FRONTPAGE Intel i7 4790K Devil's Canyon CPU Review

RGE · Jul 4, 2014

weakssauce said:
so if i was to delid my 4790 what paste should i use? i only have as5 and antec nano diamond :/

Best advice is dont delid, unless you want to for curiosity and you consider your cpu disposable and will just buy a new one if it dies. Also I wouldnt plan on your 24/7 going much higher unless your heat limited or borderline stable with high temps. Though it did help me with benching on edge of stability.

Regarding tims, I wouldnt use either of those, as5 is conductive and no better than non-conductive tim and diamond might scratch die surface.

For paste tims you will be dealing with pump out, meaning may have to reapply every few months as temps increase based on others experience that delidded with haswell. Though paste tims can be nonconductive and dont have to worry about vrm components under IHS.

Im going to use CLP (liquid metal) and paint the vrm components with clear nail polish, and hope CLP forms a semi-permanent bond.

But for me this is just an experiment out of curiosity.

Woomack said:
For me it's 8th bad haswell in a row

ouch...and yeah I got a decent clocker in the DC lottery.

Mpegger · Jul 4, 2014

I decided to leave my Windows install as is and try a little overclocking. The new bios for the board included alot more options for voltages, but also a "Overclocking Profile" with 4.5, 4.6 and 4.7 Turbo settings that automatically set various voltages for you. I went straight for the 4.6 and did a quick 1hr run with OCCT, Large Data set. See my previous post for my setup. Temp at case intake was around 79F-80F this time around, and the voltage set for the cpu was 1.278 (I forgot exactly what it was so don't quote me on that

). What I found strange was how inconsistent the temps were compared to previous setups.

RGE · Jul 4, 2014

4.6 looks good, temps look the norm....on auto watch for some motherboards are overvolting vring too much on early bioses....like 1.4 on some, when stock is 1.05 and I only need 1.22 up to 47 uncore.

Spawn-Inc · Jul 4, 2014

so it looks like my thoughts of running 5ghz daily isn't going to happen unless i get lucky

my only hope now is that hopefully my 3x3 rad setup will give me an edge.

weakssauce · Jul 4, 2014

do you recommend that liquid metal for ihs to waterblock? or just for the die

RGE · Jul 4, 2014

weakssauce said:
do you recommend that liquid metal for ihs to waterblock? or just for the die

only between die and IHS...most use col. liquid pro or liquid ultra....doesnt seem to suffer from pump out and thermal conductivity is ~40 w/mk (ignoring the inflated clp listing) or about 4-5X that of paste.

between IHS and waterblock just use nonconductive tim pk1, mx2/4 etc as there is only 1-2C difference between tims at higher surface area between wb/ihs, and pump out isnt a big issue since paste tims designed for that location.

Spawn-Inc said:
so it looks like my thoughts of running 5ghz daily isn't going to happen unless i get lucky my only hope now is that hopefully my 3x3 rad setup will give me an edge.

I think if you get a good chip you can bench 5ghz at ambient, but havent seen any yet get close to 5ghz 24/7. Unless going subambient with good chip.

weakssauce · Jul 5, 2014

cool cool ima order some and see if it helps

Mpegger · Jul 5, 2014

rge said:
4.6 looks good, temps look the norm....on auto watch for some motherboards are overvolting vring too much on early bioses....like 1.4 on some, when stock is 1.05 and I only need 1.22 up to 47 uncore.

Although the temps are good, I'll eventually play with the VID to try and get it lower, as well as try for 4.7. What bothered me about the temps was how much it fluctuated. On all my previous cpus, you can clearly see the average temp. The 4790K however, is just all over the place with its temps.

RGE · Jul 5, 2014

I was going to ask if your last cpu had solder tim....looks like it did. Welcome to the temp fluctuations of the crappy tim1 club!

Spawn-Inc · Jul 5, 2014

What's the max daily voltage for the 4790k?

RGE · Jul 5, 2014

23C drop in core temps using CLP for ihs/die after delid, at 4.7 w/1.287v, prime 28.5 small ffts, from 98C max to 75C max (see pic below)
(all runs use pk1 for wb/ihs tim)

prior to delid,
4.7 ghz w/1.287v, prime 28.5 on small ffts repeating 12k for 7 mins, temps 98C on hottest core.

After delid with pk1 tim between ihs/core (note this pk1 tim was not very well applied, ie too thick)
4.7ghz w/1.287v, prime 28.5 on small ffts repeating 12k for 7 mins, temps 85C on hottest core

After delid with CLP between ihs/core
4.7 ghz w/1.287v, prime 28.5 on small ffts repeating 12k for 7 mins, temps 75C on hottest core.

Pic of CLP applied: this is one really small drop spread around.

RGE · Jul 5, 2014

Spawn-Inc said:
What's the max daily voltage for the 4790k?

vid goes up to 1.27v, so intel thinks that is ok.

Some say stay under 1.3v

Some say stay under 1.35v.

Nobody really knows, since no one but intel knows mttf given a volt/temp.

some using 1.4+ have claimed degradation.

Im testing 4.8 w/1.31 now, im expecting that to fail prime blend after a while. Im probably going to end up 24/7 at 4.8 w/1.315 to 1.325 somewhere in that range.

Mpegger · Jul 5, 2014

rge said:
23C drop in core temps using CLP for ihs/die after delid, at 4.7 w/1.287v, prime 28.5 small ffts, from 98C max to 75C max (see pic below)
(all runs use pk1 for wb/ihs tim)

prior to delid,
4.7 ghz w/1.287v, prime 28.5 on small ffts repeating 12k for 7 mins, temps 98C on hottest core.

After delid with pk1 tim between ihs/core (note this pk1 tim was not very well applied, ie too thick)
4.7ghz w/1.287v, prime 28.5 on small ffts repeating 12k for 7 mins, temps 85C on hottest core

After delid with CLP between ihs/core
4.7 ghz w/1.287v, prime 28.5 on small ffts repeating 12k for 7 mins, temps 75C on hottest core.

Pic of CLP applied: this is one really small drop spread around.

So lower temp by 10-20C, but not much higher overclock? Although cringe worthy, I may go for delidding if temps are a problem here (no a/c, summers in the 90F+ easily). These past two days haven't been bad though. Today has been the highest at 85F (82F intake).

RGE · Jul 5, 2014

Mpegger said:
So lower temp by 10-20C, but not much higher overclock? Although cringe worthy, I may go for delidding if temps are a problem here (no a/c, summers in the 90F+ easily). These past two days haven't been bad though. Today has been the highest at 85F (82F intake).

4.6ghz and 4.7 ghz require same vcore for stability before and after delid on mine, within my ability to detect difference, ie ~0.005v which granted is large range.

4.8ghz, jury still out, testing now, but just crashed at 1.315, so probably going to end up near same vcore.

5.0 ghz cinebench was clearly stable with less vcore after delid (tested that many times tightening ram/uncore), but 5.1 still utter fail even with CLP.

So yeah bottom line, no higher mhz benching or 24/7. But delidding allows me to prime/test 4.8ghz with reasonable temps no longer reaching tjmax.

Mpegger · Jul 8, 2014

Is anyone using the on-die video along with overclocking? Would it affect overclocking of the main cpu at all (added heat)?

I noticed my 780TI isn't down clocking and it's probably the stupid dual monitor "bug" or "feature", however you look at it, that nVidia has in their drivers. Strange though, as I clearly remember it down clocking perfectly fine on the previous setup.

Evilsizer · Jul 8, 2014

thats a good question, my cpu being different. i notice my temps are the same, i was using ondie for a bit. at [email protected] with the igp all stock speed, load temps in prime95 for each core were 70/69c area. seems about the same now using the GTX660 when the cpu is under load, maybe 2c if that. i notice full load temp changes a bit depending on the size of the work being done in prime95. max load at some point according to coretemp from the max temp always hits 70c on one of the cores.

outhouse · Jul 12, 2014

rge said:
vid goes up to 1.27v, so intel thinks that is ok.

Some say stay under 1.3v

Some say stay under 1.35v.

Nobody really knows, since no one but intel knows mttf given a volt/temp.

some using 1.4+ have claimed degradation.

Im testing 4.8 w/1.31 now, im expecting that to fail prime blend after a while. Im probably going to end up 24/7 at 4.8 w/1.315 to 1.325 somewhere in that range.

Your chip sounds like mine.

4.8 at 1.325 and 75C under load, prime95. 55c gaming.

I really need to delid this thing. I like your results. All I have is the EK GPU tim at the moment. At least it is non conductive

RGE · Jul 12, 2014

outhouse said:
Your chip sounds like mine.

4.8 at 1.325 and 75C under load, prime95. 55c gaming.

I really need to delid this thing. I like your results. All I have is the EK GPU tim at the moment. At least it is non conductive

For running prime 28.5 I had to delid to prevent throttling at 100C during small ffts on blend, even at 4.7ghz it came close.

outhouse · Jul 12, 2014

rge said:
For running prime 28.5 I had to delid to prevent throttling at 100C during small ffts on blend, even at 4.7ghz it came close.

Maybe 75C at 1.325v isnt so hot afterall.

I swear my temps jumped 10C going to this new EK block over my old RBX.

RGE · Jul 12, 2014

outhouse said:
Maybe 75C at 1.325v isnt so hot afterall.

I swear my temps jumped 10C going to this new EK block over my old RBX.

What are your temps with prime 28.5 small ffts at 1.325v and 4.8. If they are under 90 that is unusually good.

FRONTPAGE Intel i7 4790K Devil's Canyon CPU Review

Senior Member

Member

Attachments

Senior Member

Member

Registered

Senior Member

Registered

Member

Attachments

Senior Member

Member

Senior Member

Senior Member

Member

Senior Member

Member

Senior Forum Spammer

Senior Member

Senior Member

Senior Member

Senior Member

Similar threads