Overclock BSOD Despite low temperature

[390382450]

I overclocked my I7 6700 K to 4.5 ghz with 1.35 volts. When running Prime 95 for 5 seconds, it went BSOD and the temperature was around 65 Celsius, I heard newer version of P95 is harmful to CPU is that true? Also, why does my computer still BSOD despite other prosper voltage and temperature?

Rig:
I7 6700K
Asus Maximus VIII
Corsair Vengeance @ 2133 32GB
No graphics card
2 SSD and 2 HDD
EVGA 650 Supernova G1 PSU

 

[JamesSneed]

Because the CPU is getting errors running at that frequency. It inst because of thermals. This is normal when overclocking. You either up the voltage or lower the overclock to overcome the errors.

 

[bgunner]

Newer Prime95 releases are not harmful tot CPU's. The issue is that anything past version 26.6 does not reflect the way Intel actually tests there CPU's. It actually stresses them more so a closer eye needs to be kept on temperatures.

If the PC is blue screening the the system is unstable. Try lowering the multiplier and retesting.

 

[greens]

Stability and Temperature have very little to do with each other in modern systems.
Thermal problems are almost unheard of with regards to overclocking.

Regarding something being harmful - no. No more harmful than driving a car hard is "harmful".

I would start over. First get your RAM up to spec. Did you enable XMP 2.0? That is going to corrrect your timings and frequency. Once that is correct, move to your CPU.

OC the CPU marginally, without touching voltage. Test for stability. Get your CPU to a point where it BSODs. NOW you increment your voltage up. Test for stability. Once you confirm you are stable again, start to OC the CPU again, marginally, without touching voltage. Once you are unstable, increment your voltage.

It can be a time consuming process, but once done you never have to worry about it again. I personally much prefer a voltage offsets over manual voltage assignments.

Upping your BCLK, Multiplier, and voltage all at once is not advised - what works on your system may not work on another's.

 

[johnbl]

when you overclock, you can violate the setup and hold time requirements for the electronics.
for example, each bit is represented by a voltage and when a bit changes from a 0 to a 1 the circuit is actually changing the voltage from a low voltage to a higher voltage. if you overclock you change the timing of the sampling of the circuit and the voltage might not have stabilized yet. this results in the bit being read when its voltage is in flux (between what would be considered a zero bit and what would be a 1 bit) when this happens the circuit will just pick a one or zero for the bit. if it is wrong then errors will crop up in the instructions given to the CPU and bad memory addresses will be used and cause drivers to crash. The CPU internal memory controller will catch some of the errors and call a bugcheck 0x124 error to shutdown the CPU.

When you actually look at the circuit and what it is doing you will be surprised that it actually works.
the faster the clock rate the more errors are introduced. The warmer the circuit the more errors you get.
but you still can have a cold circuit and have the errors. Also, when you overclock, there is a lot involved, various connections in the CPU have to have the data transfers synchronized or you get data corruption when cache memory is moved to the cpu for execution. The synchronization is done by applying different voltages via lookup table inside the bios it is based on the clock frequency. It is one reason you need to make sure the BIOS is up to date and your overclocking software is not old. It is very common to have old overclocking software that will apply the wrong voltage for the clock rate.
(happens a lot with the newer low power cpu's, make sure your bios date is newer than the release date of your cpu and generally you will be ok)

sometimes when you get a overclocking bugcheck 0x124
you can look at the memory dump and check the system up timer and dump the error record to see why the system was shut down.
most of the time it will be shutdown by request of the CPU because the CPU memory controller detected a bit error while transferring contents of memory inside the CPU cache memory.

the command in the windows debugger is
!errrec (on the address provided in the bugcheck)

generally if it is a heating problem, you can look at the system up timer and it will be longer than 15 seconds.

if the voltage to the CPU is too low or too high then the up timer will be under 15 seconds.

if you run
!sysinfo smbios
it will dump the BIOS info, you can find the date and compare it to the date that the CPU was released (check google)
often if the BIOS does not know about your CPU, it will show that the current speed of the CPU is faster than the max speed (when the system is not overclocked)

most of the time I just do not overclock, when you overclock you try to find a balance between your machine crashing do to the error rates the overclock introduces and how long you plan to run the machine before it crashes. (length of a gaming session)
but be sure to reboot often because the errors accumulate and when windows detect a error in its data structures it will shutdown the system. I would also turn off the sleep functions since you do not want errors in memory to be saved to your pagefile and get loaded after you reboot.