Overclocking WTF?!?

[Sandman[SA]]

This is more of a rant really.  But feel free to ask questions, comment or add overclocking stories of your own if you like.

So I have been toying around with my old system just trying to see how far it can overclock it.  So I had been using D.O.C.P. profile as it gives higher memory overclocking presets.  I was able to push that Intel Core i7 930 2.80 ghz processor to 4.21 ghz and it would be somewhat stable for the most part.  Grant it, I had to push memory voltage to 1.90v.  1.50v is default.  There is 24 gb of DDR3 1600 MHz ram in triple channel.  Even with water cooling using a 120mm rad and a 240mm rad in series, temps were hitting upper 80C to mid 90C under load and hover around the upper 50C at idle.  Also overclocked the R9 390X to 1125 MHz (default clock is 1060 MHz)  Temps on that was also upper 80C under full load.  Had to push core voltage +10 and power limit to +20 to keep it from artifacting.

So running Heaven Benchmark with it set to the ultra preset, the opening scene couldn't get above 64 fps, and had some random micro studders here and there.  The amount of heat those radiators were putting off while load testing was shockingly hot!!  Like might melt plastic level of heat.  The computer crashed during the 3rd run and wouldn't boot back up until I cleared the CMOS.  It took a few hours but I was finally able to regain that clock speed again.  But it was less stable than before.

So I went back to the BIOS and switched over to X.M.P profile.  Set CPU ratio to 22.  Disabled speed stepping and turbo mode.  Set BCLK to 165 and set memory voltage to 1.64.  Basically left everything else to auto or default settings and got just over 3.6 ghz.  The CPU now never gets over 77C and the heat coming off of the radiators is much, MUCH cooler.  Re-ran heaven benchmark with the GPU set to default settings, and that same opening scene now shows high 70's fps.  So am I missing something here?  How is it that this system runs faster at slower clock speeds?

[One Eyed Wonderweasle]

 It could have been the  one of the chipsets over heating on the mother boards I always ended up putting extra cooling on the northbridge cooler of the SKT 1366 they tended to overheat on non overclocked systems.

[Sandman[SA]]

Sorry about the crappy pictures.  But you can sort of see what I am doing with this system.

I re-did all the thermal paste with thermal grizzly kryonaut and replaced the thermal pad under the mosfet cooler.  There is a 120mm fan aimed at the chipset coolers and ram modules.  And yeah, they still get pretty warm.  It also helps cool the gpu backplate which gets crazy hot, even with the 2 front case fans going.

[Sandman[SA]]

Okay.  So as it turns out, the 2 AIO's that I modded into a open loop cooling system had a small flaw.  I have a very old Corsair H100 240mm cooler for the CPU and a much newer Corsair H55 120mm cooler for the GPU, hooked together with a EK 150 mil reservoir.  It's mostly just distilled water and a small part of Nissan blue automotive antifreeze as the coolant.

Anyway, the H100 has a 3 speed pump control on the face of the pump housing with 4 additional 3 pin PWM headers on the top of the housing for fans.  The pump controller had died a long time ago.  The fan headers on it also no longer worked, but the pump still ran.  Although, what I wasn't aware of was that the pump was stuck on it's lowest speed setting.  Well this explains why the 240mm rad (first rad in the loop) was heating up much more and far sooner then the 120mm rad.  And took so long to cool back down.

So I took the H100 pump off, took it apart to try and find a way to by-pass the pump speed controller so it can just run at full speed.  What I found is that the pump motor has a 3 wire PWM connector that plugs directly into the pump controller board.  I unplugged the pump motor from the controller and plugged it in to molex to PWM adapter.  Now the pump is running at full speed and both rads are heating up and cooling down much faster.  Also drove the temps down nearly 15C under load.

But now, I fear that I've run in to another issue.  The very old 750 watt PSU is getting very hot when the system is a full load for extended runs.   After a while, the computer just shuts off, then tries to reboot several times all on it's own.  But can't make it through post until the PSU cools down some. 




Edit:  Found a problem with the PSU overheating.  The air filter on the bottom of the case (Fractal Design Focus G) has a plastic film covering most of it.   

[One Eyed Wonderweasle]

 Glad to see that it looks you fixed most (or all?) of the big problems.

[Sandman[SA]]

So far, so good.  It's been under a full stress load for the past 3+ hours with no issues what so ever, running at 4ghz.  CPU temps leveled off in the high 80's C and the GPU is staying in the low 60's C.  And the PSU is staying nice and cool. 

Who the hell puts plastic film over a PSU intake filter anyway??  Fractal Design. 

[One Eyed Wonderweasle]

 Maybe it's for people that have their power supplies intake facing up? don't know, I don't think I've seen that before.

[Sandman[SA]]

Eh.  Maybe?  They do make RGB PSU's that have RGB fans too.  So turning it up side down would give you more bling in a case like the Focus G. 

But yeah, I've never seen that before either.  It could have just stuck there during packaging because it didn't look like it was intended to be there.  I just never noticed it until I pulled off the filter mesh from the bottom of the case. 

[Sandman[SA]]

I figured that since this is a thread about overclocking, I'd just put this right here.  Good educational stuff for those who may want to try overclocking on a modern UEFI bios.