After OC'ing my i7 920 to 4Ghz following a guide online, I've not noticed any performance improvement and every thing that reads my CPU tells me the CPU is at stock settings in terms of Ghz. When I go into the BIOS though it tells me it's 4Ghz at the top but as far as I'm aware it's not. Is it normal for monitoring programs/benchmarks to not read what the CPU is clocked at or am I running with a stock CPU?
Yep, CPU-z and make sure your multiplier and fsb are correct. Then you will need to stress test it with Prime95 for at least an hour to make sure its stable. Most monitoring programs will always give the correct reading, especially CPU-z. I have never known them to get it wrong, as its something that the system can easily find. Some things often come out wrong, for example my GPU interface reports as PCI-E x8, when it is infact PCI-E 2.0 x16.
Run Prime95 and then check CPU-Z while it's running. If your ratio is on auto, it will lower itself when it has very little to do so as to save power.
Didn't realise the 'i' series had all the auto stuff - in that case, turn the energy saving stuff off in the BIOS, and make sure your PC is set to 'Performance' in control panel (Vista and 7) - that disables cool 'n' quiet and whatever the intel equivalent of that is
People have told me that stressing the CPU with prime95 for 24hrs is the standard for testing your overclock. If your PC can keep stable over that amount of time, you're pretty much as close to stable as you need to be. You can never really tell if your overclock is 100% stable.
Stressing it for 24 hours seems like overkill really. You're just wearing it out- if the applications you run are so important that you'd want to be that sure of stability you'd never overclock anyway. Sure, it's a pain if you get a BSOD while gaming, but it's not the end of the world. 24 hours of uninterrupted 100% CPU with raised voltages and temperatures is just going to hasten the degradation of your expensive hardware.
Linpack for 4 hours with maximum amount of allocated memory. Stop unessential programs and services beforehand to free up extra ram. Mine never misses a beat, despite way too much gaming. Gaming in itself is a good stability tester or rather proof of stability.
This is one of the reasons why you should always run benchmarks BEFORE you overclock. Running a benchmarking program before the overclock means you have a base measurement to see how much you improved by. This is more important when you're tweaking your memory as sometimes when you do the memory you actually lose some speed.
Wearing it out? It's a CPU not a set of tyres. Stressing is essential in my opinion, even if you don't overclock. If you don't, you'll only end up using real life to test stability.. meaning a crash during a game, or lost data during something important. I would run Prime95 or Linpack, Intel Burn Test.. whatever, just to establish it's stable when CPU and RAM are being run flat out. 24 hours is probably not needed, but leave it running overnight at least. Running graphics benchmarks also stresses the GPU which will make any faults, or voltage/power problems apparent. You won't wear it out. You will however find any problems before they find you. Each time I build, I will always run Memtest, Prime95, and furmark or some other graphics stress test even if you don't overclock.
I use Intel burn test as it works very well and have not had any stability issues when its past that test. Simon.
