2010-11-07, 01:48 PM
Polantaris Wrote:I should ask my brother if I can test my GPU in his computer, that would be a better test I think, since a high load crash may occur regardless during a GPU test because of how the ones I've seen work.Do not swap parts. That is a fundamental No-No in any facility that fixes things right the first time.
If you can create a crash repeatedly, then the next step easily identifies the problem. No, that test does not say the GPU is defective. It says a shorter list of suspects exist.
Only way to have a valid voltage measurement from those six wires is to have the GPU executing stressfully - up to the point that a crash might happen.
If power to a GPU is defective, then numbers from a multimeter will say so loudly and without doubt. Without those numbers, you can be chasing your tail. Do not remove anything. Do not even disconnect one wire. Always collect relevant facts before removing anything. That also means reviewing system (event) logs and Device Manager.
Useful to identifying a defective part before replacing anything is to make the intermittent hard. If you can stress the GPU to always recreate a failure (and do so without Windows), then a hard fact exists. And the number of suspects is down to less than 10 a serious reduction.

