I have an i5-3570K, on an ASUS P8Z77-V LK motherboard, with an Antec Kuhler H620. Well, I have a solid OC of 4.6 going. During Prime95, worker 3 keeps failing. This has happened to me on every voltage with every clock speed. Is it possible that I got a CPU with a faulty core? Does that happen?
One time, with a 4.4 OC, during Prime95 the other cores ran 6 hours without error, never above 65C, and this one core failed after about 3 hours. Now, at 4.6, with 1.24V, all my cores run fine forever, at less than 68C, except this one. It keeps failing on worker #3. During Cinebench, it hasn't hurt my scores. But Prime95 kills this core every time. So, could it be a bad core? I don't want to increase V even more. I'm already pushing my comfortable heat limits.
You may want to try some other stress testing programs to get a good read on this core #3 so that if you end up calling for a RMA you will have some testing results toack you up. It would be a good idea to keep a file with all the results , because you may have to prove the issue since the cpu is actualy working and the core is failing under stress testing.
I think you'll be hardpressed to justify a claim an RMA that the part isn't working, because it only fails when you're overclocked.
as you mentioned if all cores run much better at 4.4, and they all run fine at stock, so it's not a "bad" core. If it's failing at stock then you have a case.
Nobody expects the CPU to run at infinite speed. so at some higher speed it will error out. But what will cause the error? It will be the weakest link in your chip. In this case, your weak link is more clearly core3 rather then failing somewhere random, but still it is meeting the specs of what you bought.
This being said, if the return policy is for "satisfaction guaranteed for any reason" and not return "only due to error", then yea maybe if you are not satisfied you can get a different chip... But don't fool yourself with the justification that you got a bum chip