FRONTPAGE Intel i7 4790K Devil's Canyon CPU Review

EarthDog said:

Not sure what to tell you. There had to be other variables involved? Across dozens of CPUs I have inadvertently tested this on, I saw no changes. Its commonly held it makes no difference unless you go cold or have a much bigger difference.

Glad you got to drop the voltage regardless of the reason though.

Click to expand...

Like i said, i dont doubt you are right. I just have a personal experience, that differs from the countless testing you have done.

The one and only reason that could have an effect as far as changes, would have to Bios update, as i highly doubt that adding a third GPU will lower CPU volt usage

EDIT:
I also did change from EK block to heatkiller blocks on the GPU's which took the GPU temps at least 10degrees per card.

In my case:
[email protected] 1.25V = ~80*C max , ~1.15 VID
[email protected] 1.25V = ~62*C max , ~1.09 VID
the same cooling, only switched CPU

CPU vrm frequency at 400kHz let me stabilize CPU clock one ratio higher than on stock frequency ( 300kHz ) and I'm using Z87M OCF board.
Additional options like PLL voltage, vrm frequencies etc can help in lowering CPU voltage. All depends from CPU itself. Ambient temps are not really changing voltage requirements. As ED said it's generally starting to affect CPU voltage at sub zero temps.
Actually not many CPUs are overclocking better after delidding and in best cases it's ~100MHz higher clock. All have visible temp drops but in most cases there are no differences in required voltages or max clock after OC. I think that most users who see OC improvements were reaching thermal throttling limit or stability limit ( which depends from CPU and can be below 90*C above some clock ).

What I see so far is that DC is overclocking better than last 4670/4770K series. Average results in the web are at least 200MHz higher for DC at similar voltage. When average 4770K needed 1.35V for 4.5GHz then now we see average 4790K which needs 1.25V for the same clock. Even if it won't make higher clock then at least it will run cooler so it also won't require any special cooling.

Another bios could certainly do it... If you were comparing apples to apples, I would buy it, but since other variables changed, I'm going to stick with what I learned and have seen confirmed on the web over the past few years.

temps getting on my nerves, and want to see if benches different...though mainly out of curiosity. I just used vice only method, on pcb on one side, on ihs other, and with duct tape on vice...turned slowly til gave. Take me a while to clean it...have cl pro coming in 2-3 days, but will use some other paste til gets here.

Duct tape to delid? That's BALLER!

The toughest part was cleaning off the adhesive.

it is clean, boots and runs fine, but the ihs was sliding all around so crappy tim install, but made sure it works. temps cooler already, but going to use different tim1 and then some testing.

here it is cleaned.

I'm leak testing and trying to get the air out of the new loop right now. I should be up and running by tomorrow and will report temps stock, and after I fiddle with some overclocking.

Mpegger, let us know your temps.


holy crap, after delidding, I just ran cinebench 3x in a row at 1.36v core 5ghz. I needed 1.45 to run about 1 in 3x successfully, and 1.465 to consistently run it. I just walked it down to 1.36v havent crashed yet. Some on oc.net were saying they saw no changes with ivy in mhz, but were seeing it in DC. I dont know about 24/7 volts, but benching on ragged edge of stability, so far made a huge difference.

my temps dropped 13C at prime 28.5 small fft load 4.7ghz 1.28v, before and after delidding 10 min runs of 12k small ffts prime blend 28.5.

edit: well thought I might get 5.1 cin R15 now, but that is still a no go. Made it through about 1/2 to 2/3 but cant make it all way with 5.1hz, tried up to 1.47v.

made a really crappy video of 1.366v running 5ghz cr15, can see volts couple times, but not a good video.

you dropped 13c from deliding?!?!?!

yep, about 10C on 4.6ghz 1.245v setting, 13C on 4.7ghz 1.285v.

dont think it changed my vcore necessary for 4.6 stable though, still crashed at same too little vcore. dont know about 4.7/4.8 yet.

but did reduce vcore necessary for 5gz cinebench, and temps dropped from 90's to 80's C at same vcore.

dang.... my 4790 does 4.7 at 1.23v and temps are about 55 how how was yours at 4.7??

im running 2 480 rads for my cpu and videocards

weakssauce said:

dang.... my 4790 does 4.7 at 1.23v and temps are about 55 how how was yours at 4.7??

im running 2 480 rads for my cpu and videocards

Click to expand...

To my eyes, that looks like a good chip.

so far it looks like I got the worst chip on OCF

Quick setup info: Swiftech MCR320-QP, Swiftech MCP655 w/Alphacool HF D5 Top, and Swiftech Apogee HD cpu block w/AS5 as the TIM.

With all UEFI settings to Auto (current air temp at intake 81F, water temp 78F [not sure how accurate water temp probe is]): Idles around 37C (about 0.8ish VID), 100% load around 60C (about 1.18 VID during test, but I have seen the VID go as high as 1.2, though not during the test).

I'm debating re-installing Windows from scratch on the new SSD before I try to overclock this chip (I just cloned it from the old). Although things are running like normal, I've already had various issues and wanna try to get the best possible overclock I can with this cpu, so I'd like to start with a nice clean slate. But I really, really, really, really, REALLY dread having to re-install everything and redo all my settings to have things work like normal again. Ugh. I may end up spending the entire 3 day weekend doing so.

weakssauce said:

dang.... my 4790 does 4.7 at 1.23v and temps are about 55 how how was yours at 4.7??

im running 2 480 rads for my cpu and videocards

Click to expand...

I can boot, game, run prime for 20 mins on 4.7 at 1.23v, but to be prime 28.5 and 27.9 10hr stable, I need 1.28v, and always run mine .005 to .01 higher, I want to have same stability as stock.

Gaming temps now in high 40's, low 50's.

For rest of my temps, see post 121. Though now I am 10-13C cooler temps, depending on voltage/prime version.

i cant get any higher than 4.7 though

i tried 4.8 at 1.35vcore but fails on prime instantly

im just messing with vcore and block not too sure about all the other voltages...

I made some tests on 4690K using ss cooling ( about -45*C idle ) and I simply wasted half of the day.
- 5GHz 1.5V is passing cinebench 11.5 and R15
- 5.1GHz 1.55-1.60V is barely passing any longer tests
- 5.2GHz 1.60-1.70V is barely booting

Simply my 4670K even though is running at 15*C+ higher temps, is overclocking better.

Since intel said they binned these, and 4790k stock specs are 4ghz turbo to 4.4, and 4690k is 3.4ghz stock to 3.9, wonder if ones that didnt make 4790k specs within vid range make the majority of 4690k, hence 4690k clock worse on avg...or you just got unlucky in lottery, or some of both.

Mine will run superpi 32m all day at 5.1ghz, 1.44v before delidding at normal 25C ambients...havent tried lower v. after delidding. But cinebench 5.1 still no go.

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

@RGE , Each day I see more 4790K that are clocking the same as my 4690K in the web ( reviews, forums etc ). I don't think there was any special binning of DC. Only 4690K and 4790K are on the same improved chips so it's probably the same as with 4670K and 4770K where average chips were similar. 4790K is more popular among overclockers so there will be more results at higher clock. Most people who got bad chips won't share on the forums as always. Gamers will more often pick 4690K and they won't overclock it really high so making good statistics which one is better will be really hard.
You had a lot of luck with your CPU it seems. For me it's 8th bad haswell in a row. 1st DC but I don't think I will buy more.