Hey guys I have a i7 4771 running in a gigabyte GA-Z87M-D3H, this is in a fractal node 605. Was originally running a stock cooler but as expected this wasn't cooling very well so swapped it for a Arctic freezer 7 I had on the side. Under full load of IBT it see's 100 degrees in seconds Now I know this isn't the most cutting edge cooler but I thought it would be reasonable on a stock speed chip? I'm limited to sub 140MM height coolers, what are my options?
If it's hitting high temps within seconds i would say the heat sink isn't making good contact, have you checked the mounting ?
Mounting seems to be fine. Soon as load goes off cpu it drops down to low 30's instantly. May check Bios isnt pumping too many volts through it.
That can't be right. See if you know someone with a corsair unit to try it out. That's running twice as hot as mine which is crazy
Then it's defiantly caused by bad contact with the CPU, if the HS was making good contact you wouldn't see the temps jump up and down in seconds, have you removed the HS to see what footprint the TIM has left, my guess is it will show some parts of the HS aren't making contact.
that doesnt sound right at all, maybe a malfunctioning temp sensor ? it shouldnt EVER get that hot ( even with stock cooling ) and shouldnt cool off that fast either could you post a screenshot of how fast it cools down ? if it is instant its almost definitely a temp sensor issue as the heat cant magically disappear from the chip
Is the mounting plate the right way round ? The mounting holes that the HS attach to should be facing front to back... Why isn't that YouTube embedding working ? http://youtu.be/GKWh24l5-tQ?t=1m50s
Yup all as it should be. Wouldn't usually make a thread about such things but I expected to see mid 60's if that.
Bad contact between the die and the heatspreader in my opinion mate, I had an I7 4770K that did exactly the same at stock re instant temp jumps to 100c under load. I ended up delidding as it was a second hander and replacing the bone dry TIM, afterwards it sat at around 68c with intel burn test and a zalman CNPS7000C-CU on silent mode. Before delidding it was so bad the same happened no matter what cooler used, anything I tried from stock up to h100i saw the instant temp jump with burn test. If it's still under guarantee just rma it, no point voiding your warranty. similiar results? https://www.youtube.com/watch?v=XXs0I5kuoX4 https://www.youtube.com/watch?v=7tPK6CdV_WI
I will see what happens with a different cooler but this has crossed my mind. It's 2nd hand also so not sure if I am brave enough to delid it!
lol well if you do don't use a razor or the vice method, simple plastic does the trick like a guitar pick etc, less chance of taking a resistor off!
Will report back. It is a retail chip and I have the hsf, i guess i maybe able to rma direct if this is the case.
In your OP you say it used to use the stock cooler and temp weren't very good, did the stock cooler exhibit the same sudden rise and fall in temps or was it just not doing a very good job ? If it's the latter then it can't be a problem with a malfunctioning temp sensor or bad contact between the die and the heatspreader, if it's the former and you had the same sudden rise and fall in temps with the stock cooler then it could be.
RMA is your only option. My very first 4770K hit 100c at stock under a Silver Arrow. Very similar to yours. Idle temperatures were fine but the minute you placed load on it, throttle time. I delidded my second one and never saw above 70c after that, even at 4.5GHz. Delidding and lapping the chip shaved off 23c.
Intel will usually replace actually faulty CPUs no-questions-asked as long as it is in the warranty period (you can check here). This only applies to processors that were bought as a box with a HSF (retail package), not ones bought on their own ('OEM' package), but it;s worth a try anyway if you don;t know which yours was sold as and the warranty checker says it's valid. I had a 2nd hand i5-3570k replaced with them never asking for any invoice information.
Swapped the stock cooler back and it instantly hits 100 across all cores. Haswell *thumb down* Will RMA this one then, see what I get back! Thanks for the input guys.
Sounds like a die->IHS interface issue. Of course, I would RMA it too, but I'm willing to bet a de-lid would sort it.
