• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Going to OC my 4790k, input appreciated

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
My next pass at 4.9 I'm going to remove the heat spreaders from the 2 sticks of RAM and lower that one fan down to get the most airflow I can over the lower portion of the heatsink. I just want to ensure I have done all I could as far as the cooler is concerned. Then I will pull the cooler and check the spread of the thermal paste.

Temps do look good but I will feel better knowing I checked things out.

Something to note when it was at 4.9 and 1.3V, when I did a benchmark on the RAM, I noticed my cursor lagging somewhat and my short video render took slightly longer. Could this be signs of instability of a high clock or low voltage....or both?

Also, with the CPU at 4.8 or 4.9, how does the memory keep pace or does it go stride for stride when the CPU is oc'd.?

I'm learning as I go so bear with me......:)
 
I would say a sign of slightly unstable, raise the vcore .05 and try again.
just leave the ram be for the time, all messing around with ram will get you now is unstable.
ram clocking can come later, cpu is KING!!!
 
I would say a sign of slightly unstable, raise the vcore .05 and try again.
just leave the ram be for the time, all messing around with ram will get you now is unstable.
ram clocking can come later, cpu is KING!!!

Will do, thanks.


Just need to see if you can get it to pass at least 2 hours of Aida, 4.8-4.9 is a darn respectable Oc on any cooling especially Air.

Well, I kinda fell asleep..... I put my head down for what seemed like a second. So not many tests today except this 6 hr run.....oops .:)

I removed the heat spreaders from the RAM, lowered the fan about 1/2".

4.9 Ghz @ 1.310V The graph is I believe the last hour of the test.

I also ran Geekbench, and haven't noticed the cursor dragging, that happened before only during the AIDA memory benchmark while it was populating the values so maybe that's why.

Edited to add a few more screenshots.
 

Attachments

Last edited:
The plan today was to lower the fan, run a 30 min test then compare. I also want to remove 2 sticks of RAM and try a few 30 min tests before a 2 hr run. Will have plenty of time this weekend, obviously the 2 hr temp averages will be lower and I would like to find a more consistent peak core temp in the mid 70's even if that means backing down the clock a little unless what I have is acceptable.
 
ok, give us a direction you want to go with this rig, gaming, rendering videos, bench for grins and giggles, like I do, all of the above and more, none of the above?
benchmarks will load and heat a rig much more than gaming or.... a lot of other stuff.
 
Your temps are fine, as Caddie mentioned, you most likely will never see to see temps during normal usage. Again that seems like a pretty good chip you have there. Those temps are a lot better then the heat monger 4770k I have that's for sure.
 
I would like to see a prime95 run.
just to see if his is better than mine, as much as I have into it I might make him an offer.............
 
Sorry for not properly quoting the last few replies as this phone is sort of a pain to multi quote.

I want to use this as my day to day rig, including video rendering, batch photo editing as well as gaming. Benchmarking is definitely part of the plan as this has become quite interesting.

I enjoy getting my hands dirty and trying to figure out what makes all this tick, has intrigued more than I thought. I would like to see what this rig can do and then tweak it to my tasks.

I may not keep it at the highest level of OC but not completely sure of that yet either....depends on what we can get out of it.

I will run Prime 95 as well.

Definitely have to give credit to the awesome D15 cooler as this has exceeded my expectations. ...again I appreciate the push for it.....:thup:

CD.....I may not be able to give this one up...lol.

Forgot to mention.....I read that Prime 95 and the Haswell chips didn't work well together...something about overvoltage. I assume that's with earlier versions of Prime and would the fixed cpu voltage set in the BIOS override that anyway?....Just want to double check.
 
Last edited:
For the AIDA64 tests, should I leave the local disks and GPU disabled? I assume this is the case but I noticed they were not checked for the stress test.
 
For the AIDA64 tests, should I leave the local disks and GPU disabled? I assume this is the case but I noticed they were not checked for the stress test.
Yes you don't need to stress the Local disks or Gpu.
 
Which Prime 95 test do you want me to run? I tried a couple real quick and this thing really jacks the temps up...this program is brutal...lol.
 
you still have a way to go on the cores yet buddy.......
time to change the all core selection in bios and start in on per core clocking.
it's like this, when it's only using one core (90%of the time) it runs at the one core clock, when it uses 2 cores, it runs at the 2 core clock and so on.
save this clock as a profile in the bios and start core clocking.
mine is set up 1core=5.2, 2 core= 5.0, 3 core 4.8, 4 core, 4.8.
it takes a hefty shot of vcore but with most of what we do being 1 core it boots like a rocket!!!!!

- - - Updated - - -

p95, blend, turn off hyper threading.
it make a lot of heat and has little to do with real world.
get it to run 30 minutes, core temps to MAX, peak at 80c.
 
Great...thanks. :thup: I used the newer version of Prime 95...ver 28.7 and apparently the Haswell chips should not use that version, only 26.6 & earlier. The chip throttled down almost immediately. I will try 26.6 and go from there.
 
yes, 28.7 will make a lot of heat and it'llll take more vcore than 26.6 and most will wind up around 1.3-1.35 vcore with it at 4.8.
 
yes, 28.7 will make a lot of heat and it'llll take more vcore than 26.6 and most will wind up around 1.3-1.35 vcore with it at 4.8.

Yeah no doubt....alot of heat is an understatement.....all cores hit about 100 deg in a couple seconds so I'm definitely not using that version. I don't see a ver. 26.6 for windows...do you have a link I could try?

Thanks.
 
26.6 is much lighter in heat but still more heat than intel XTU, for the game rig i tweak to XTU.
 
I did get the prime 95 ver. 26.6 loaded...thanks again for that link...currently running the blend test. I'm running AIDA64 in the background, not as a stress test but to monitor temps only. Let me know if this is the correct test but I started the stress test and took a screenshot after 30 min. I noticed only the top field was populated then I clicked on the "Status" tab then the rest of the fields populated. I don't know if that's because I clicked on the status or it just does this after 30 min.

So I kept it going for another 30 min and took another screenshot. The timers both indicate 30 min but the bottom 2 pics are after 1 hr.

As discussed earlier, I did remove the cooler to check the thermal paste pattern, consensus was the pattern was normal and all was good. Only thing I did differently was because the center contact patch was slightly elongated, I applied the paste like a grain of rice instead of a round dab. I ran 2 30 min AIDA tests and the temps were 1 deg higher in 1 core but 1 deg cooler in 2 others so at least it wasn't a step backwards. I understand each subsequent test can yield different results but should be good to go.

This is @ 4.9 Ghz, 1.310V v-core, HT enabled

I was able to post this while Prime 95 was running and didn't notice any cursor lag or odd hangups.
 

Attachments

  • prime95 30 min.jpg
    prime95 30 min.jpg
    209.9 KB · Views: 60
  • prime95-30 min-t.jpg
    prime95-30 min-t.jpg
    302.6 KB · Views: 58
  • prime 95 1hr-b.jpg
    prime 95 1hr-b.jpg
    505.5 KB · Views: 62
  • prime95 1hr-a.jpg
    prime95 1hr-a.jpg
    540.1 KB · Views: 60
I spoke to soon on posting updates when Prime was running....90 min into test it shut down while I was opening forum so I will let it go for 2 hrs (hopefully) then update.
 
Back