Making my cpu more stable

[Mitrailleuse]

hey,

my cpu is currently clocked at 4.2ghz,and i wondered if i need to disable any of the 4 lowest options in here:

I Have heard that when the cpu is lowering its speed it could cause overclocks to fail(BSOD),

but i don't want to hurt my cpu's life expectancy too much.

[wng_kingsley7]

At an overclock of 4.2Ghz what kind of temperature and voltage is required?

The C3/C6 allows your CPU at idle to drop its power usage to a lower wattage no need to turn it off. Also its helps to increase the longevity of the CPU by keeping the C-states enabled vs not enabled which would reduce the high clock/frequency the CPU is running at.

[jscho]

Your CPU should be fine with the current settings. If it can last a few hours at 100% load with something like AIDA64, then you're golden. It takes serious overclocking to reduce the lifespan of the CPU.

[Mitrailleuse]

i'm running 4.2ghz with 1.275V,

but i want to go for 4.4\4.5,so i would like to do everything that can help me with stability.

[Flojer0]

In my experience messing with C states is only helpful in extreme cases, such as taking my 2600k past 4.5GHz. As others have said if you disable them for a 24/7 overclock you will shorten the life of your CPU more than necessary. Just make sure you can pass a good stress test at reasonable volts/temps and you'll be set.

 

Edit: To add to the discussion, CPU wear and tear is a product of heat and power. Power is a product of voltage and clock speed. To increase CPU longevity you want to keep all factors as low as possible. Hence if you are going to be running this overclock all the time it is recommended to leave C-states alone because keeping the clocks up when you aren't using the computer will do nothing but wear on the CPU.

[deathjester]

i'm running 4.2ghz with 1.275V,

but i want to go for 4.4\4.5,so i would like to do everything that can help me with stability.

 

 

Many Haswell's (despite the absolute BS you will read stating they all OC to 4.7, which were articles based on highly binned ENGINEERING CHIPS), will not go past 4.2 at 1.3v. 

 

You are lucky if they reach 4.5-4.6 before 1.3v to be honest. Past that? You need a very good closed loop and a delid. With a delid the h100i might be enough to get 4.4. You aren't "doing anything wrong" and the MB isn't at fault. When you go past 1.25v temps start to skyrocket, depending on the batch and some run VERY HOT. Past 1.3v you should have dual rad water. 

 

If you are worried about the life of your chip? Going past 1.3v is going to cut into that and so are temps. 

 

Now there is possibly one thing you could have "messed up". Try lowering the cache to 3500 and check stability. Not all chips will do 1:1 cache (or anywhere even close) and many need to be at 3500 to reach 4.4 

 

There are many guides out there telling you "go 100 lower on cache". This is BS. Cache means even less then a memory oc. Set cache to 3500 and find max oc first.  After that you can increase other voltages to raise cache. You also might need 1.9 on the vccin to get the OC. This will also raise temps though.

 

Haswell is not a one voltage OC. It is multiple voltages and a complete headache to get it maxed. Do one thing at a time and it will be easier.

 

When dialing in cache after the max clock? Use Aida 64 and click only cache to test it. It will find stability much faster then hours of testing. 

 

http://www.tomshardware.com/forum/id-1722630/intel-god-quick-dirty-guide-4ghz-haswell.html

 

Half the people can't even OC to 4.4 on this thread and he tells you what voltages to change. That is just Haswell. . 

[Biased Opinion]

Many Haswell's (despite the absolute BS you will read stating they all OC to 4.7, which were articles based on highly binned ENGINEERING CHIPS), will not go past 4.2 at 1.3v.

You are lucky if they reach 4.5-4.6 before 1.3v to be honest. Past that? You need a very good closed loop and a delid. With a delid the h100i might be enough to get 4.4. You aren't "doing anything wrong" and the MB isn't at fault. When you go past 1.25v temps start to skyrocket, depending on the batch and some run VERY HOT. Past 1.3v you should have dual rad water.

If you are worried about the life of your chip? Going past 1.3v is going to cut into that and so are temps.

Now there is possibly one thing you could have "messed up". Try lowering the cache to 3500 and check stability. Not all chips will do 1:1 cache (or anywhere even close) and many need to be at 3500 to reach 4.4

There are many guides out there telling you "go 100 lower on cache". This is BS. Cache means even less then a memory oc. Set cache to 3500 and find max oc first. After that you can increase other voltages to raise cache. You also might need 1.9 on the vccin to get the OC. This will also raise temps though.

Haswell is not a one voltage OC. It is multiple voltages and a complete headache to get it maxed. Do one thing at a time and it will be easier.

When dialing in cache after the max clock? Use Aida 64 and click only cache to test it. It will find stability much faster then hours of testing.

http://www.tomshardware.com/forum/id-1722630/intel-god-quick-dirty-guide-4ghz-haswell.html

Half the people can't even OC to 4.4 on this thread and he tells you what voltages to change. That is just Haswell. .

.

And here I am with a 4.6 ghz 4670k on air in the

Australian summer a only 1.225 volts. Wow I got lucky