Can I already be at the limit of my PSU?

[systemlord]

Hello everyone

When I installed my water cooled GTX 480 I also installed some EK LED lighting using two 4-pin molex connectors total, I was pushing my overclocked GTX 480 to 880MHz|1760MHz|4200MHz @ 1.125v (@ 48-51C loaded) finding it 100% stable! Here is where the BSOD started to occur, when I would go for an overclock of 905MHz|1804MHz|4200MHz @ 1.125v (@ 52-54C loaded) using OCCT and OC Scanner would crash to a BSOD with the message attached at bottom of post. Now after removing all LED lighting I was able to overclock my GTX 480 to 905MHz|1804MHz|4200MHz @ 1.125v without any issues or BSOD...?

How can just a few EK LED lights cause such problems with instability leading to BSOD? Can I already be at the limit of my Enermax Infinity 720 Watt power supply?









Uploaded with ImageShack.us

 

[rolli59]

You should not, more like 500watts and change maybe http://www.anandtech.com/show/2977/nvidia-s-geforce-gtx-480-and-gtx-470-6-months-late-was-it-worth-the-wait-/19

 

[systemlord]

You should not, more like 500watts and change maybe http://www.anandtech.com/show/2977/nvidia-s-geforce-gtx-480-and-gtx-470-6-months-late-was-it-worth-the-wait-/19

Alright 500 watts is for the stock clocked GTX 480 in a given system, I'm pushing my 480 to the moon, who knows how much more power draw I'm pulling...? I can tell you that at stock clocks my radiator pushes cool air out of my radiator, however at 905MHz core it feels like a space heater only while stress testing with FurMark, OCCT and EVGA OC Scanner! Is there any info about heavily overclocked GTX 480s regarding power draw? I would really like to find out how much wattage and amps I'm pulling from my power supply, right now I just don't have enough info whether or not I need a larger more powerful PSU!

-Systemlord

 

[rolli59]

Even with a overclock you are not adding hundreds of watts because that would melt the components. That was 479watts running Furmark which is not normal load!

 

[systemlord]

Even with a overclock you are not adding hundreds of watts because that would melt the components. That was 479watts running Furmark which is not normal load!

Then why did two LED kits (CPU & pump top) cause serious instability and BSODs? The power supply clearly seemed to have stability issues with them installed which begs the question of having enough power.

 

[13thmonkey]

pumps, lights etc. all create electrical noise, which might destabilise a system on the edge as a result of an overclock.

 

[systemlord]

pumps, lights etc. all create electrical noise, which might destabilise a system on the edge as a result of an overclock.

Question, has anyone ever come across this scenario? I personally haven't and was just wondering.

 

[13thmonkey]

I recall a similar thread from several years ago.

Try switching the pump and led's to a diferent molex connector, hopefully on a different rail?

 

[13thmonkey]

Also try it with lighting kit installed, but physically moved away from the rest of system. Running mains electric next to telephone causes bandwidth drop, think of the bandwidth required in the GPU and the incredibly precise nature of the timings and then the slightest outside influence could cause issues.

 

[systemlord]

Also try it with lighting kit installed, but physically moved away from the rest of system. Running mains electric next to telephone causes bandwidth drop, think of the bandwidth required in the GPU and the incredibly precise nature of the timings and then the slightest outside influence could cause issues.

I just finished 2 hours of OC Scanner with full power draw enabled @ 902MHz/1804MHz/4200MHz @ 1.138v without a single error after installing both LED kits wiring them away from other components and main board connections. Before when I had BSOD I had one of the dual LED wires (2 possitive + 2 negitive) wrapped around the 24-pin mobo main connector and around the RAM modules to the CPU water block, now the wires are isolated.

What I don't understand is why my load temps were 7C cooler with the same ambient temperatures, when I used OC Scanner @ 880MHz/1760MHz/4200MHz @ 1.125v my load temps were 57C, yet 52C @ 902MHz/1804MHz/4200MHz @ 1.138v! I'm starting to think my video cards BIOS is buggy, it's not just the varying temps but the amount of heat dump my radiator was doing with the smaller overclock versus the bigger overclock...?

 

[13thmonkey]

but is it now stable with leds on?

no idea about temps

 

[systemlord]

but is it now stable with leds on?

no idea about temps

It has to be either the drivers, OC programs and/or the BIOS causing these varying load temps. Perhaps I run this by Nvidia/EVGA and see if this is a known issue with 400 series cards. These BSOD happened when the card was a few hours old, perhaps now the card is broken in being place on a dyno for weeks on end! All electronics need a break-in period.

-Systemlord

 

[13thmonkey]

ahh! my 470 was ok for the first week, then sporadic crashes, then it wouldn't boot, swapped card, all ok.

 

[systemlord]

ahh! my 470 was ok for the first week, then sporadic crashes, then it wouldn't boot, swapped card, all ok.

There's still a problem at 902MHz/1804MHz/4200MHz, some days its stable while other days its not no matter how high the Vcore setting! This card just doesn't like 902MHz core and 1804MHz shader on 1.138v, 1.150v or even higher, there must be some component on the PCB that just isn't up to par, I wish I could transfer this golden GPU (P0=0.988) to another card because something other than the GPU is hindering performance!

I find it strange that my first card also would not OC past 880MHz without becoming unstable, so as before both my first GTX 480 and my second 480 become unstable after 880MHz on the core with any Vcore setting. My first card could only become stable at 880MHz core with 1.175v and my second card is stable at 880MHz core at 1.125v, I find it strange that both cards hit an overclocking brickwall at exactly the same frequency of 880MHz core! And people say that no two cards overclock exactly the same, except in my case they do!

 

[13thmonkey]

perhaps they just can't go past 880, this is early silicon, the 460's and 450's are apparently better OC's