Core i7 4770K OC Voltage?
Core i7 4770K OC Voltage?
What is the highest voltage you should use with your Corsair H105 cooler? You mentioned 1.35v as the maximum, but you've been using 1.5v and still faced stability issues at higher speeds like 4.5GHz or 4.4GHz. It seems the settings you're applying aren't working consistently, even though you remember some previous stable configurations.
Wow, you got a bad chip. The maximum voltage should be 1.35v. Don't exceed it.
My 4770k isn't working out of control... I barely manage 4GHz, even in that area, why is the boost frequency set to 4GHz when the actual operating frequency is... 4GHz?? These chips are faulty, period... many don't work well at all, contrary to what people claim.
My 4770k doesn't OC... at all! I barely reach 4GHZ, even there, why is the boost frequency 4GHz? These chips are poor, really... many don't OC well as people claim. It's true that Haswell didn't perform as well as Devils Canyon, but it's not as bad as Skylake versus Kaby. There are plenty of 4770ks that can reach 4.4/4.5ghz. In fact, it's good for you to overclock to 4ghz, Redgarl. Turbo boost only raises it to 4ghz when one core is under stress, then drops to 3.6/3.7ghz when fully loaded.
After posting this, I adjusted the voltage to 1.35 and set the multiplier to 45. When I started testing with OCCT, it crashed immediately, but after 45 minutes AIDA64 has been running smoothly. It's unclear why OCCT fails so quickly while AIDA64 handles it better.
The op doesn't suffer from a bad chip. Only a small fraction of the 4770Ks reach above 4.5ghz, and most require extremely high voltage such as 1.5v. The best I've achieved was 4.7ghz at 1.45v, though it wasn't stable. Running the chip at that voltage for 100-200mhz isn't worth it. I ended up with 4.2ghz at 1.2v.
Wow really? Haswell was that bad, huh? I can reach 4.7ghz at 1.35v on my 4690K.
OCCT has a slightly different workload compared to AIDA64. OCCT appears to be more accurate, and if any of those tests fail, it means the OC is problematic.
TechyInAZ:
Really? Haswell was that bad? I can reach 4.7ghz at 1.35v on my 4690K.
OCCT handles the workload a bit differently compared to AIDA64. It seems more accurate, and if any of those tests fail, it means the OC is problematic.
Devil's Canyon performed significantly better for overclocking thanks to extra capacitors that provide steadier power delivery and better thermal interface material.
My current setup is still running fine. I've been playing games without any issues.
Haswell are not "bad overclockers", the issue here is the temps of these when they are not delidded. They run insanely hot and make many overclock attempts a joke. (Unlike you like to see 90s C. or higher when you stress test).
Here is the funny part:
I have a "meh" 4770K which I did NOT delid (an entire story why I didn't, lol...but let's keep this for another time) and I run this PC with this chip at 4.4 at 1.250 Vcore set in Bios FOR YEARS. In general daily use and also in games (FC5 etc, far cry series had always been picky when it comes to overclocking), the PC is *STABLE*.
I just, after many years where I didn't bother with more overclocking yet did a few runs OCCT, and I get WHEA errors or crashes or freezes within minutes, at temps approaching 85. Which would indicate the system is really not stable at all.
But I am long not bothering with Prime95, OCCT etc. anymore since their "real life value" for testing CPUs, with new i5/i7 CPUs is questionable. They really don't do anything else but heating up your CPU and put them an extremely unrealistic scenario. (Let's not even mention later version of Prime with AVX instructions which add even more volts and heat up your chips like crazy)
As said, I am running this PC *for years*, so should I worry because OCCT freezes after 2 minutes? Hardly.
If you want to do better real life testing, use Asus "Real Bench", or Intel Extreme Utility Bench, with Asus Real Bench possibly the best test out there.