System switches off during 3DMark-Vantage

[cruiseoveride]

I recently put together a machine to game with and I haven't really started using it yet, but I've noticed that it switches off consistently on a particular 3dmark-vantage benchmark.

Specs:
CPU: Athlon X3 440
MOBO: GA-MA790FX-UD5P
RAM: 2Gb OCZ kit
GFX: 2x HD4870s (815mhz core /950mhz mem) in crossfire
PSU: 700watt OCZ GamerXstream

All stock cooling. Only the GPUs have been overclocked. They do run a little hot 60C.

The system powers off on the same test everytime

Its the "Feature Test" that happens just after the person surround with green dots.

I think the test is some sort of blending test, i see a bunch of colours for a second and then it powers off. Its doing it on the same test everytime.

The numlock light stays on, but I need to poweroff the PSU and power it back on in order to start the machine again.

I'm wondering if its caused by the GPUs over heating, but how can I tell? Is there a program that logs system health to a file while I run the 3dmark test, so if it crashes, I can just reboot and look at the logs?

I have a strange feeling it could be the PSU as well, because I used to run the same system with a 480watt Antec PSU and it did the same switch-off trick whenever I tried to run the window experience index test.

 

[MCMunroe]

Feature Test on Vantage draws the most current and runs the GPUs the hardest.
It doesn't sound to difficult to run two tests to narrow down the source of the issue.

First run the test with both cards at stock speeds. If the test is completed successfully it was your clocks. If the computer still shuts down, it probably is the current draw.

Second try running the test at your over clock speeds but only one card by dissabling Crossfire or removing the second card.

I ran 2x HD4870s my self. I found it a hot, noisy affiar, and ended up not liking it at all and just running one of the cards. System kept setting off the over-draw alarms on my back up power supply.

 

[JackNaylorPE]

Try OCCT and let it record voltages and temps (CPU / Memory / GPU / PSU torture tests...read the graphs when done

http://www.ocbase.com/perestroika_en/

 

[cruiseoveride]

Looks like its the PSU that cant take it. Both GPUs are under 65degC and the CPU is frozen (<30C).

Thanks MCMunroe

 

[cruiseoveride]

Best answer selected by cruiseoveride.