Prime95 performance suffers after installing GT 640 - Page 4

Aramis Wyler · 2013-03-23, 01:59

Maybe the card was just seated poorly.

Rodrigo · 2013-03-23, 04:00

Quote:

Originally Posted by swl551

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.

All right, new weird symptom.

As swl551 suggested, I put the case cover back on. This time, I could not get the power switch (the physical button at the top of the case) to light up after pressing it to turn the PC on. However, the indicator light on the back of the case did light up, as did the Ethernet port with the cable hooked up. Both fans (in and out) also started whirring, normally. But in addition to the power switch not lighting up, there was no video signal to the monitor.

Curiously, pressing the power switch did not turn off the computer. I had to keep it pressed for 6 seconds (as if the PC had in fact been turned on and I was doing a hard reset). And yet, the XP startup jingle never played either.

Tried this 4 times -- no dice. Reopened the case, took out the GPU, and rebooted (with the case open). Now the power switch did light up and XP booted normally, with a video signal. Prime95 even opened automatically as per my settings to launch at startup.

So I shut down the system, unplugged all the cables, put the GPU back in, buttoned everything back up -- and once again the power switch isn't lighting up and there's no video signal.

For good measure, I unplugged the power cord, removed the case cover, plugged the power cord back in and hit the power button. Same thing -- power button not lighting up, no video.

What could the power switch lighting or not lighting up have to do with the rest of it? This may be an important clue. But it wasn't happening before tonight!

Over to you.

Rodrigo

P.S. FWIW, the PSU info is: Bestec ATX0300D5WC, H-P Part No. 5188-2625. (This computer was built in 2009.)

swl551 · 2013-03-23, 04:20

Quote:

Originally Posted by Rodrigo

All right, new weird symptom.

As swl551 suggested, I put the case cover back on. This time, I could not get the power switch (the physical button at the top of the case) to light up after pressing it to turn the PC on. However, the indicator light on the back of the case did light up, as did the Ethernet port with the cable hooked up. Both fans (in and out) also started whirring, normally. But in addition to the power switch not lighting up, there was no video signal to the monitor.

Curiously, pressing the power switch did not turn off the computer. I had to keep it pressed for 6 seconds (as if the PC had in fact been turned on and I was doing a hard reset). And yet, the XP startup jingle never played either.

Tried this 4 times -- no dice. Reopened the case, took out the GPU, and rebooted (with the case open). Now the power switch did light up and XP booted normally, with a video signal. Prime95 even opened automatically as per my settings to launch at startup.

So I shut down the system, unplugged all the cables, put the GPU back in, buttoned everything back up -- and once again the power switch isn't lighting up and there's no video signal.

For good measure, I unplugged the power cord, removed the case cover, plugged the power cord back in and hit the power button. Same thing -- power button not lighting up, no video.

What could the power switch lighting or not lighting up have to do with the rest of it? This may be an important clue. But it wasn't happening before tonight!

Over to you.

Rodrigo

P.S. FWIW, the PSU info is: Bestec ATX0300D5WC, H-P Part No. 5188-2625. (This computer was built in 2009.)

I had a card die on me one day after I powered off the computer. Just 30 seconds later when I turned it back on, nothing.... The machine refused to start. By experimentation I discovered removing the GPU allowed it to start up and boot. The card was warrantied and all ended well.

To prove it was the GPU I put it another PC and bingo the second PC could not startup either.... Was all trial and error until I found a pattern.

If you have another PC try the GPU swap around. see what's what.

Sorry to hear about it. I know how frustrating it is.

Aramis Wyler · 2013-03-23, 04:29

I guess a card with a short ciircuit would explain a lot of the power issues. So would a motherboard short maybe. Watch out for any metal touching (or underneath) the board.

Dubslow · 2013-03-23, 05:55

Hmm.... if it was a GPU farting, why would that affect the power switch light? That smells of a bad motherboard. Either way though, swl551 has excellent advice -- test the GPU in another box and see what happens. If it is in fact the GPU, then contact the manufacturer, but be cautious if they ask about the power supply; that's quite suspect (as in it seems a bit underpowered, but who knows).

Rodrigo · 2013-03-23, 22:16

Update: I had another GPU waiting to be installed on a different machine. It's a GT 630 by Galaxy.

Popped it into place this afternoon, on the problem XP machine. The PC came up normally and I installed the latest drivers for it.

Per-iteration times are back to where they were when I'd installed the 640 -- at 111 and 86 ms.

But at least it's all working.

Of course, this is just how things started out with the 640: it, too, worked at first and with the higher iteration times.

Will keep you posted. I have yet to try setting up the 640 on a different machine.

One other thing. I noticed that the specs on the box for the 640 say, "minumum 12V current rating of 20A." The sticker on the PSU says, "output +12V / 19A -12V / 0.8A". Not being an electrical engineer, I'm not sure what much difference this makes, or how much. But I'm confident that somebody here will know.

Rodrigo

Rodrigo · 2013-03-24, 05:52

Update 2: Been running mfaktc on an OBD exponent as a test of the new 630. LL per-iteration times in Prime95 are holding steady at 110 and 86 ms. So at least I know that LL productivity won't be any worse than this if I run mfaktc at the same time (the mfaktc self-test did significantly affect Prime95 tonight).

Rodrigo

P.S. One other thing -- the 630 in the "good" dx7500 microtower has the same power requirements (350w, 20A) as the 630 that I put in the "bad" dx7500.

chalsall · 2013-03-24, 14:33

Quote:

Originally Posted by Rodrigo

One other thing. I noticed that the specs on the box for the 640 say, "minumum 12V current rating of 20A." The sticker on the PSU says, "output +12V / 19A -12V / 0.8A". Not being an electrical engineer, I'm not sure what much difference this makes, or how much.

That's not good...

The card's minimum is 20A. The PSU's maximum is 19A. (The -12V amps rating is completely independent from the +12V.)

And remember that your GPU(s) are not the only thing(s) consuming +12V. You definitely need to upgrade your PSU.

Rodrigo · 2013-03-24, 16:44

Thank you, chalsall. I'll search for a shopping guide to help in the selection of a suitable power supply.

Rodrigo

chalsall · 2013-03-24, 16:55

Quote:

Originally Posted by Rodrigo

What are the dangers of leaving the situation as it is?

In order:

1. Overall system instability.
2. Frying your power supply. PSUs should never be run at 100% (or more) of their rated capacity.
3. Frying other components during #2. (Not likely, but possible.)

A good way of thinking about electrical power is to model it like water in a dam. The voltage is the level of the water in the dam. The amperage is the size of the pipe out of the bottom of the dam.

This analogy is useful because it can visually show that while it's possible to "over voltage" through a pipe, it isn't possible to draw more amps through the pipe than it's designed for. But, importantly, it is possible to draw less amps than the pipe can supply.

Or, put another way, voltage is pushed, amps are drawn.

Edit: Whoops. I answered a question which you removed. I hope the above is still useful.

Rodrigo · 2013-03-24, 17:14

Yes, it was very useful, thank you!

I'd removed the question because I figured that only bad things could possibly happen. But your answer tells me why, and better yet it provides that mental picture that's so helpful. I'd never heard the concepts explained that way, and it really makes all the difference.

Thanks again!

Rodrigo

2013-03-23, 22:16	#39
Rodrigo Jun 2010 Pennsylvania 947 Posts	Update: I had another GPU waiting to be installed on a different machine. It's a GT 630 by Galaxy. Popped it into place this afternoon, on the problem XP machine. The PC came up normally and I installed the latest drivers for it. Per-iteration times are back to where they were when I'd installed the 640 -- at 111 and 86 ms. But at least it's all working. Of course, this is just how things started out with the 640: it, too, worked at first and with the higher iteration times. Will keep you posted. I have yet to try setting up the 640 on a different machine. One other thing. I noticed that the specs on the box for the 640 say, "minumum 12V current rating of 20A." The sticker on the PSU says, "output +12V / 19A -12V / 0.8A". Not being an electrical engineer, I'm not sure what much difference this makes, or how much. But I'm confident that somebody here will know. Rodrigo Last fiddled with by Rodrigo on 2013-03-23 at 22:17 Reason: additional info

2013-03-24, 05:52	#40
Rodrigo Jun 2010 Pennsylvania 947 Posts	Update 2: Been running mfaktc on an OBD exponent as a test of the new 630. LL per-iteration times in Prime95 are holding steady at 110 and 86 ms. So at least I know that LL productivity won't be any worse than this if I run mfaktc at the same time (the mfaktc self-test did significantly affect Prime95 tonight). Rodrigo P.S. One other thing -- the 630 in the "good" dx7500 microtower has the same power requirements (350w, 20A) as the 630 that I put in the "bad" dx7500. Last fiddled with by Rodrigo on 2013-03-24 at 05:59 Reason: none at all

2013-03-24, 16:44	#42
Rodrigo Jun 2010 Pennsylvania 947 Posts	Thank you, chalsall. I'll search for a shopping guide to help in the selection of a suitable power supply. Rodrigo Last fiddled with by Rodrigo on 2013-03-24 at 16:46

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Problems installing Prime95 on Mac OS X 10.4	peterhowarth	Software	4	2015-02-18 10:40
Prime95 performance on Core i5 3570K	akselsm	Hardware	14	2013-02-10 15:23
Dropbox dropped Prime95 performance 47%	Rodrigo	Information & Answers	26	2011-06-21 05:06
Performance of Prime95 on Windows XP x64	Schamschi	Software	5	2009-09-20 20:45
Processor Performance vs. Price For Prime95	louis_net	Hardware	5	2004-07-15 05:11

2013-03-23, 01:59	#34
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 2³×53 Posts	Maybe the card was just seated poorly.

2013-03-23, 04:29	#37
Aramis Wyler "Bill Staffen" Jan 2013 Pittsburgh, PA, USA 2³·53 Posts	I guess a card with a short ciircuit would explain a lot of the power issues. So would a motherboard short maybe. Watch out for any metal touching (or underneath) the board.

2013-03-23, 05:55	#38
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3·29·83 Posts	Hmm.... if it was a GPU farting, why would that affect the power switch light? That smells of a bad motherboard. Either way though, swl551 has excellent advice -- test the GPU in another box and see what happens. If it is in fact the GPU, then contact the manufacturer, but be cautious if they ask about the power supply; that's quite suspect (as in it seems a bit underpowered, but who knows).

2013-03-24, 17:14	#44
Rodrigo Jun 2010 Pennsylvania 1663₈ Posts	Yes, it was very useful, thank you! I'd removed the question because I figured that only bad things could possibly happen. But your answer tells me why, and better yet it provides that mental picture that's so helpful. I'd never heard the concepts explained that way, and it really makes all the difference. Thanks again! Rodrigo