as I had already waited two months on the dang board to hit the market. Given the choice again, I would've gone with the Expert and 7600GT/7900GT, but I'm still happy with the amount of tweakability. The other boards sporting RD580 don't have nearly the same. I honestly thought SATA was so mature that ULi would've gotten it right, but NVIDIA swooped in and mucked things up. If I had the smarts I'd reverse engineer the Asus board's BIOS to see how they fixed it through the BIOS without a special driver.
  • Create New...