Here's the story, a long one.
2 weeks ago, I built a SLI system using all NEW parts. A GIGABYTE GA-K8NXP-SLI, AMD64 3500+, Mushkin 3200 LVL2 with TCCD chips, and a Rosewill 550W Rp550 in a Sunbeam Transformer case.
I started tweaking it in to see what it would do. Everything was going FAIRLY well until I set it for 4x HT, 250 fsb, x11 multi, and 1.65 vcore. It booted into windows and and started running Prime 95. About 2 minutes in, I smelled insulation burning and shut it off. I checked the board and cables, didn't see anything suspicious, so pulled the power supply. Sure enough, there was a strong smell of burnt material at the fan openings. I put an older Enermax 431W into the system and everything powered and booted. So, I immediately ordered an Enermax Noisetaker 485W from the Egg and RMA'd the Rosewill.
I got the Noisetaker 2 days later, and put it in. Booted into the system. Then started tweaking settings, again. But, this time the board would falter at post with even stock setting, randomly. After about a day (changed power supplys, again) I finally came to the conclusion that the board had fried the first time. Mainly what convinced me was that the vcore would flutuate from about 1.45 - 1.49 while monitoring it during prime AND it would not power up without the DPS card at all. When I first got the board, it would power up without the DPS. So, I figured that the MOSFET array had gone.
Of course, at this time, I was cursing the Rosewill power supply (now I believe, unduly). I RMA'd the board back to ZZF for cross shipment for a new one. Got it in two days. Installed it (with the Noistaker 485W) and everything was fine. It booted right up without the DPS (take note of this), but I installed the DPS card anyway. After about a day of tweaking, the board showed some good oc potential. but I settled for 4x HT, 240 fsb, X11 multi, 1.55 vcore, and ram at 2-3-3-8 T1. I used Clockgen to change the settings at OS load, as the board doesn't like to post with these settings, but will run Prime stable in XP.
Note that the Gigabyte board has no .5 multi's, vdimm 2.85 max, and the vcore has a jump from 1.55 to 1.65 in bios.
Now or the interesting (sad) part.
It ran fine for about 5 days. Last night, rebooted for a software install, look over and seen a "Xp has recovered from a serious error" window. Canceled out of that and and a Clockgen error popped up. I fired up CPUz and it showed the settings that I had set in bios, EXCEPT, the vcore was back to 1.408 (I had it set for 1.55 in bios). I rebooted into bios, and the vcore was set to 1.55. Rebooted back into the OS (without Clockgen) and CPUZ, Sandra, and Easytune all showed 1.4 vcore. Used Clockgen to manually set it to 1.55, but all the monitoring showed a fluctuating 1.48. Funny thing is, Easytune would set it at 1.55, and the monitoring apps would show 1.55, fluctuating, and the system would crash with any kind of stress.
I then pulled the DPS card, and guess what. It wouldn't power up. Put it back in, powered up. Another MOSFET arrray gone bad. Tried 3 different GOOD power supplies from 431w to 550W, same.
So, I have to call ZZF today and RMA another one. I haven't even gotten the refund from the last RMA, this is gonna be fun (not).
I am not going to get another Gigabyte. I am going to get the MSI. Here's why.
The Gigabytes' have only been available a short while. And, they are not the most purchased of the NF4's available. So, when you search the forums, you don't find much about them. However, there is a peppering of "board bad after six days" posts. MY OPINION is that the MOSFETS are getting hot and popping out. I think the engineers know of the heat problem. POINT: The Gigabyte does not have any cooling solution on the MOSFET array, but the MSI, Asus, and DFI SLI's
do. Maybe I am being Chicken Little here, considering I only have experience with two boards, but my gut feeling is that this will be a problem for Gigabyte.
I sure wish I had noticed this prior to ordering the first time.
One other thing to consider, tho. I have a Big Water radiator mounted on the back panel above the MOSFET array. So, in my case, there is not a lot of air flow over them. But, with the air flow in the case, the system temp monitors say it stays a couple of degress C above room temp with the case closed up, so it still should be able to cope, normally.
Wish me luck with Zip Zoom Fly.
2 weeks ago, I built a SLI system using all NEW parts. A GIGABYTE GA-K8NXP-SLI, AMD64 3500+, Mushkin 3200 LVL2 with TCCD chips, and a Rosewill 550W Rp550 in a Sunbeam Transformer case.
I started tweaking it in to see what it would do. Everything was going FAIRLY well until I set it for 4x HT, 250 fsb, x11 multi, and 1.65 vcore. It booted into windows and and started running Prime 95. About 2 minutes in, I smelled insulation burning and shut it off. I checked the board and cables, didn't see anything suspicious, so pulled the power supply. Sure enough, there was a strong smell of burnt material at the fan openings. I put an older Enermax 431W into the system and everything powered and booted. So, I immediately ordered an Enermax Noisetaker 485W from the Egg and RMA'd the Rosewill.
I got the Noisetaker 2 days later, and put it in. Booted into the system. Then started tweaking settings, again. But, this time the board would falter at post with even stock setting, randomly. After about a day (changed power supplys, again) I finally came to the conclusion that the board had fried the first time. Mainly what convinced me was that the vcore would flutuate from about 1.45 - 1.49 while monitoring it during prime AND it would not power up without the DPS card at all. When I first got the board, it would power up without the DPS. So, I figured that the MOSFET array had gone.
Of course, at this time, I was cursing the Rosewill power supply (now I believe, unduly). I RMA'd the board back to ZZF for cross shipment for a new one. Got it in two days. Installed it (with the Noistaker 485W) and everything was fine. It booted right up without the DPS (take note of this), but I installed the DPS card anyway. After about a day of tweaking, the board showed some good oc potential. but I settled for 4x HT, 240 fsb, X11 multi, 1.55 vcore, and ram at 2-3-3-8 T1. I used Clockgen to change the settings at OS load, as the board doesn't like to post with these settings, but will run Prime stable in XP.
Note that the Gigabyte board has no .5 multi's, vdimm 2.85 max, and the vcore has a jump from 1.55 to 1.65 in bios.
Now or the interesting (sad) part.
It ran fine for about 5 days. Last night, rebooted for a software install, look over and seen a "Xp has recovered from a serious error" window. Canceled out of that and and a Clockgen error popped up. I fired up CPUz and it showed the settings that I had set in bios, EXCEPT, the vcore was back to 1.408 (I had it set for 1.55 in bios). I rebooted into bios, and the vcore was set to 1.55. Rebooted back into the OS (without Clockgen) and CPUZ, Sandra, and Easytune all showed 1.4 vcore. Used Clockgen to manually set it to 1.55, but all the monitoring showed a fluctuating 1.48. Funny thing is, Easytune would set it at 1.55, and the monitoring apps would show 1.55, fluctuating, and the system would crash with any kind of stress.
I then pulled the DPS card, and guess what. It wouldn't power up. Put it back in, powered up. Another MOSFET arrray gone bad. Tried 3 different GOOD power supplies from 431w to 550W, same.
So, I have to call ZZF today and RMA another one. I haven't even gotten the refund from the last RMA, this is gonna be fun (not).
I am not going to get another Gigabyte. I am going to get the MSI. Here's why.
The Gigabytes' have only been available a short while. And, they are not the most purchased of the NF4's available. So, when you search the forums, you don't find much about them. However, there is a peppering of "board bad after six days" posts. MY OPINION is that the MOSFETS are getting hot and popping out. I think the engineers know of the heat problem. POINT: The Gigabyte does not have any cooling solution on the MOSFET array, but the MSI, Asus, and DFI SLI's
do. Maybe I am being Chicken Little here, considering I only have experience with two boards, but my gut feeling is that this will be a problem for Gigabyte.
I sure wish I had noticed this prior to ordering the first time.
One other thing to consider, tho. I have a Big Water radiator mounted on the back panel above the MOSFET array. So, in my case, there is not a lot of air flow over them. But, with the air flow in the case, the system temp monitors say it stays a couple of degress C above room temp with the case closed up, so it still should be able to cope, normally.
Wish me luck with Zip Zoom Fly.