5930K overclocking help

[Metro]

Hello

 

So I have MSI X99 Godlike, 5930K, Corsair Dominator 32GB 3000Mhz and EK Predator 360mm

 

When I tried my 4.4Ghz overclock using 125.00Mhz BLCK and 1.3V I got crashes in games and Intel Burn Test says my overclock in unstable. So I went back to stock core clocks and kept my 125.00Mhz (as my RAM needs it) however during Intel Burn Test my system froze twice and then the third time it said unstable overclock.

 

Temperatures for both overclocked and stock are fine (71c highest stock and 85c highest overclock, however it normally is around 76c)

 

Could I have some help please.

 

Thanks

[Guest]

Don't BLCK Overclock

Use the CPU Multiplier 

[Guest]

I agree with @Ronnie76. You can't OC with BLCK on haswell, you need to overclock using the multiplier and bumping up the voltage when needed.

I just realized how many things is wrong with haswell

It doesn't play nice with Voltage, you can't BLCK OC, and it feels like Im forgetting something...

[Guest]

and you can only overclock the K-skews unlike skylake where you can overclock i3s? On top of that mainstream haswell CPUs have a horrible TIM that make the CPUs get really hot.

oh yeah and they "Fixed that with Devils Canyon"

Still I have no comment on the Skylake i3 Overclocking

You could also count the overclockable Pentium strange and off too

[harrynowl]

I agree with @Ronnie76. You can't OC with BLCK on haswell, you need to overclock using the multiplier and bumping up the voltage when needed.

I just realized how many things is wrong with haswell

It doesn't play nice with Voltage, you can't BLCK OC, and it feels like Im forgetting something...

This is LGA2011 not LGA1150

 

Running 1.25 platform (125.00 BLCK) is required for most XMP kits of DDR4. Most boards set it to 1.25 automatically.

 

@Metro have you tried the Intel XTU stress test?

[Guest]

This is LGA2011 not LGA1150

 

Running 1.25 platform (125.00 BLCK) is required for most XMP kits of DDR4

 

@Metro have you tried the Intel XTU stress test?

Haswell-E not Broadwell-E

still the same generation just different chipset, same architecture.

~js

[harrynowl]

Haswell-E not Broadwell-E

still the same generation just different chipset, same architecture.

~js

Do a bit of research..

 

Haswell-E i.e. LGA2011 requires use of a 1.25 multiplier to run high speed XMP DDR4 kits

[Guest]

Do a bit of research..

 

Haswell-E i.e. LGA2011 requires use of a 1.25 multiplier to run high speed XMP DDR4 kits

Who said the user had XMP enabled?

 

Regardless, mainstream haswell still ran really hot.

To be honest, I'm not that fond of having my CPU at STOCK SPEEDS with my h100i at 63~ deg Celsius under full load from AIDA64.

It seems way too hot, I agree with you there.  

edit: Yeah, Haswell runs too hot 1.35 volts

[harrynowl]

and you can only overclock the K-skews unlike skylake where you can overclock i3s? On top of that mainstream haswell CPUs have a horrible TIM that make the CPUs get really hot.

The Z170 spec includes a dedicated BLCK generator that's why. It's not tied to other system buses.

[harrynowl]

Who said the user had XMP enabled?

He's running 3000MHz it's a pretty safe assumption considering the JEDEC standard is 2133-C15.

 

And if you knew about the Haswell-E/X99 platform you'd know that XMP will automatically set your base to 125 if you go for a high RAM speed. Hence why OP is running 125.

[Metro]

He's running 3000MHz it's a pretty safe assumption considering the JEDEC standard is 2133-C15.

 

And if you knew about the Haswell-E/X99 platform you'd know that XMP will automatically set your base to 125 if you go for a high RAM speed. Hence why OP is running 125.

 

Hello

 

Thanks for the responses, I have not tried XTU stress test. I will do it now.

[harrynowl]

Hello

 

Thanks for the responses, I have not tried XTU stress test.

It would be worth a go. It's a bit more reliable than Linpack (IBT) and covers the instruction sets better in modern CPUs.

 

Are you sure you don't just need to up the voltage of the CPU? You do have some temp headroom.

[Metro]

It would be worth a go. It's a bit more reliable than Linpack (IBT) and covers the instruction sets better in modern CPUs.

 

Are you sure you don't just need to up the voltage of the CPU? You do have some temp headroom.

 

Do you mean at stock or when overclocked, as it my computer crashes on both.

[harrynowl]

Do you mean at stock or when overclocked, as it my computer crashes on both.

Hmm.

 

Is this only with 125 base?

 

Will leaving it at 100 and running your RAM at the standard 2133-C15 cause a crash in XTU?

 

A lot of people increase the 'system agent' voltage when they are running 125 & 3000MHz.

 

I think Tweaktown did a preppy comprehensive article on X99 & high speed DDR4 platform clocking a while back.

 

Give me a moment and I'll try dig it up.

[Metro]

Hmm.

 

Is this only with 125 base?

 

Will leaving it at 100 and running your RAM at the standard 2133-C15 cause a crash in XTU?

 

A lot of people increase the 'system agent' voltage when they are running 125 & 3000MHz.

 

I think Tweaktown did a preppy comprehensive article on X99 & high speed DDR4 platform clocking a while back.

 

Give me a moment and I'll try dig it up.

 

So I tried Intel Extreme Tuning both CPU and memory stress test for 15 minutes and it gave me no crashes or stability problems.

 

It is only with 125Mhz, clock speed is at stock clocks. Could Intel Burn test be the problem for it having stability problems without overclock.

[harrynowl]

So I tried Intel Extreme Tuning both CPU and memory stress test for 15 minutes and it gave me no crashes or stability problems.

 

It is only with 125Mhz, clock speed is at stock clocks. Could Intel Burn test be the problem for it having stability problems without overclock.

The Linpack test is pretty old. I'd just do an hour on XTU then start dialing things back up again.

[Metro]

The Linpack test is pretty old. I'd just do an hour on XTU then start dialing things back up again.

 

So I done CPU stress test on XTU for an hour, it was fine, on stock clocks with BCLK 125.00Mhz. Then I done memory stress test and got Kmode_Exception_not_handled Windows 10 error.

[harrynowl]

So I done CPU stress test on XTU for an hour, it was fine, on stock clocks with BCLK 125.00Mhz. Then I done memory stress test and got Kmode_Exception_not_handled Windows 10 error.

I'd try running memtestx86, you'll need a USB drive to run it from.

[Metro]

I'd try running memtestx86, you'll need a USB drive to run it from.

 

Yeah, I just got two Windows 10 blue screens when testing memory with XTU, so I think my RAM is bad

[Metro]

I'd try running memtestx86, you'll need a USB drive to run it from.

 

I will try memtestx86 now.

[Fetzie]

You'll likely need to increase the system agent voltage if you're running 125MHz BClk.

Haswell-E not Broadwell-E

still the same generation just different chipset, same architecture.

~js

Haswell-E and Broadwell-E are both on the X99 Chipset using the LGA2011v3 socket.

Haswell-E (i.e. i7 5820K, 5930K, 5960X) supports base-clock overclocking, which is REQUIRED to run RAM kits at clock speeds above 2666MHz.

[Metro]

I'd try running memtestx86, you'll need a USB drive to run it from.

 

It came back with no errors. So now I am not sure what is going on.

[harrynowl]

It came back with no errors. So now I am not sure what is going on.

Hmm.

 

Gotta keep it interesting.

 

I'll get back to you in the morning with any ideas.

[Metro]

Hmm.

 

Gotta keep it interesting.

 

I'll get back to you in the morning with any ideas.

 

I will do another memtest86 later and will do the UEFI version as well

[Fetzie]

I will do another memtest86 later and will do the UEFI version as well

 

Slight increases to the VCCIO (also known as VCCIN VTT on some motherboards) and the System Agent voltages should increase the stability of the IMC when running high-speed RAM. I haven't overclocked my memory higher than 2666 (I'll be honest, the RAM did that all by itself too) so I don't have any values to aim at for those two settings though.