Hi guys. I am having stability issues with my pc, here it is
I just put together my new system. its a A64 3000+, BFG 6800GT, 1gig Corsair valueselect ddr400, Enermax EG495P 485w psu. using 2 older WD IDE HD's. the ide HD's and the 6800GT and corsair ram were not new, the HD's were from my old pc and my bro gave me his 6800gt when he got his new card. The ram i had been using for over 6 months. Only the motherboard, cpu and psu were new.
Everything went fine during the install and was stable until I ran 3dmark05 to. This is when the problems started. Everytime i tried to run it it would restart instantly just before the first test was about to start. I have tried run in single channel and double channel mode. I have set the ram to various different timings and tried 1T/2T with no luck. I even underclocked the ram to 333. The weird thing is that sometimes it would start the test and reboot during one of the runs. So i investigated the videocard and noticed that my idle temperature on the 6800gt was 70 degrees, during one of the runs i noticed it hit 90 degrees, so i thought this might be the problem. Then i installed the asus probe and recorded the voltage rails while I was running the benchmark and noticed that on idle the psu runs at 11.904v/5.026v/3.264v. During the stress test it would dip down to 11.712/5.025/3.248. At this point I looked into the psu so I removed the PCI-x cable from the psu and hooked it up to one of the molex connectors using the pci express adapters, to try to use a different rail. I started it up and ran the test and it completed the whole test and gave me 5095 points. for this run my bios was all at auto default settings and was only running 1 512mg stick of corsair ram. So to make sure if it was all ok I ran the test again right after and then halfway thru the test the thing rebooted again. my rails were at 11.82v/5.025v/3.250v when it rebooted. So I then reconnected the pci-x cable from the psu again and ran the test again. once again it completed the test, but then it would crash halfway thru the test the second time again.
If i switch between these 2 cables I am able to run the complete test but not the second run, Can someone explain to me why?
I want to mention that before running 3dmark05, i had run prime95 for 5 hours at 99% load and it ran it great. my 12v rail was fluctuating between 11.901v and 11.84v the whole time. I even ran aquamark and got 56000 points. So i thought it was 3dmark05, so i installed counterstrike source and sure enough it rebooted 5 minutes into playing so I know I have a problem somewhere. I have tried every possible settings imagineable, even overclocking:
When i ran prime95 for 5 hours my bios settings were: 2.4ghz at 1.45v,3xHT at 266mhz, 333 divider at 210mhz@2.85v dual channel, 2.5-3-3-8, 1T. It ran everything fine until i tried 3dmark05 and crashed. I tried a stick of my Corsair ballistix pc3200 ram and the same thing happened. cpu temp never went over 41 degrees the whole time.
I have tried bios 1005, 1007 and this morning i put the 1010 beta bios in and it still randomly reboots during 3dmark05. I don;t thing the high temps on the video card are the problem because it went thru a couple of 3dmark runs with no artifacts...maybe its the powersupply? is it supposed to dip that low during stress?
If anyone has any idea, I would really appreciate it.