Perscitus Posted December 3, 2017 Posted December 3, 2017 I have overclocked my CPU using the following bios settings: CPU Multiplier x45BCLK: 100mhzShort and Long duration: 230/200.core current limit 200Vcore: OFFSET mode +0.020VoltsLoad line calibration: LEVEL 4Spread spectrum disabledSpeedStep: onEverything else left alone, mostly on auto. I ran a torture test with Prime95 and found that I have to stop it due to temperature issues. What can I do to reduce my top end temperatures? Is there anything that seems off, or too much voltage? Share this post Link to post Share on other sites More sharing options...
SpikeSoprano Posted December 3, 2017 Posted December 3, 2017 Some chips run hotter than others, you may need a better cooler to get to 4.5, also prime will make the cpu run a lot hotter than any game will . Share this post Link to post Share on other sites More sharing options...
ccokeman Posted December 3, 2017 Posted December 3, 2017 While the Hyper 212 is a good cooler it is still limited when you start pushing high levels of vcore through your silicon. Go from offset mode to manually set the vcore and start low and see what you get. If it reboots you have to push up the vcore a little bit more. Adjust and test until you get the right vcore/temperature/stability curve you can. As you are thermally limited then push the speeds to what your cooler can handle or buy a more substantial cooling system. Just for comparisons sake my Core i7 6950X is thermally limited even with a well built custom water cooling setup using 2 radiators and a d5 pump. Share this post Link to post Share on other sites More sharing options...
cchalogamer Posted December 3, 2017 Posted December 3, 2017 First off what temps are you seeing with your current settings? Second as mentioned before me Prime 95 is an unrealistic workload for gaming temp comparison and even over the top for things like rendering. A CPU hitting 80C in prime may run 55C max gaming. TCase max for your CPU is around 72C (that's the max temp measured at the IHS) and is different than the now more commonly listed TJunction (max temp measured at the die) meaning that you should be safe to 80C+ readings by most software that's looking at the reported core temps (For example HWMonitor) If you still need to drop temps looking at realistic loads the only option short of a better cooler is less voltage and as a byproduct to maintain stability maybe less speed. Share this post Link to post Share on other sites More sharing options...
Perscitus Posted December 3, 2017 Posted December 3, 2017 I left it overnight, and my temps from the idle are lower-mid 30s( Share this post Link to post Share on other sites More sharing options...
ccokeman Posted December 3, 2017 Posted December 3, 2017 While Prime 95 is an unrealistic workload it does impart the worst case scenario for your CPU/Memory/Motherboard combination. I used a 24 hour run as my benchmark for stability for my usage scenario. You can use other programs to simulate this loading in a shorter time window but the long term heat soak of the components mounted in the case provide a more telling story. I have had memory failures and cpu related failures at the pre and post 24 hour mark. Every small error will add up in time and you will end up with things not working or corrupted files. Test stability at the worst possible case so you know you wont have issues. We all have our own opinions on overclocking and what defines stability. I promote worst case scenario testing so you know for sure. Others are happy with good enough at light loading. I got tired of reloading the OS every couple weeks and is the reason I test the way I do. Like I said each of us has opinions so rather than take just one or two gather a larger consensus to make your decisions. Share this post Link to post Share on other sites More sharing options...
Perscitus Posted December 4, 2017 Posted December 4, 2017 I ran some games and such during the days and peaked out at 79 Share this post Link to post Share on other sites More sharing options...
cchalogamer Posted December 4, 2017 Posted December 4, 2017 I have had memory failures and cpu related failures at the pre and post 24 hour mark. Before I say anything "negative" I want to be clear, there is no right or wrong way to OC and stability test. If you're not killing parts and you're happy with the results that's the important part, but that one sentence is why I stopped bothering with "24hr stress tests" a long time ago. I still fully support you doing just that if it makes you feel better about the overclock. Personally I've found running a few benchmarks and looping them a few times testing what I just OCed finds me more issues than any 24hr run of prime/furmark in way less time. However much like running 24hrs of Prime95 can yield a "stable" system that crashes on the same test 20 mins later sometimes you'll find an issue elsewhere. A certain game will crash for seemingly no reason or you get a BSOD, first step then for me is to dial back/eliminate the OC to see if that was the cause and go from there. I've just found running something like the Heaven benchmark on loop for a few runs finds more GPU issues for me than furmark in less time and oddly enough Cinebench's multi threaded back to back and then just encoding a video or two finds issues with a similar consistency for me in far less time than longer tests. With all that out of the way I would be reasonably happy with the OC you have now. Based on current temps I would personally focus on trying to dial back to voltage if you can get it stable with less and get a bit lower temps, but if not you should still be fine where you are. Share this post Link to post Share on other sites More sharing options...
Perscitus Posted December 4, 2017 Posted December 4, 2017 Tonight I'll probably reduce the offset to +0.010V and see what temps I get from Prime95 torture.. and if they are OK, I'll give it an hour on Prime to see if any crash occurs. If there is an issue, I think I'll just start dialing back the multiplier. As mentioned, the difference it makes for my frames in game are minimal.. Shame I'm stuck at work an unable to test now...... Share this post Link to post Share on other sites More sharing options...
Perscitus Posted December 6, 2017 Posted December 6, 2017 Ended up taking it down to 4.4 and set the offset to +0.015V. Didn't have a lot of time to much yesterday since I had to install the GTX 1060. Everything is playing nice and smooth, so I might take the lazy approach and leave it as is.. but I'll continue to monitor my temps and such to see what my normal use pushes them to. Share this post Link to post Share on other sites More sharing options...
WarWeeny Posted December 6, 2017 Posted December 6, 2017 If you really are concerned about temps, just put it at 4GHz. This should be easily doable with lower voltages and you will not notice it in games. Share this post Link to post Share on other sites More sharing options...
xenkw0n Posted December 25, 2017 Posted December 25, 2017 I just want to say that your 212 Evo should be able to cool better than that. I'm kind of leaning towards bad airflow or needing to re-apply thermal paste being your biggest issue. 90c after only -7- minutes on Prime95 means you were going to get even hotter than that. 80c while gaming is extremely high for a normal 4.4-4.5ghz overclock on a 2500k. If I remember correctly these chips still had soldered heatspreaders so they should run even cooler than an IvyBridge. How much vcore actually gets applied while you're monitoring in Windows with your current offsets? Something is off, or your ambient is like 35c. Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now