Solved

Safe 24/7 Overclock and Reasonable Temperature i7 4790K

i7 4790K with Phanteks PH-TC14PE.

Voltage manually set to 1.2 Off-set stuff seems more complicated than I'm really wanting to get into.

I'm overclocked to 4.7 Ghz, stable after 1 hour of Prime95.
Max temperatures:
78 #0
80 #1
80 #2
72 #3

Power consumption was 122.29 (38% over TDP of 88)

So my questions are:
Is this a safe OC for 24/7 usage? I really don't feel like getting involved with the voltage offset stuff, but if this voltage 24/7 will put the longevity of my processor at risk, then I will.

Is my PH-TC14PE working properly, and are my temperature OK for Prime? (Obviously in regular use I won't ever go above 75-78 degrees)

Thanks for any replies!
34 answers Last reply Best Answer
More about safe overclock reasonable temperature 4790k
  1. If you can do 4.7 with a vcore of 1.2, congratulations, you won the bin lottery.

    The consensus is that vcore voltages higher than 1.30 are not a good thing for 24/7 usage.
    I have been unable to find any official Intel recommendation on what is a safe vcore limit.

    I would take the time to set up speedstep(C1E). That lowers the multiplier and voltage when the cpu has little to do.
    It is a very confusing area, and the settings are named differently in different motherboard bios'es.

    I have abandoned prime95 as a stress tester. Plus I can't seem to get rid of the program, and that bothers me.

    I use OCCT which is probably more representative of what we normally do.
    It will shut down the test at 85c.
  2. MisterSprinkles said:
    This is very odd to me. You shouldn't be hitting 122W @ 1.2V in my experience. Something is not doing what you want it to here. Your temps seem high for 1.2V unless you just have horrible airflow in your case or something. That's a pretty decent air cooler you have on it.

    Geofelt when you tell people to turn speedstep and C states back on you need to include a disclaimer that this can cause instability with an overclocked system.


    So I was typing a response to you, in which I was going to take a screenshot during Prime95 testing and I crashed after 2 minutes.
    Odd because the same settings were stable for an hour long test...
    Regardless, I increased my Vcore to 1.225.

    Air flow in my case should be good:
    (Imagine a third fan to the left there)


    And here is a screenshot of HWMonitor, RealTemp, and CPU-Z just after a fresh run of Prime95:


    I'm brand new to this, so I don't know the in's and out's, however I do notice a sizeable discrepancy between what HWMonitor reports as my vcore, and what CPU-Z does (1.248 vs 1.223)

    Any ideas what, if any, the issue would be to cause such a high power usage?
    You can see that, with the voltage increase I am now up to 83 on my highest core, and a power usage of 132W (51% over TDP)
    The fact that my power usage increased 10W from a voltage increase of .025 strikes me as a bit odd, but again I am new to this.


    I can also take pictures of my BIOS settings if that helps, the ONLY things I have altered are:
    Disabled turbo boost so I can start OC'ing.
    Set vcore to manual from auto
    enable XMP 1 for my RAM

    Thanks for any/all help!
  3. jordan1794,

    Just so we don't overlook the basics, which version of Prime95?

    Blend or Small FFT's?

    What is your ambient temperature?

    Are you running the latest BIOS?

    Have you read this? Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html
  4. CompuTronix said:
    jordan1794,

    Just so we don't overlook the basics, which version of Prime95?

    Blend or Small FFT's?

    What is your ambient temperature?

    Are you running the latest BIOS?

    Have you read this? Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html


    Yes, I am very familiar with that guide, have probably read it 15 times now (Kinda paranoid about temps lol!)
    I am using the prime version listed there, with small FFT's

    Ambient temperature is around 22-24

    Latest BIOS is being used.
    However I'm not too happy with how my BIOS was handling my voltages when set to auto : http://www.tomshardware.com/forum/id-2579489/investigate-auto-voltage-settings-gigabyte-bios.html
  5. Thank you for that information.

    122 watts does seem high for a Vcore of 1.248. Although there could be several reasons for this, it appears to scale corresponding to Vcore.

    My i7 4770K is O/C'd to 4.7 at 1.280 volts (adaptive) with 114 watts at 76C (22C ambient) on air with Cooler Master TPC 812 (Dual 77 CFM fans) bareback on the processor die.

    Keep in mind that ambient plays a role as a minor variable. As ambient increases, resistance increases, which increases wattage.

    Try Core Temp to compare wattage with HWMonitor. Let's see what you get.
  6. CompuTronix said:
    Thank you for that information.

    122 watts does seem high for a Vcore of 1.248. Although there could be several reasons for this, it appears to scale corresponding to Vcore.

    My O/C is 4.7 at 1.280 volts (adaptive) with 114 watts at 76C (22C ambient) on air with Cooler Master TPC 812 (Dual 77 CFM fans) bareback on the processor die.

    Try Core Temp to compare wattage with HWMonitor. Let's see what you get.


    Essentially the same result:
  7. Best answer
    So to sum it up, using identical settings in BIOS at a consistent ambient temperature:

    (1) Wattage readings between Core Temp and HWMonitor agree.

    (2) Vcore readings between Core temp and CPU-Z do not agree with HWMonitor.

    (3) Core temperature readings between Real Temp, Core Temp and HWMonitor agree.

    This would seem to indicate that HWMonitor is misreporting Vcore, however, it obviously calls into question the accuracy of the processor's wattage being reported via the motherboard's super I/O chip.

    In simple terms, watts = volts x amps (current). I suspect that current is not being correctly measured and / or calculated, which would explain why the power is abnormal. This would point toward the motherboard as the cause.

    I believe your Core temperatures are normal and consistent with your Vcore, ambient temperature and your Phanteks PH-TC14PE high-end air cooler.

    Since your temperatures and Vcore are quite plausible, I believe the wattage readings are erroneous. I suggest that you simply keep a mental note of this issue and continue to run your rig.

    The bottom line is that I don't think this is enough of a problem to cause you any sleepless nights, or to RMA your motherboard.

    CT :sol:
  8. Makes me think of what kind of OC I could have gotten with a water cooler...

    On the other hand, maybe it's better to not tempt myself lol.
  9. You've done a good job.

    If you are indeed stable at 4.7 with 1.225, then since an increase of 40 to 50 millivolts is required for an increase of 100MHz, it's likely that you should be stable at 4.8 with 1.275.

    Your Core temperatures should still test less than 80C with P95 v26.6 Small FFT's running at 22C ambient. Since the wattage on your particular motherboard is erroneous anyway, just ignore it.

    What's important is to keep it under 1.300 volts and 80C.

    CT :sol:
  10. I actually tried 1.25v and 4.8 Ghz, made it into windows and partly into a stress test, but temperatures were around 85 and then I got blue screened.
    I have noticed that my VTT voltage is over 1.775, and I've read that it should not go over 1.4 (However, I'm not sure if that information is correct or not, be it outdated or just wrong)

    Do you have any thoughts on this, or does anyone with an OC'd i7 have any insight? I do have xmp enabled for my RAM, and I've read the that the VTT is for the memory controller in the processor, however my RAM only needs 1.5v so I'm not sure what my VTT should be at(if it's even at the wrong setting now)


    Sorry for all the questions! I just really want to make sure my system is running properly.
  11. jordan1794 said:
    Makes me think of what kind of OC I could have gotten with a water cooler...

    On the other hand, maybe it's better to not tempt myself lol.


    Your oc is limited by the vcore you can tolerate, not by the cooler.
    It is only vcore of over 1.3 that needs extra cooling, and that is not a good 24/7 way to go.
  12. geofelt said:
    jordan1794 said:
    Makes me think of what kind of OC I could have gotten with a water cooler...

    On the other hand, maybe it's better to not tempt myself lol.


    Your oc is limited by the vcore you can tolerate, not by the cooler.
    It is only vcore of over 1.3 that needs extra cooling, and that is not a good 24/7 way to go.



    Well as I understand it (And I am new to this, so I could be completely wrong) I could probably get a clock of 4.9 with a Vcore right at, or just over 1.3
    However with my temperatures reaching 85 at 1.25v with 4.8 Ghz, I think an increase up to 1.3v would definitely have me in the 90+ zone with my current cooler, which I'm not willing to go.

    If I did do such an overclock though I'd set-up with voltage offset, that way I wouldn't be at that voltage 24/7.
    Regardless though, I think I'm just proving more and more to myself that I shouldn't tempt it lol.
  13. As I mentioned above, you would need an increase of 40 to 50 millivolts (0.040 to 0.050) for an increase of 100MHz, so I would never expect that you would be stable at 4.8 with an increase of only 25 millivolts (0.025).

    Regarding your Core temperatures, in my previous post I somehow had it in my head that you were running on the edge of 70 instead of 80, so I apologize for my mistake. I follow so many temperature threads, and it's sometimes difficult to keep it all straight after being away from the Forums.

    Regardless, it seems that you were indeed at your ideal overclock with 4.7GHz. Keep in mind that the difference between 4.8 and 4.7 is only 2.1%, which is insignificant and imperceptible.

    jordan1794 said:
    I have noticed that my VTT voltage is over 1.775, and I've read that it should not go over 1.4. I do have xmp enabled for my RAM, and I've read the that the VTT is for the memory controller in the processor, however my RAM only needs 1.5v so I'm not sure what my VTT should be at.


    Concerning VTT, which is memory termination voltage, it's generally recommended to leave it on "Auto" As you haven't posted your motherboard model I couldn't do a search, but an increase of more than +0.25 over the default value is not recommended.

    VTT should typically be 1/2 the value of VDIMM, which in your case is 1.5, so VTT should be 0.750. If you add 0.025, then the max setting would be 0.775.

    Are you certain that you read 1.775 and not 0.775?

    This probably isn't your motherboard, but you might want to give this a read: GIGABYTE Z97X OC Guide - http://www.overclock.net/t/1490835/the-gigabyte-z97x-overclocking-guide

    CT :sol:
  14. CompuTronix said:
    As I mentioned above, you would need an increase of 40 to 50 millivolts (0.040 to 0.050) for an increase of 100MHz, so I would never expect that you would be stable at 4.8 with an increase of only 25 millivolts (0.025).

    Regarding your Core temperatures, in my previous post I somehow had it in my head that you were running on the edge of 70 instead of 80, so I apologize for my mistake. I follow so many temperature threads, and it's sometimes difficult to keep it all straight after being away from the Forums.

    Regardless, it seems that you were indeed at your ideal overclock with 4.7GHz. Keep in mind that the difference between 4.8 and 4.7 is only 2.1%, which is insignificant and imperceptible.

    jordan1794 said:
    I have noticed that my VTT voltage is over 1.775, and I've read that it should not go over 1.4. I do have xmp enabled for my RAM, and I've read the that the VTT is for the memory controller in the processor, however my RAM only needs 1.5v so I'm not sure what my VTT should be at.


    Concerning VTT, which is memory termination voltage, it's generally recommended to leave it on "Auto" As you haven't posted your motherboard model I couldn't do a search, but an increase of more than +0.25 over the default value is not recommended.

    VTT should typically be 1/2 the value of VDIMM, which in your case is 1.5, so VTT should be 0.750. If you add 0.025, then the max setting would be 0.775.

    Are you certain that you read 1.775 and not 0.775?

    This probably isn't your motherboard, but you might want to give this a read: GIGABYTE Z97X OC Guide - http://www.overclock.net/t/1490835/the-gigabyte-z97x-overclocking-guide

    CT :sol:

    VTT is displayed in HWMonitor as 1.775:


    Here are my BIOS settings regarding CPU voltage, if that helps in any way:
  15. I wouldn't count on HWMonitor being correct.

    Again, which model is your Gigabyte motherboard?

    VTT may be shown in BIOS under memory rather than CPU settings. Does your BIOS show any voltages which might be appropriate for VTT, such as Memory Termination Voltage or VDIMM / VDDR Termination Voltage?

    Does your BIOS even have a setting for VTT?
  16. Sorry, forgot to say my MB model is in the BIOS picture.
    (At the bottom, Gigabyte - Z97X-UD5H-BK)

    I actually think there was a memory termination voltage, let me check real quick.
  17. Only other voltage settings are under chipset voltage control, and they are PCH Core and PCH I/O.
    Both auto, Core was 1.090 I believe, and I/O was 1.5

    I'll check my "classic" BIOS and update this reply if there are any additional settings
    (I have a "smart tweak" mode and a "classic" mode)

    EDIT: Alright so in my regular BIOS I checked the voltages under PC Health Status and the VRIN voltage was displayed as 1.774. This is essentially the same as what HWMonitor is reporting as my VTT, perhaps HWMonitor is accidentally reporting this voltage because my VTT voltage is not available?

    My MB actually has a spot for me to check, with a multimeter, the voltage of VDIMM, unfortunately I don't have one.

    Maybe I'll grab a cheap one just to check over the next week...

    How plausible do you think it is that HWMonitor is reporting my VRIN as my VTT?
  18. jordan1794 said:
    This is essentially the same as what HWMonitor is reporting as my VTT, perhaps HWMonitor is accidentally reporting this voltage because my VTT voltage is not available?

    How plausible do you think it is that HWMonitor is reporting my VRIN as my VTT?


    As I mentioned in my previous post, I wouldn't count on HWMonitor being correct. Like your Gigabyte motherboard, my ASRock has no BIOS setting for VTT, so it's a non-issue.

    What clock speed is your memory?
  19. jordan1794 said:
    Well as I understand it (And I am new to this, so I could be completely wrong) I could probably get a clock of 4.9 with a Vcore right at, or just over 1.3
    However with my temperatures reaching 85 at 1.25v with 4.8 Ghz, I think an increase up to 1.3v would definitely have me in the 90+ zone with my current cooler, which I'm not willing to go.

    If I did do such an overclock though I'd set-up with voltage offset, that way I wouldn't be at that voltage 24/7.
    Regardless though, I think I'm just proving more and more to myself that I shouldn't tempt it lol.


    Your last underlined sentence is acquired wisdom!
    Let me say that again.
    Your last underlined sentence is acquired wisdom!
    Quote:
    I think I'm just proving more and more to myself that I shouldn't tempt it


    Every overclocker has to come to the realization they've overclocked as far as their cooling is safely going to allow and learn to stop before you damage your hardware.

    Either stop now, or get much better cooling than you have, to comfortably go further temperature wise.
  20. CompuTronix said:
    jordan1794 said:
    This is essentially the same as what HWMonitor is reporting as my VTT, perhaps HWMonitor is accidentally reporting this voltage because my VTT voltage is not available?

    How plausible do you think it is that HWMonitor is reporting my VRIN as my VTT?


    As I mentioned in my previous post, I wouldn't count on HWMonitor being correct. Like your Gigabyte motherboard, my ASRock has no BIOS setting for VTT, so it's a non-issue.

    What clock speed is your memory?


    2133 Mhz, I enabled XMP 1 in my BIOS and it appears (to my limited knowledge) that it set-up everything correctly to allow it to run at that speed. When I first booted up I think is was only running at 1600 or so.

    And to 4Ryan6 I'm quite happy with 4.7Ghz, especially with a lower voltage than what my processor was using for 4.4 turbo boost.
    I will admit, I expected lower temps from my Phanteks cooler, as it was supposedly the best air cooler on the market, but I guess I far under-estimated the heat that this processor puts out.
    From what I gather, some people have to get an aftermarket cooler just to keep this chip cool at stock speeds.

    Honestly I'll probably swap to water-cooling, but not until I need/really, really, want it.
    (Read: 2-3 years from now, at least)
    My old computer, that my wife is using now, has a i7 920 at only 2.67 Ghz, and it still games perfectly fine.

    Honestly, might be time to start thinking about OC'ing hers, WoW actually told her she didn't meet the minimum requirements! lol
    (Game runs smooth as butter though)

    Right now I'm at 1.225v for 4.7, going by the model of .04-.05v for each 100 Mhz I could, in theory, get right up to 4.9 Ghz with 1.305-1.325v, which is a little higher than what is deemed safe, but I'm sure using some creative ways (from googling I can see that there are many ways to stay stable with a lower voltage by changing some other things) I could probably get to 4.9 Ghz at 1.3v
    (And for pure short term bragging rights I'm sure 5 Ghz wouldn't be out of the question, but definitely not a 24/7 OC)

    But, as I said before and you quoted me twice lol, it's probably best that I don't have water cooling right now. Perhaps by the time I am ready for it (2-3 years or so) I'll also have the skill-set to make it work.

    Thanks again for all the help guys, I really appreciate the time you have taken to help me learn new things, from this one post I feel like my troubleshooting skills/overclocking knowledge have increased 10 fold haha.
  21. jordan1794 said:
    CompuTronix said:
    What clock speed is your memory?


    2133 Mhz, I enabled XMP 1 in my BIOS and it appears (to my limited knowledge) that it set-up everything correctly to allow it to run at that speed. When I first booted up I think is was only running at 1600


    Keep in mind that your i7 4790K only supports memory up to 1600MHz: Intel® Core™ i7-4790K Processor
    (8M Cache, up to 4.40 GHz)
    - http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

    This means that 2133MHz memory is forcing the memory controller on board the processor to run overclocked, which can increase temperatures within the CPU die. There are also reports that some BIOS versions with certain motherboards will run the processor hotter with XMP enabled.

    Here's a good Sticky on the topic which 4Ryan6 wrote: Overclocking the CPUs Memory Controller, Effect on a Raised Multiplier CPU Overclock - http://www.tomshardware.com/forum/id-2450639/overclocking-cpus-memory-controller-effect-raised-multiplier-cpu-overclock.html

    CT :sol:
  22. I think, just for testing/curiosity, I'll de-active XMP and re-run a stress test and see if my temps are significantly different or not.
  23. Disabling my XMP did absolutely nothing lol.
    VTT voltage was the same. Temperatures were precisely the same, CPU wattage was the same, and so forth.

    I do remember reading up and making sure that my RAM would operate on 1.5v because intel CPU's don't like RAM over that (And really, really don't like it over 1.65)

    But at this point I'm far beyond my limited knowledge.
  24. Lol
  25. When you disabled XMP, did you downclock your memory to 1600MHz?
  26. Automatically did it(I checked with CPU-Z)
  27. Unless we've missed something, I think you've chased down the bugs. If I were you, I would declare victory, sit back and enjoy numerous, large overclocked beers! :D

    Enjoy your rig!

    CT :sol:
  28. CompuTronix said:
    Unless we've missed something, I think you've chased down the bugs. If I were you, I would declare victory, sit back and enjoy numerous, large overclocked beers! :D

    Enjoy your rig!

    CT :sol:


    Yep, thanks again for all your help!
    If I could select two of your answers for best answer I would! haha
  29. Just thought I'd revisit this and give an update.
    I was able to get to 4.8 Ghz with a vcore of 1.24
    And am now at 4.9 Ghz with a vcore of 1.3

    I swapped out my Phanteks for a LEPA Aquachanger 240 AIO Closed Loop cooler and my temperatures at 4.9 w/ 1.3 vcore are still cooler than the Phanteks was running at 4.7 with 1.225 vcore.
    (82-83-83-79)
  30. My system agent is actually at 1.050.
    The only way to set it with my MOBO is via an offset from the vring voltage, which is currently stock 1.050 (with 4.4 uncore overclock)
    I was bumping it up to +.150 (so 1.2) while trying to overclock my RAM, but my RAM does not overclock lol.
    Runs perfectly at 2133 with 1.5 DRAM, but even with 1.65v I can't get anything higher to be "game stable"
    So I'm back to +.000 (so 1.050)

    As for my actual vcore, using cpuz I get 1.3
    Using HWMonitor, which is decidedly inaccurate, I get 1.32
    (HWMonitor tells me my VTT is 1.776v....in fact, is is reporting my vrin override instead...)
    When booting into my BIOS, it fluctuates between 1.299 and 1.300
    I could get out my multimeter and check on the MOBO, but after you read the next paragraph you'll see it's not needed right now.

    I'm actually sticking with 4.8 GHz at 1.25 vcore for my daily usage.
    I am stable with 4.8 GHz at 1.24, but that's running Prime...and sometimes you can still crash when running multiple individual heavy loads even if you are stable with Prime, so 1.25 is where I'm at.
    I didn't test it to see if gaming would be unstable, but the drop of 15 mv only brought my temperatures down a couple of degrees, so I'm not worried about running it as low as possible.


    To give an idea of how inaccurate HWMonitor is, I present a table from my testing:

    Set vcore | CPUZ reading | HWMonitor reading
    1.300 | 1.300 | 1.320
    1.295 | 1.295 | 1.320
    1.290 | 1.291 | 1.296
    -------No data for this range-------
    1.265 | 1.266 | 1.284
    1.260 | 1.261 | 1.272
    1.255 | 1.257 | 1.272
    1.250 | 1.252 | 1.260
    1.245 | 1.242 | 1.260
    1.240 | 1.237 | 1.260
  31. geofelt said:
    If you can do 4.7 with a vcore of 1.2, congratulations, you won the bin lottery.

    The consensus is that vcore voltages higher than 1.30 are not a good thing for 24/7 usage.
    I have been unable to find any official Intel recommendation on what is a safe vcore limit.

    I would take the time to set up speedstep(C1E). That lowers the multiplier and voltage when the cpu has little to do.
    It is a very confusing area, and the settings are named differently in different motherboard bios'es.

    I have abandoned prime95 as a stress tester. Plus I can't seem to get rid of the program, and that bothers me.

    I use OCCT which is probably more representative of what we normally do.
    It will shut down the test at 85c.


    i can do it with lover vcore
  32. Riivio said:
    geofelt said:
    If you can do 4.7 with a vcore of 1.2, congratulations, you won the bin lottery.

    The consensus is that vcore voltages higher than 1.30 are not a good thing for 24/7 usage.
    I have been unable to find any official Intel recommendation on what is a safe vcore limit.

    I would take the time to set up speedstep(C1E). That lowers the multiplier and voltage when the cpu has little to do.
    It is a very confusing area, and the settings are named differently in different motherboard bios'es.

    I have abandoned prime95 as a stress tester. Plus I can't seem to get rid of the program, and that bothers me.

    I use OCCT which is probably more representative of what we normally do.
    It will shut down the test at 85c.


    i can do it with lover vcore



    Erm, not that it's a competition or anything...but you are running 4.6 @ 1.19 vcore.
    I can do 4.7 @ 1.2
    And 4.8 @ 1.25
    And 4.9 @ 1.3 (But temps reach around 82-85 so I stick with 4.8)
    I've also swapped to a LEPA Aquachanger 240 since this posting, the switch dropped my temperatures around 10 degrees across all cores.

    I could actually probably do 4.7 at around 1.15 or 1.1
    I never even bothered to see how long I could go at 4.7

    With 4.8 though I know 1.25 is the lowest, 1.24 boots and runs prime but crashes in real world use (2 hour long video editing)

    You still have a good chip though, and with the temps you have there you surely could hit 4.7 or 4.8 with a vcore still at or under 1.3
    I'd suggest a better cooler though, i see your max temps are reaching over 80
  33. jordan1794 said:
    Riivio said:
    geofelt said:
    If you can do 4.7 with a vcore of 1.2, congratulations, you won the bin lottery.

    The consensus is that vcore voltages higher than 1.30 are not a good thing for 24/7 usage.
    I have been unable to find any official Intel recommendation on what is a safe vcore limit.

    I would take the time to set up speedstep(C1E). That lowers the multiplier and voltage when the cpu has little to do.
    It is a very confusing area, and the settings are named differently in different motherboard bios'es.

    I have abandoned prime95 as a stress tester. Plus I can't seem to get rid of the program, and that bothers me.

    I use OCCT which is probably more representative of what we normally do.
    It will shut down the test at 85c.


    i can do it with lover vcore



    Erm, not that it's a competition or anything...but you are running 4.6 @ 1.19 vcore.
    I can do 4.7 @ 1.2
    And 4.8 @ 1.25
    And 4.9 @ 1.3 (But temps reach around 82-85 so I stick with 4.8)
    I've also swapped to a LEPA Aquachanger 240 since this posting, the switch dropped my temperatures around 10 degrees across all cores.

    I could actually probably do 4.7 at around 1.15 or 1.1
    I never even bothered to see how long I could go at 4.7

    With 4.8 though I know 1.25 is the lowest, 1.24 boots and runs prime but crashes in real world use (2 hour long video editing)

    You still have a good chip though, and with the temps you have there you surely could hit 4.7 or 4.8 with a vcore still at or under 1.3
    I'd suggest a better cooler though, i see your max temps are reaching over 80



    I can hit 4.9 @ 1.27 and 5.0 @ 1.33 stable ;)
Ask a new question

Read More

Overclocking Intel i7 Temperature