Issues with my new system

[Geowil]

alright, so I just got my system built and finally got it to boot this weekend. Got vista installed, tried to run hal-life 2 and got an error message:

failed to lock vertex buffer in CMeshDX8::LockVertexBuffer


already contacted steam support though, so that's taken care of for now.


So I decided since hl2 would not run, lets go try cod4 and crysis to let my 4870 x2 flex a bit.

So i downloaded the crysis demo and cod4. I ran crysis first, and after starting a new game, about 8 seconds into the opening video, I got blue screened.

So i decided, fine cod4 then.

On it I got to just before it loads you up for the training course before I got another bsod.

So, I google it, see that it could be driver related. uninstalled the drivers, reinstalled em (CCC/DD 9.1).

I also read it could be caused by over heating, so I went into the advanced CCC and lowered the gpu/mem clock on the card by 50 mhz and pumped the fan up to 65%.

tried running crysis again, same thing happened.

by now I am getting more po'ed, I decide that I will just test it in 3dmark.

tried to run it and it caused my computer to restart without any bsod or error messages.



Now, I think I may know what the issue is, but just want to check with others first.

My new psu decided to commit suicide the second time I tried to use it, dont think that it got to anything else though. It's being rma'ed and I am getting a replacement on tuesday (its an 800w ultra X3).

For right now I am using a 680w Apevia.

I am thinking that my system uses maybe 400 to 500 on idle, then when I go to play a graphics intense game, it maxs it out and its bsod time.


I think this because I am able to play COD 1 without any issues.

So, if someone could take a look at my setup and see if that is the case, I would appritiate it:


OS: Windows Vista ultimate 32-bit
MB: Asus maximus II
Ram: 4 x 1 gb sticks ddrs pc8500
Gfx CArd: Sapphire Radeon HD 4870 x2
Sfx Card: Crative X-Fi extreme gamer fatal1ty edition
Proc: Intel 2.4 Ghz Q6600
1 sata HD
1 120mm fan connected at the moment
Thermaltake Big Water 760is
and a sony dvdr/cd rw combo drive

Idle Stats (in case its an overheating issue):
cpu: 20 to 25C
gpu: 42 to 46C

Full Load Stats:
cpu: 28 to 35C
gpu: 51 to 60C

 

[xthekidx]

Check the Amperage on the 12v rail on that PSU, you will need probably 35A or so for that GPU. Wattage isn't as important as the amperage on your PSU, ATI says a 650w psu is recommended for 1 4870x2, and your system definitely isn't going over that.

 

[shortstuff_mt]

BSOD's are often caused by RAM problems. Did you manually set the RAM timings and voltage to the manufacturers specs in the BIOS? Four sticks of RAM are harder on the motherboard than two sticks, especially DDR2 1066 RAM. You may even have to bump up the Northbridge and RAM voltages a notch higher to get it stable. What brand/model is your RAM? We need a little more info than "4 x 1 gb sticks ddrs pc8500". Have you run memtest86+ overnight to test for RAM errors?

 

[xthekidx]

Good point, have you done any overclocking at all on your system? Run memtest86+ and Prime95 to make sure your system is stable before you blame your psu.

 

[Geowil]

i ran memtest a few days ago, and no errors were reported. here is what I have:

http://www.ocztechnology.com/products/memory/ocz_ddr2_pc2_8500_platinum_sli_ready_edition_dual_channel

the 12 rail on my psu i am using now is under 30amps, i think its was like 28.


I have not done any overclocking yet.

@shortstuff_mt

I have not manually done anything with the ram in the bios. should I have?

EDIT:

I over stated the psu amps, rail 1 has 22 amps and rail 2 has 24 amps.

I also ran a prime95 test and after a minute, cores 0-2 stopped due to an error:

Fatal Error: Rounding was 0.5, expected less then 0.4


core 3 did not get this error.

Was monitoring with Speed Fan, the temps topped at 39c (which is the normal temp according to SF) during the test.

 

[shortstuff_mt]

You should manually set the RAM timings and voltage in the BIOS. That RAM needs a lot of voltage to work correctly. The standard for DDR2 is 1.8v and the specs on that RAM are 2.1v - 2.3v! The motherboard may not be giving the RAM enough voltage on AUTO settings. Your system isn't stable if prime95 fails. I'd start with setting the RAM timings and voltage and see what happens.

 

[xthekidx]

Enter the timings as the manufacturer specifies; 5-5-5-15, 2.1v. If its still unstable then bump up your voltage +.1v.

The reason for your blue screen is the unstablity that Prime95 found. If you can run Prime95 for a couple hours without errors then its pretty safe to assume you are stable, although I would recommend running it at least 10-12 hours to make sure.

 

[Geowil]

@shortstuff_mt

my ram was already running at 2.097v I bumped it to 2.133 and during the test only one worker thread errored this time. should I increase a little more?

 

[xthekidx]

Yes, generally speaking higher voltage gives more stability, but also higher temps. So you want to run at the lowest possible voltage that gives you stability. Since you are still getting errors though, increase your voltages until you are at the limit and if you still aren't stable, get a different kit of ram that is on the QVL of your mobo.

 

[Geowil]

ok, I changed the volts to 2.15 and ran prime for 8 minutes without errors.


quick question, should I be running one thread for each core on my proc? so 4 in this case (thats the default).

so, I deicded to test out cod4 again to see if the issue was gone.

got bsoded in roughly the same place again.


I am going to run prime for the rest of the night and then tommorow when i get home from work, i will install my replacement psu in and see if i still get bsoded.

 

[xthekidx]

Did you get BSOD when you had your other PSU before the apexia one? or did you not get that far?

 

[Geowil]

i never got close to that far lol. It popped right after my second boot up test.

running 4 threads in prime cause my cpu to over heat i think (came back down to check on it and it had restarted).

going to run it on 2 threads if that still counts for testing the memory.

 

[xthekidx]

Are you running stock speeds on your cpu? you should not be overheating under 100% loads on prime95 if you are using stock speeds, if you are then your heatsink isn't seated correctly, and needs to be fitted again.

 

[Geowil]

i am using a liquid cooler, I probably didnt get the arctic silver across the entire processor surface.

on the other hand, my 2 threads are stil running for prime, no errors, again, with only 2 threads, not sure how accurate the test is.

 

[Geowil]

ok, I replaced my psu with my new 800w. While the bsod that normally occurs is now gone a new one takes its place (before the map loads in crysis but at the same place in cod4). Posted it in the general h.b. section.

looks to be a problem with the drivers.

 

[xthekidx]

If prime95 is still giving you errors then its not drivers, its hardware.

 

[Geowil]

ok, I have solved this bsod I was getting.

It was not my ram at all, it was a setting in the CCC.

The Catalyst AI setting was causing all of my issues (the bsod and vertex lock error).

I am still going to replace the card and see if the CAI is still bugged with the new one, im my mind, that shouldn't cause a bsod or anything on a properly working card.

thanks for the help though.