Intel 5820K - First Overclock...guidance????

[JrClocker]

OK - I finally got the memory to overclock...

Do you want to hear something funny...do you?

If I use the XMP memory profile, it sets the BCLCK to 125.0 MHz. At 125.0 MHz, the system will not post.

If I set the BCLCK to 125.03 MHZ...the system will post.

Right now I have my memory clocked at 3.0 GHz at 1T (set to 1.3 V just to be safe)...doing stability testing now.







125.03 MHz...are you kidding me??????? Really????????

Is this a 5820K issue or an issue with my motherboard?

Question though...why is my CPU package power not being reported correctly? It's only reported correctly in the various apps (AIDA64, etc.) when the BCLKC is at 100 MHz.

Also...why does Windows task manager report the wrong CPU frequency at a non-100 MHz BCLCK?




Pics are below.

 

[JrClocker]

Update: As long as I use any frequency other than 125.00 MHz it works...go figure???

 

[JrClocker]

Does anybody have any advice/recommendation on how to get the correct CPU Power Dissipation and Clock frequency reported in Windows?

Is this a motherboard issue or something else?

 

[EarthDog]

CPU power dissipation?

What software in windows is reporting incorrect clocks (and w/e this CPU power dissipation is)?

 

[JrClocker]

Windows Task Manager is reporting the wrong clock (posted this and CPU-Z above).

Various software tools (Intel Extreme Tuning, AIDA64, etc.) report CPU power dissipation.

If I use a BCLK other than 100 MHz, these all report wrong numbers.

For instance, at 100 MHz, the power dissipation is reported to be about 125W for the processor.

At any other BCLK, the power dissipation is reported to be about 2.7 W.

I know the power is in the ~125 W range by the CPU temps reported.

 

[EarthDog]

Missed those above, my bad!

Not sure any of the 'power dissipation' (power use?) values are accurate anyway.

As far as why the wonky behavior with BCLK being higher, not sure why that happens... I don't recall it happening in the OC Formula I have...

 

[JrClocker]

If you could check one or two I would appreciate it. The screen shots above are from Windows 10.

The power dissipation numbers seem pretty close by doing the math related to the heat transfer equations based on the numbers from the specs on the various parts.

For instance, if I use AIDA64, Intel Extreme utility, pre 28.5 Prime95, etc. with my overclock I get a CPU power number of about 125W and max core temperatures of about 70 C. If I used Prime95 28.5 I get a CPU power number of 190 W and max core temperature of about 93 C (on just 1 core!). The math works out, so the numbers are within 5%.

 

[JrClocker]

It appears as though I have done well in the silicon lottery:

4.3 GHz at 1.200 Vcore...everything else on auto.

Power draw for OCCT, Intel extreme, etc is about 160 W max (70 C). Prime95 ver 28.5 it's 220 W (85C). This is with fans set to low noise.

So the new 280mm Corsair H110i GTX ( I.e. Non-Swiftech - hehe) is doing the job.

Now, if I let the AIO soak, the temps rise about 5 to 7 C. Still OK...I am happy with the Overclock

Now, the major reason for the heat soak is the CPU. However:

1. The air across the radiator is from inside the case
2. The air flow comes across the board VRMs
3. I have 2 970s that dump 200 to 250 W into the case

The case has good ventilation.

Should I reverse the airflow across the radiator to blow from outside the case? My concern is that I would be blowing hot air onto the VRM heat sinks...

Thanks!

 

[JrClocker]

@ mimart7

About the power supply from earlier in this thread...I think you get to say a qualified "I told you so".

With my old power supply, any sort of overclock on my 970s in SLI was not stable. I read some 970 forums and this seems to be an on and off issue...with no real answer other than the typical "it's an NVIDIA conspiracy", "SLI sucks", etc.

I tried playing around with SLI overclocks last night on my new EVGA power supply. I was able to get it "stable" over a few hours. I set the 970 clocks back to stock after messing around for a bit. Don't really need an overclock there...yet.

However, it seemed to be more stable. However, the other delta in my system is an NVIDIA driver upgrade. The release notes did not say anything about SLI overclock stability changes...but I have written more than enough release notes over my career and know that it's impossible to include everything.

So, it could be the power supply or it could be the drivers. I don't have the patience to install the old power supply to rule out that variable...so it's a qualified "you told me so".

 

[mimart7]

@ mimart7

About the power supply from earlier in this thread...I think you get to say a qualified "I told you so".

With my old power supply, any sort of overclock on my 970s in SLI was not stable. I read some 970 forums and this seems to be an on and off issue...with no real answer other than the typical "it's an NVIDIA conspiracy", "SLI sucks", etc.

I tried playing around with SLI overclocks last night on my new EVGA power supply. I was able to get it "stable" over a few hours. I set the 970 clocks back to stock after messing around for a bit. Don't really need an overclock there...yet.

However, it seemed to be more stable. However, the other delta in my system is an NVIDIA driver upgrade. The release notes did not say anything about SLI overclock stability changes...but I have written more than enough release notes over my career and know that it's impossible to include everything.

So, it could be the power supply or it could be the drivers. I don't have the patience to install the old power supply to rule out that variable...so it's a qualified "you told me so".

I would never say say that to anyone. Glad you are getting this sorted out.