6800k overclock fail

bruceGH4 · Feb 17, 2017

I'm trying to crank up my performance a bit on my new rig. My bios has a lot of tools for automatic overclocking. The ez overclock put me at 4000mhz.
The extreme tweaker ran it up to 4600 before it failed. I have been using my system with no problems at 4200 so I thought that I would crank it up and test it at 4300 with my best core(1) at 4400. This is shown in the first screenshot, and this failed sometime after 30 minutes. So I've dialed it back a bit and am running prime95 blender as shown in screenshot 2. I set my ram up to 2448.
Any comments/info on this would be appreciated.

ATMINSIDE · Feb 17, 2017

What's your cooling? That's getting warm.

bruceGH4 · Feb 17, 2017

ATMINSIDE said:
What's your cooling? That's getting warm.

I have Captain 240ex by DEEPCOOL. It came with Pre applied thermal paste.

- - - Updated - - -

Looking like this after 2 hours of prime95. Is this too hot?

M@rk · Feb 17, 2017

bruceGH4 said:
I have Captain 240ex by DEEPCOOL. It came with Pre applied thermal paste.

- - - Updated - - -

View attachment 188265

Looking like this after 2 hours of prime95. Is this too hot?

Ehh, Wouldnt run my CPU higher than 80c, 95 On hottest core seems too much, It wouldnt crash but its not so healthy for it. Have you set your voltages manually? Set it at 1.35volts and see if it crashes @4.4GHz, If it does add 0.01Volts and check again, When it finally doesnt crash, Run P95 for 30 mins, If it doesnt crash run it for 3 more hours, If it crash add 0.01volts and do all of it again. WATCH YOUR TEMPS AS YOU STRESSTEST!
If it doesnt crash, Do all of the above but instead of adding voltage, Lower it.

ATMINSIDE · Feb 17, 2017

You're approaching the thermal limit, which induces some instability. Mark's advice is solid.

idkfa · Feb 17, 2017

Doesn't prime stress all cores equally? Why is there is an over 20 °C difference in temperatures across the cores? Personally I'd redo the thermal paste.

bruceGH4 · Feb 17, 2017

idkfa said:
Doesn't prime stress all cores equally? Why is there is an over 20 °C difference in temperatures across the cores? Personally I'd redo the thermal paste.

Yeah, this is my first build but the pre-applied paste looked as old and dried as the paste on the computer I ripped apart that got replaced.
It wasn't really sealed in the box, just a loose plastic cover.
I think I should step back my overclock and benchmark, redo the paste, and see what I get.
Thanks for the idea.

wingman99 · Feb 17, 2017

Paste will only give 1-2c difference good or Bad.

bruceGH4 · Feb 17, 2017

wingman99 said:
Paste will only give 1-2c difference good or Bad.

Do you mean versus no paste at all? Or different types of paste.

ATMINSIDE · Feb 17, 2017

wingman99 said:
Paste will only give 1-2c difference good or Bad.

Dried up paste vs fresh paste is much more than a 1-2°C difference.

wingman99 · Feb 17, 2017

ATMINSIDE said:
Dried up paste vs fresh paste is much more than a 1-2°C difference.

Not for me.

Kenrou · Feb 18, 2017

VID right ? isn't 1.381v for 4.2ghz/4.3ghz a tad high ?

idkfa · Feb 18, 2017

I was thinking of the paste being unevenly spread or so. Clearly not all cores will have the same temperature but 20 °C difference seems rather high to me.

And I'm with wingman, bad or dried up paste is more than 1-2 °C.

Kenrou · Feb 18, 2017

~20c when I switched my 660ti's paste, 4yo and never been cleaned beneath the shroud but this is an extreme example

caddi daddi · Feb 18, 2017

yes, redo the paste but, just don't expect great things.
temps vary from core to core, some cpu's more than others, nothing to get excited about.
80c on the hottest core should be your peak limit, lower is always better

bruceGH4 · Feb 18, 2017

I was using auto to set my voltages. Maybe they could be lower. Haven't got to the point where I will play with that.
My current plan: Take the pizza out of the oven. Then,
no wait, right now.....
Ok
Reduced my oc to 4000 and benchmarked it as shown,
I ordered some new thermo paste (arctic silver $8) from China, should get it in 2 weeks.
Was going to use my son's leftover paste but determined it to be sh*t based on a testing site.
Will redo my cooling block mount with arctic silver and retest.
Very cool to have six cores to compare.
This is my bench after 1 hour 40 minutes blending.

trents · Feb 18, 2017

Be careful when you apply the Arctic Silver 5. It is electrically "capacitive" and you don't want to get it in the wrong places. Arctic Cooling MX4 might have been a better choice. It is safer and probably conducts heat a little better than AS 5.

ATMINSIDE · Feb 19, 2017

You'll want to use manual voltage control, you'll shave a lot of vCore off.

Kenrou · Feb 19, 2017

Kenrou said:
isn't 1.381v for 4.2ghz/4.3ghz a tad high ?

ATMINSIDE said:
You'll want to use manual voltage control, you'll shave a lot of vCore off.

bruceGH4 · Mar 4, 2017

new thermal paste

Got my $8cnd tube of arctic 5 in the mail and re-did my thermal paste.
The effect is 5 degrees c, which I think is significant.
Thanks for the advise. I guess the next step is to adjust my core voltage.
By the way, the pre-applied paste on my Deepcool rig looked fully spread so I think maybe just lower quality than Arctic 5.
Also I applied using tinting/stripe method from arctic 5 manual.

6800k overclock fail

Registered

Sim Racing Aficionado Co-Owner

Registered

Member

Sim Racing Aficionado Co-Owner

Member

Registered

Member

Registered

Sim Racing Aficionado Co-Owner

Member

Member

Member

Member

Godzilla to ant hills

Registered

Senior Member

Sim Racing Aficionado Co-Owner

Member

Registered

Similar threads