Just to be clear guys, when you say your system is now "stable", this should mean passing prime95 and/or memtest86 for I'd say
at least 6 hours. Just cause you were able to boot or POST with no problem, or even play games with no BSODs doesn't mean it's stable.
For those of you who are new to the thread, welcome. I think we'd be all interested in your exact setup and what kind of problems you are or are not experiencing, so please post your specs. As you can see, we've been struggling to work through the problems related to this board/chipset/processor since August and have made slow (but steady i think) progress. After being harassed by almost all of us, Gigabyte has put out several BIOS updates to try and fix the problems.
Since the latest is F6, a result of about 10 beta versions, I think we should take a step back and review the inital bugs and what has or has not been addressed.
1) dual channel memory issue - I'd say this is the most aggravating problem. It turns out that many memory modules do not run very well in dual channel unless they happen to be on the "recommended" list. There is one put out by AMD, which I would say is accurate, but it is a very small list. The one we have by Gigabyte is unofficial, and it was sent to someone via tech support email. Based on the experiences of the people in this thread, I believe we can say that Crucial Ballistix runs well. Since most review sites (including anandtech) used OCZ modules during their test of this board, I think we can give the thumbs up to this brand also. Corsair memory seems to cause the most problems, but I have heard of users having some degree of success. For some reason using DIMM slots 3 and 4 seem to be more stable than 1 and 2. And it's been established that if you want to fill all 4 slots, a command rate of 2 should be enabled. We're not really sure where the problem lies here, if the instabilities are caused by the integrated memory controller, or the nforce3 chipset, or the BIOS. I was tempted to believe AMD's explanation that only high end memory should be used for dual channel DDR400 (cheap stuff should run DDR333 just fine). But then why (I'm not exactly sure about this) are there so many complaints about instabilities with THIS board and not the ASUS A8V? That would lead one to believe the motherboard manufacturer is at fault. AMD tech support was not able to tell me for sure if this was an issue that could be improved through the BIOS.
2) incorrect temp readings - This was finally solved through the F4 BIOS versions. It took a damn long time to fix and now according to Ballistyx the problem has resurfaced in the F6 version. I'm not a BIOS programmer, but I really don't get how outputting the data from the temp sensor diode on the processor is that hard. I have called Gigabyte tech support to ask for some sort of documentation on the various BIOS revisions, and so far the BIOS dept has not provided any. I know it's not common for most companies to do this, but I dont think it is that unreasonable for me to ask. There have been MANY beta BIOS versions that tech support has asked me to try, and I would be more comfortable doing this if I had some idea as to what has been modified. For all we know, they could have just hard-coded in the temp readings instead of fixing the actual problem.
3) BIOS not accepting raises in the FSB - I'm not sure as to if this problem has been fixed in F6. According to one person in the
house of help forums, the F4K version accepted raises and the F4 did not. He speculates that this is because some aspect of the dual power system was changed in order to provide more stability. I don't know if this is true or not, but I will say that the DPS is
definitely a suspect for causing instability issues. Many people think their instabilities are memory related (and probably most are), but I think that some are because they have a less than perfect power supply. I get the feeling you need to have a good quality PSU for this board. Also, Gigabyte released an identical motherboard (K8NS Ultra-939) without the DPS. That action would lead me to believe that either something is wrong with the DPS or that the company is just leaving it out so they don't have to charge $220 for the board. Or that there is more wrong with the K8NSNXP-939 and that it's actually a revision and we are guinea pigs. I called Gigabyte sales dept about this and the guy (who didnt really seem to know dick) told me that the K8NS Ultra-939 was discontinued. So who knows.
4) hard drive LED not responding - I believe this was fixed in the F4 versions.
5) bad firewire performance - don't know, I don't use firewire. I'm just repeating what I've heard on other forums.
6) "keyboard not working after a warm reboot that all nForce boards suffer from" - Ballistyx can tell you more
7) slight overclock of AGP, Hypertransport bus, memory bus - on stock settings, HT bus is at 1005Mhz, mem is at 402, and AGP is at 67Mhz. Don't know if many others are having this problem, (and it hasn't seemed to cause instabilities for me), but I always set it back after booting. Im not sure if this is just an acceptable deviation that's within the "tolerance", or if it's just cheating to get better benchmarks. I don't think it will really cause any problems, but it's annoying.
If you guys know of any other issues, let us hear about it, and also let us know if you are experiencing any of the problems stated above.