Prime95 performance suffers after installing GT 640

[Aramis Wyler]

From that same page, it says the minimum system power requirement is 350 watts, compared to only 300 watts for the 630.

You said you had a 300 watt power supply, so I guess that'd be it right there.

[ATH]

This old review measured 138W but that is the peak during game testing:
http://www.tomshardware.com/reviews/...w,3214-10.html

But it might be the 300W PSU is not enough. Most PSU's works best at around 60-80% of maximum which is 240W max and if the card draws 60-140W.....

[Rodrigo]

Originally Posted by chalsall

If you really want to get empirical...

Cut out a bit of cardboard in the exact shape of the board in question (less the card's slot insert). (With the machine off) electrical tape it into the same position.

Then monitor the temperatures and the performance of the machine.

Some of us do this kind of thing for a living....

Before going out to buy electrical tape (cardboard isn't a problem ), I tried putting the GPU back in its slot, and leaving the case open, to see if it could be an airflow issue.

The per-iteration times with the 640 in place, Prime95 running, and the PC case open, are back to 110 and 85 ms.

Temperatures are a couple of degrees lower all around, as you might imagine with the case open.

Rodrigo

[Rodrigo]

Originally Posted by Aramis Wyler

From that same page, it says the minimum system power requirement is 350 watts, compared to only 300 watts for the 630.

You said you had a 300 watt power supply, so I guess that'd be it right there.

I have more information on this. According to my Kill-A-Watt, with the GPU in place the PC's power draw went from 55-60W before Prime95 launched, to 95W with Prime95 (and not mfaktc) running. That should be well below any threshold, no?

Hope this yields further useful clues.

Rodrigo

[Rodrigo]

OK, this is bizarre.

Ever since the last reboot to plug in the Kill-A-Watt (that is, since the previous post) it's been yielding values of 76 and 67 ms. No further changes have been made to the system.

Rodrigo

[chalsall]

Originally Posted by Rodrigo

Ever since the last reboot to plug in the Kill-A-Watt (that is, since the previous post) it's been yielding values of 76 and 67 ms. No further changes have been made to the system.

Is the cover still off? Was it offline long enough to cool down to ambient?

[Rodrigo]

At that point, and now an hour later also, the cover was/is off.

The PC was offline only for as long as it took to unplug it, put the Kill-A-Watt in place, and plug everything back in. (A minute or so.)

The difference in temperature with the case on or off seems to be 2-4 degrees.

The same good per-iteration times are still holding. In fact, these are just about the values I'd expected considering how things went with the GT 630 in the other dx7500 box.

What do you think? (Not that I'm complaining! )

Rodrigo

[swl551]

Originally Posted by Rodrigo

At that point, and now an hour later also, the cover was/is off.

The PC was offline only for as long as it took to unplug it, put the Kill-A-Watt in place, and plug everything back in. (A minute or so.)

The difference in temperature with the case on or off seems to be 2-4 degrees.

The same good per-iteration times are still holding. In fact, these are just about the values I'd expected considering how things went with the GT 630 in the other dx7500 box.

What do you think? (Not that I'm complaining! )

Rodrigo

Put the cover back on and confirm the timings slow down again. You need to be sure you can reproduce the problem in order to solve it correctly.

If the temps go back up you know the new card is impacting the system. After that it is up to you. Leave case open, add other fans, buy new system.

Assuming putting cover back results in problem returning and then you start using your GPU for something like Trial Factoring or Gaming
1. It will increase the heat problem even more
2. fry or trip the circuit of your PS as I think a 300w leaves NO headroom.

[kracker]

Originally Posted by swl551

...
2. fry or trip the circuit of your PS as I think a 300w leaves NO headroom.

I believe if it is a good, quality PSU it will be fine.

[swl551]

Originally Posted by kracker

I believe if it is a good, quality PSU it will be fine.

OEM power supplies, for your typical PC, don't exactly rank up there with best of breed. However there is no conclusion possible since we can't test its performance or its power overload features.

Life's a gamble....

[kracker]

Originally Posted by swl551

OEM power supplies, for your typical PC, don't exactly rank up there with best of breed. However there is no conclusion possible since we can't test its performance or its power overload features.

Life's a gamble....

No-name PSU's I would say 80% don't do what they say.

EDIT: I had a no-name 500w psu, my computer kept crashing... and when I swapped it for a good 460w psu... the crashes stopped.