Originally Posted by tzvia Warning: Spoiler! (Click to show)
Craziness. I built a 'whitebox server' to run VMWare ESXI 5.5 with the evaluation of Vsphere 5.5 (studying for certification) around an FX-8350/Asus M5A99FX PRO R2.0 and 32gigs Patriot 1866. Not wanting to overspend, I put a cheap Zalman air cooler on it, thinking I was going to run it stock so I didn't need anything more. Used Liquid Pro TIM. Well, I added an old 500gig drive just to put win 7 to verify stability in Prime. Not stable on auto settings. Set manual vcore and turned off all the auto power saving stuff and had gone as high as 1.380 vcore (1.394 under load) and it would still crash after some time in prime. Lockup or reboot (usually lockup, even running ESXI). Heat was an issue, the CPU would hit 72c and the pkg would hit 53c. I wasn't going to raise vcore any higher without better cooling. I was cursing my bad cpu luck too. So I bought a CM Seidon 240m just to keep things simple; this isn't my gaming rig, it's running my domain and 4 other servers and doubling as a test bed for my classes. I removed the Zallman, and noted that the mount was good, cleaned it up and mounted the CM with Ultra instead. Let it run Prime all night at 1.38 vcore. Expected it to be frozen solid when I got up, but it was chugging away, the core at 58c, a big improvement, but I don't get that it did not crash. Was the heat before bad enough to crash it? So I ran it for 3 hours this morning at 1.340 vcore and it did not crash (before, 10 minutes max). Now running for an hour so far at 1.32vcore, occasionally hits 1.34, which would have crashed it in less than 5 minutes on the Zallman. What gives? If I get 12 hrs out of it, I will lower vcore again and see what happens. It just seems so odd. In a good sort of way but I was beginning to think that the memory was an issue or something and was about to run a memory test. I wish I had the time to read through all these posts; I am sure my answers to 1: what the hey, and 2: what a typical vcore for these chips at stock speed (I actually set the CPU to 4.2ghz manually) would be to give myself a good starting voltage to test it at.
the cpu temp is just the socket temp which is generally around 8-12 degrees higher then package temp but largely ignored. Your package temp is what you want to watch. and 53 is within safe parameters you could push up to 62 without issues. Usually recommend using hwinfo64 as it is generally more accurate with FX boards and leaves less questions of whats what (cpu temp is labeled under cpu as "CPU 0", cpu socket temp is labeled under motherboard as "CPU" in hwinfo64 aswell as northbridge, etc generally labelled correctly).
part of your issue may be LLC or maybe some other power delivery setting asus may have. My guide in signature may help a little bit atleast with LLC has a few asus digi settings at the bottom aswell though that may help a bit. Maybe it is possible that maybe the zalman was pushing too much hot air down towards the cpu socket and VRMs causing higher temps there and making it unstable? I dont know seems fairly strange to me aswell.
You should be able to push to 4.5ghz ATLEAST daily overclock without even worrying about temps with the seidon.
O yea and the stock vcore is labelled in hwinfo64 as Core #x VID. which differs from chip to chip. The general rule is the lower your VID the higher the OC you can push, Sometimes a high VID can be pushed further without touching volts though (if its a good chip). Mines 1.36v (Higher then average that I have seen, not the greatest chip either)Edited by dmfree88 - 11/10/13 at 12:30pm