Hi guys,
Attempted to overclock my 4670k 2 years after I bought it and first tried and so far have had some reasonable results.
Overclocking it on a z87 Asus Hero motherboard, I have managed to clock it to 4.4Ghz at 1.2 volts, with my XMP profile enabled (2133Mhz, 8GB).
At these settings I have done one 8 hour RealBench stress test, with temps hitting a peak average of 71 degrees. I have also passed the high setting on IBT with temps in the high 80s.
There are a couple of (probably beginner) questions I was hoping you could help me with.
1)
I've asked about this on these forums before I think, but when stress testing (and messing around with my overclock generally) I opt for a manual cpu voltage in the BIOS and specifically set it to 1.200v. Even manually entering this value for some reason causes the CPU to pull 1.216v at load (whether that be stress testing or games). I know that the adaptive settings can cause this, but I was under the assumption that the manual setting is a maximum it should never go over? Is this simply a bug in the way the voltage is being reported in CPU-ID & HWMonitor?
2)
At the moment my cooling solution (h80i + 2x NF-F12s) seems to be underperforming. I have had problems with the mounting of it in the past, but seem to have fixed all the obvious issues (backplate being mounted incorrectly, thermal paste being applied incorrectly etc). The rad hasn't been thoroughly cleaned in over a year though, so at the weekend I'm going to remove it, clean it, and reseat it, in the hopes of ironing out some of my issues.
That being said, I want to ensure that I thoroughly test my system stability with my overclock enabled. At the moment however, I'm pretty sure Prime 95 would cause my temps to approach 90 degrees after a few mins and potentially approach 100 during an 8 hour stress test, on a small FFT test.
I use my computer 99% of the time for gaming + general web browsing and so will never come even close to the temps output by Prime 95 in general use.
If I configured my system so that an 8 hour Prime 95 run didn't go over, say, 80 degrees, I don't think I'd be able to move above stock at all.
Basically, what I am asking is using Prime 95 and other synthetic benchmarks really necessary when testing an overclock? As long as something like RealBench runs with decent temps and gaming runs with decent temps I should be okay to use the clock 24/7?
3)
Finally, what sort of peak temps are safe (by this I mean, won't cause excessive cpu wear) to use 24/7 on Haswell? Most overclock guides seem to tune their targets based on achieving no more than 80 degrees on synthetic benchmarks, but that will probably mean low 60s during gaming, so is that the limit I should be aiming for?
On my current clock I got an average maximum temperature of 70 degrees after 4 hours of playing Borderlands 2, but after I've cleaned / reseated my h80i I assume that will drop to below 70.
In truth, because each person seems to have different values they aim for in terms of stress test and normal load (gaming), it's difficult for a beginner like myself to truly understand what sort of temps I should be aiming for, and what is safe etc.
Hope this hasn't come across as too much of a ramble!
Attempted to overclock my 4670k 2 years after I bought it and first tried and so far have had some reasonable results.
Overclocking it on a z87 Asus Hero motherboard, I have managed to clock it to 4.4Ghz at 1.2 volts, with my XMP profile enabled (2133Mhz, 8GB).
At these settings I have done one 8 hour RealBench stress test, with temps hitting a peak average of 71 degrees. I have also passed the high setting on IBT with temps in the high 80s.
There are a couple of (probably beginner) questions I was hoping you could help me with.
1)
I've asked about this on these forums before I think, but when stress testing (and messing around with my overclock generally) I opt for a manual cpu voltage in the BIOS and specifically set it to 1.200v. Even manually entering this value for some reason causes the CPU to pull 1.216v at load (whether that be stress testing or games). I know that the adaptive settings can cause this, but I was under the assumption that the manual setting is a maximum it should never go over? Is this simply a bug in the way the voltage is being reported in CPU-ID & HWMonitor?
2)
At the moment my cooling solution (h80i + 2x NF-F12s) seems to be underperforming. I have had problems with the mounting of it in the past, but seem to have fixed all the obvious issues (backplate being mounted incorrectly, thermal paste being applied incorrectly etc). The rad hasn't been thoroughly cleaned in over a year though, so at the weekend I'm going to remove it, clean it, and reseat it, in the hopes of ironing out some of my issues.
That being said, I want to ensure that I thoroughly test my system stability with my overclock enabled. At the moment however, I'm pretty sure Prime 95 would cause my temps to approach 90 degrees after a few mins and potentially approach 100 during an 8 hour stress test, on a small FFT test.
I use my computer 99% of the time for gaming + general web browsing and so will never come even close to the temps output by Prime 95 in general use.
If I configured my system so that an 8 hour Prime 95 run didn't go over, say, 80 degrees, I don't think I'd be able to move above stock at all.
Basically, what I am asking is using Prime 95 and other synthetic benchmarks really necessary when testing an overclock? As long as something like RealBench runs with decent temps and gaming runs with decent temps I should be okay to use the clock 24/7?
3)
Finally, what sort of peak temps are safe (by this I mean, won't cause excessive cpu wear) to use 24/7 on Haswell? Most overclock guides seem to tune their targets based on achieving no more than 80 degrees on synthetic benchmarks, but that will probably mean low 60s during gaming, so is that the limit I should be aiming for?
On my current clock I got an average maximum temperature of 70 degrees after 4 hours of playing Borderlands 2, but after I've cleaned / reseated my h80i I assume that will drop to below 70.
In truth, because each person seems to have different values they aim for in terms of stress test and normal load (gaming), it's difficult for a beginner like myself to truly understand what sort of temps I should be aiming for, and what is safe etc.
Hope this hasn't come across as too much of a ramble!
Last edited: