Testing temperatures on a i7-4790k

ssmana

Reputable
Jul 5, 2014
114
0
4,680
hi i have a little concern about overheating my processor. i have a i7-4790k and it is a 4.8ghz. it is stable when i run prime 95, and i tested it also on intel burn test high setting test. my question is how else would i go about testing the temperature, i have 2 software programs called core temp and HWmonitor. im thinking of running vally with core temp open for one hour and checking the results it displays, is that what people do?
 
real temp and prime95 version 26.6.

4.8ghz is rather high. im going to guess your at 1.40v vcore at least which is too high. even with a tc14pe/nhd14/h100i coller that high of a vcore is still not safe.

http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

stick to 1.30v vcore max and whatever speeds that gets you. 4.5-4.6ghz is not much slower than 4.8ghz in the grand scheme of things and your chip will thank you.
 
what kinda temps are you getting? as long as your around or below 80c with prime 95 small fft that should be sufficient. your real world temps are almost never actually going to go as high as they do with p95. testing with a second program will just give you more unnecessary heat.

as nikoli707 stated, you should definitely keep it at 1.30v max. you should definitely give the temp guide a read too and take note of high voltages causing electromigration.

im running my i7-4790k@4.7ghz/1.25v. no issues to date.
 


nice chip rage. good vcore.
 

ssmana

Reputable
Jul 5, 2014
114
0
4,680


i used to try to do a manual overlock to get it to 4.7ghz but it kept failing. i used ez tune wizard in my bios and it set it to 4.5ghz than i would manual set the ratio to 4.6ghz and it would be stable there. i have a asus hero maximus 7 motherboard, and then i found out about a setting called cpu level up, it says that some overlocking may not have suffecient margines to reach the target frequency, no to sure what that meant but it gave 3 options, example would be 4.6 than by value of 2 4.8, so i set it on 4.8 ghz and it was stable. to make sure it was running at 4.8 ghz i used cpuid ROG and yes it does go to 4.8 ghz wheni put it under load. but im not sure of the voltages because in my BIOS it says auto.
 

ssmana

Reputable
Jul 5, 2014
114
0
4,680


i think i might just go into my BIOS and set it to 4.6ghz auto voltage.

 
i don't recommend auto voltage. it could cause voltages to go too high and give you heat issues.

you are best off setting 4.6 with manual voltage. start out at 1.28v, test and bump down to 1.27 if it passes. then 1.26, 1.25 and so forth until you get a fail. then bump it up .005 at a time after the failure. overclocking can be time consuming.


same thing for 4.7 but start out at 1.30.


you'll get the results you get. every chip gets different results. i got fairly lucky with mine.
 

ssmana

Reputable
Jul 5, 2014
114
0
4,680


it is a ez tunning wizard that sets a XMP profile with my cores at 4.5ghz, if i do a manual ovecrlock i can get 4.7 ghz at 1.295 v
 
manual overclock will always be better. ive grown accustomed to manually tuning my ram and skipping xmp as well, but xmp works fine. 1.295v is what the bios is telling you? or is that what realtemp/hwmonitor is reporting as a max vcore when you load up prime95 small fft?
 


this^. the svid on the chip will automatically up the voltages for you. once you find your max svid controller auto voltage limit at a certain frequency.... you can either stick to that speed and manually lower your voltage, or add voltage to become stable at the failed stability frequency.
 
yeah 1.377 is far too high unless you have something better than an h100i and can keep core temps around 60c or lower under load. what cooler do you have? if you say you have a xspc dual 120mm rad and your temps are at 63c under load then that is completely a different story.
 

ssmana

Reputable
Jul 5, 2014
114
0
4,680


ok i will do a manual overlock than. i figure using a h100i corsair water cooler that my chip running at 4.6ghz will be cooled enough. so i will start at 4.6 ghz with 1.28 v, and then reduce the voltage each time the test passes. thanks guys appreciate the help.
 
that is an excellent cooler and you can probably hit 1.32v peak vcore and you would be fine. but i doubt 1.32v will get you a full 0.1ghz higher than 1.30v anyways. but if you end up with 1.32v at 4.7ghz then so be it, as long as you are making good cpu contact with the block and your temps are lower than 72c under prime95 26.6 then you could run it out in my opinion.

yeah my 2500k doesn't have that problem. i lapped it to shiny mirror copper
101_5062.jpg