shimon

Even so, I'd like to know how power is delivered to the CPU. At this speed power consumption will be in the hundreds of watts. The N2 cooling will have to include much of the MB. Even the memory sounds emazing at 400+ MHz. Anyway, I thought I was doing well at 3.15 (running anything, including MP3 encoding) without errors for weeks on end. Simon

I have the same board with P&C 850 PSU (I did not want to hear none of this "your PSU is too small nonsene" _ I have stable systems on 420W and unstable ones on 850.). From day one this board was not a smooth booter. I still have to be able to put 2 IDE devices on the primary ATA channel. I doubt it is the video card, especially if you verirfied it. I also doubt most major composnents. I have no idea why the board is flaky but if I had to guess I would look at the reset circuits. BTW, there are some strange interactions on this board. The stability of the CPU voltage is an inverse function of the clock speed; At 200MHz, I see visible spikes (small but visible) similar to the previous generation. These disappear at 300MHz. Have you tried this board with other-than-opteron CPU? Yes, they are supposed to work, but if everything was working like it is _supposed_ to, you will not be posting frustrations but victories :-) Simon

Did you try the pencil erasar treatment? In case this is a lost art: Start with a regular yellow pencil equipped with pinkish eraser tip - A new or very clean one. Prime the eraser by trying to erase a clean rough surface - the back of a new yellow pad, stretched clean cotton cloth. Take the card out of the computer, statically grounded etc. Erase all contact points while holding the card with contacts down. Do _not_ go vrazy, just a few strokes will do. Saturate a paper towel (I use clean, white cotton FoL, or camera lens tissues if you have those with pure alcohol. (No beer, wine, or such :-) Wipe said contacts carefully, but do not allow alcohol to run into the cicuit. Use a soft brush to brush off all lint, eraser bits, etc. Do not blow with your mouth. Such activity is to be between consenting adults, not people and computers who do not appreciate out chemical composition Make sure the slot in the MB is as clean as possible. I have dug out hairs, paper bits, parts of dead bugs, and other objects from these slots. The card goes into the slot streight: from front to back, side to side, and centered. Visualy ensure the card is sitting evenly; many MBs in cases cause the card to sit at an angle; the silly PC bracket mayneed to be re-adjusted with the proper tools, but vracked fiberglass will not improve the board. While running, wigle the card about 8-10 degrees side to side. Should not make a difference. If all this fails, visually inspect the contacts again. If it still does not work, try the card in another setting. If it does notwork yet, scream loudly. _AFTER_ you are done screaming, call the vendor for RMA.

A Guess: When MB + Expensive GPUs are inSLI mode, some memory is needed to exchange messages between cards that really do not know about each other. Enabling this option probably steals some memory, asigns it to this communication task and puts its address, size, etc. in a known place. Leaving it at Auto is a very sound advice: If the BIOS is well behaved the aperture will be setup only if needed. If the BIOS is ill-behaving, you can blame DFI for the failure instead of yourself :-) Besides, get the SLI gadgetry together, and try the various options, observing the failures. Nothing should smoke, but ... :-) Simon

My 2 cents: RMA anything that was within 10 feet of the explosion. There is nothing on any of the devices connected to the PSU to protect them from whatever the poof sent on the wires. Theoretically, there is nothing you can do to smoke a PSU. They are supposed to be OC, SC, OL, UL, OV, UV OT, UT protected. But like anything earthly, they can and do go bad. For example, I went through 4 LanParty boards in as many months. Each with a diferent failure, for a different reason. Poop happens! Simon

Add a test to the POST to pick the version number (vendor, sub, vendor, etc.) from the cards and put them in one of the BIOS screens. The 1-2 days work will be well worth it in avoided RMA's. This list should include the version and revision numbers for the MB itself; it does not show in the blue setup screens! To be extra nice, the POST can check for matching versions, etc. on GPUs and post a warning if an offensive combination is detected. To be Ultra nice, have an Auto option for the PCI-e config on the Expert; In Auto, if two matching cards are found, the BIOS will enable SLI automagically! I volunteer to do the work myself, at no cost, if DFI will provide me with the software tools needed. I have done such work before. At least once :-) - First BIOS was on an 8080A, last was on a 486...

Every Cooler Master product I owned cracked, or leaked, or cracken and then leaked. They are beautifully made but of crappy materials. And try to get any reply from the company. For a $100 or so, I have had good suxxess with the KingWin unit. It cools a CPU, VN 6600, and the NF4 rather well. You might consider replacing the tubing with 8mm ones. I could show you how to install them for a leak-proof setup. Get a non-conductive coolant. Worth the peace of mind. Frozen CPU has a non-toxic and a non-non-toxic (toxic in plain English :-) for 1/3 the price. Make sure everything is ultra clean for the non-conductive liquids, and test the liquid for conductivity before use. Main advantage of liquid cooling is: The cooling is done where you want it (outside the box, typically), and will normally be much quiter. Also, you put 4oz device on your expensive CPU, not a pound and a half, and the device leverage is one inc or less instead of 4-6 inches. To the cooling Nuts: I am designing an insulation-less active cooling for UOC. Details may follow. Simon

MotherBoard 1: (Ultra-D) Runs along mildly OC'd for some months and then dies mid-writing (I use the system to write sooftware, etc.) MotherBoard 2: (SLI-DR) runs fine for some minutes and then problem in this thread so severe had to be junked. Board 3: (Ultra-D) Without the problem but will not be stable no matter what. MotherBoard 4: Seems alive, in a new case but symptoms: Cold boots fine Shuts down fine Few seconds later yellow light next to RAM goes out If said light is on, pressing the power button will cause it to extinguish When booting 1st diag LED to exitinguish will flicker. Is this normal? Only Cure: Unplug AC, count to 10, plug AC and boot. The best (albeit partial) cure came from new memory (G.Skill 1GBx2 DDR500 which while it will not do 300MHz at all, it will do 250 very nicely. But only after I upped the vDIMM to 3.1v (my heart skipped a beat when I did that :-) My 2cents: Like others I suspect power sequencing on the MB. Observe the 24pin connector, the 4pin connector, the HD-like (you call it Molex, but Molex designed _all_ these crappy connectors way back when), and just not to be left alone, the 3.5" floppy connector. I would suspect the new motherboard (Expert?) has better power circuits. Will DFI upgrade mine? Thanx! Simon BTW, how many of you troubled souls are housing the MB in an aluminum case? A crazy theory at work...

Here are the symptoms. I need your help, please. * Pleaase forgive my typos. Due to sickness and resultant handicaap I must use a wierd (and expensive - 260 dollars each! ) keyboard, and this one is old and defective. * System OC'ed mildly as per AG recommendations. * System has been running fine for several monthts. Been very happy. * Several days ago the sudden heat phenomenon occured. I posted a detailed description but due to limited knoledge of this web software I sortt of lost it to the big bittt bucket. Cannot give you a link: but the expression "sudden heat Slow death" should be in the subject field. * Replaced PSU with a Power and Cooling 850. Had to cut tthe case out as thKoolance radiator nipples point down 1/2" too far back and radiator cannot be rotated :-( * If all the DC cables are atttached to the MB, the amber light next to the DRAM sockets stays off. The one at the bottom behaves normally. This problem waas observed witth the Hipper PSU as well, so the P&C unit is a nice solution for the wrong problem. * If I leave ony the HD-type aand Floppy-type DC connectors connected but remove the 24pins one, the next-to-raam amber LED glows but system will not staart. Not tht I expect it to - sans cables, etc. * Please reply on this thread as I cannot reaad e-mail unttil the dead box is resurected (please do not ask why - Experimenttal O/S on this box, No X (graphics) on the firewaall, etc. I really need tthis problem solved. It really is not a hobby type question this time. Thanx for a fantastic product, service,, and support! Simon

Different spacing. The Asus has 2 PC slots between the SLI socketts, the DFI has only 1. I also noted the Asus to be pasive device (no electronics, just traces), and the Leaadtek pictture shows some ICs. One wonders whaat problem is the latter correcting that the first does not feel it needs to. Simon

First, let me say I am learning sooooooo much here! Amazing! Manyy Thanx! IF this is the wrong thread/area, please forgive me and move this messaage where itt should be. I am new to this forum sttuff. Please Helpme with this myster before I starve the family replaacing good hardware. Observed: * System under normal load. * Suddenly, core temperature startsa climbing from about 42 degrees to almost 70. * It does this in steps as it normally does, but the 66-70 degrees are _not_ normal. * The climb takes about 3-5 minutes. * Video temperature climbs as well, about as much but in a more gradual manner. * Power module temperature blips up very little. * Vcore does not move * The CPU metal cap (heat spreader?) temperature tracks slowly * Water temperature follows slowly but reacts as expected; climbs from 32 tto 42 * If left to its own devices, the system will eventually crash or the CPU will halt. * Aftter power cycle, the abber LED next to the DRAM slots will not go on, * The amber LED at the bottom of the CPU does go on. * Obviously the system will not boot at all. * If I remove the (HIPPER) PSU, and attach a PSU "tester",, all lights are green which to me means the PSU is not totally dead, * DVM verifies "PSU tester"; under tester load (almost nil?) voltages seem correct. Finally, * 24 hours later the system will re-boot and run normally (but not as stable as the first 3 months), for few days. Open, perhaps irrelevant questions: * Is core temperature of 42 and CPU cover temp of 22-34 acceptable? It seems like the CPU die is not touching the cap/spreader. * Is the problem aabove indicative of a baad CPU. MB, or PSU. I am getting a PCP&C 850 so I do not have AG grumble about "PSU too small", etc :-) It is important to me to get this machine working yyesterday, or the day before, so any help will be greatly appreciated! Simon Please forgive the typos; defective KB :-(

Nah, the drive IDs itself the same way regardless who (BIOS, NF4-RAID,, XP) is aasking. If the drive IDs it is live enough to ID to all inquirers. Now,, if the drive crashes on write/read/format I would agree. BTW, there is a feaature called S.M.A.R.T, to be enabled in the BIOS and in the O/S. IT will aid in identifying flaky drives. It seems these boards were not really tested/verified much on the PCI/DISKS side. Not the perceived market, I guess. Simon

Are the drives set as RAID-1 ? Raid-1 is mirror; 2 disks appear as one with dataa duplicated (mirrored). Should one drive fail, in theory aat least :-) data will survive from other disk. RAID-1 writes slowlyy (compared to one disk), can reaad twice as fast, andwill survive one disk loss. If you set the drives as RAID-0, then the computer will ping-pong between the two every 64K or so, making the two 120's appear as one 240. Should one drive fail, you will lose all the data on both drives!!! Please note that there is a 120GGB max limit someplace in one version or another of Windoze. Cannot remember which. Also, do not ever, use RAID-0 to boot from or hold valuable data. Any one loss is a total loss. The RAID 0 and 1 on the BF4 wortks correctly. I never witnessed RAID-1 recovering though... Simon

It is not a hardware problem (my keyyboard is a hardware problem). A question Angry or another DFI employee can put to the BIOS authors: How is the memory mapped? Is it one gigantic 4GB aaddress block? Can be a problem for XP Is every slot a blocck? Is every bank a block? Best solution: Have tthe BIOS allow user to map. Had this problem on a SuperMicro server board with 8GB RAM; was maapped aas one small 0-512K and then 7GB block, which caaused the OS to barf (Pentium cannot compute on more than 4 bilion addresses even if address mode is 64bit. Ended up writing a patch that caught the (too big) block and faaked it into a series of 2GB blocks. Remember, int on Pentium is from -2billion to +2billion... :-( I may be able to help if the question above is answered. Simon PS. Forgive the typos, my good keyboaard is on the (recenttly dead) DFI based system. Reallyy!

Only 1280x1024? What is the refresh rate? Also, is the mouse somehow sharing interrupt (IRQ) with say, the video card? USB or PS2 mouse? Switch to the other type...