Intel 4790K OC Report ~

Intel 4790K ES C0
Cascade and LN2 limit testing

http://valid.canardpc.com/w90q4y

Single core: Super Pi 32M
Threaded: Cinebench 11.5
Asus MVIIG and Gigabyte z97x-OC Force
Corsair Dominator Platinum 2666C10
Corsair 240GB Neutron
Enermax 1500W MAXRevo

Thermalright 120 heat sink
4 Stage phase change cooler
K|ngp|n F1 LN2 pot

VID 1.104v
Boot into windows 1.25v @ 4.4Ghz

Mhz Vcore Cooling Input Temp Threads
4500 1.25 air stock 35c
4600 1.25 air stock 35c
4700 1.25 air stock 35c
4800 1.3 air stock 35c
4900 1.35 air stock 35c
5000 1.4 air stock 35c
5100 1.42 phase stock 35c
5200 1.44 phase 1.85 -109
5300 1.46 phase 1.88 -109
5400 1.48 phase 1.9 -109
5500 1.5 phase 1.95 -109
5600 1.52 phase 1.975 -109
5700 1.55 phase 2 -109
5800 1.6 phase 2.1 -109 4c8t
5900 1.7 phase 2.2 -109 4c4t
6000 1.8 phase 2.3 -130 4c8t
6100 1.85 LN2 2.4 -135 4c8t
6200 1.9 LN2 2.5 -140 4c8t
6300 1.92 LN2 2.5 -142 4c8t

1.95vcore = instant crash @ -140c
2.6v input = instant crash

I need to test the memory controller more
Memory controller maxed out a little over 1.4v -130c 5.5Ghz, 1.4v @ 5.2Ghz seems to be a close max stable

Cold bug factory defaults in bios at -120c
Cold bug 5.5Ghz boot -140c
Cold bug 6Ghz 4c8t idle -141c ~ loaded -150c

You gotta get chip warmer to get it colder
It clocks better and goes colder with 4c8t, removal of HT and cores makes cold bug worse.

This chip is binned for review sample, it should be close to what you get from retail.
With the new chip you get more Mhz on air and water but you can see the wall trying to reach the last few multilples on the top end.
This chip is better than any retail 4770K that I have tested in the past year.

Two of the major changes help with getting all retail chips to 5ghz on air, The TIM and the added resistor packs on the back of the CPU allow more voltage and that pushes the chip a little harder to push past 6Ghz.

I could see a lot more going on with cold and voltages, what was making the chip go faster and take more cold. Running all cores with HT and stressing the memory controller increased the internal load thus moving the cold bug from -120c to -140c. Example: Hyperthreading on and off is a 10c swing on the cold bug.

I can imagine without this new TIM and resistor packs that this chip would be a 5.6Ghz chip, so yes the improvements help the chips OC on the top end a little better vs the retails we have now, there should be a lot more 6Ghz+ chips to be found. The trade off is that it will take more voltage to push that chip up hill.

Giving the chip 1.8v and 1.9v seemed to be easy but at the trade off of max operating Mhz on all cores, so finesse must be taken to meet the temp/volt steps. My steps above in 100Mhz increments, the voltages and temps needed to make each step is greater than before for last multiples. The 4770K usually wouldn't give up the multiples when pushed too hard. I am sure we will see 2 volts again real soon. Whoever hits the 7Ghz mark will be sporting some crazy voltages.

I really hope that single stage phase change coolers can make a come back with this new 4790K, I am sure my chip could do 5.8Ghz 4c8t 24/7 and pull off 6.1Ghz screen for bragging rights.

Now that I know to push the chip it needs vcore and cooling = more Mhz. When you reach past 5.5Ghz you will need to add input voltage above 2v

My chip will do 5Ghz on air with a decent heat sink. I am using this one:
http://www.newegg.com/Product/Produc...-125-_-Product

If you need to see 5Ghz from your 4790K just crank the vcore to 1.4v to 1.5v in XTU, increase multiples to 50x with cupid open. Set it back and you are done. Do not expect 5Ghz air machines, you need water cooling for 5Ghz+ 24/7 operation.
No one promised 5Ghz air cooling processors, be realistic and settle for 4.9Ghz
My chip is an ES and I have been to 1.7v on air and would have tried 1.8v if I made progress at 1.7v.

http://valid.canardpc.com/b8wwe5

For serious overclocking a higher capacity heat sinks are needed, the water cooling segment will benefit and after many hours of plugging at temps and voltages I am back to the same conclusion that phase change is the way to go on this generation of processor to open it up fully.

Looking forward, one thing that I would like to see fixed is the multiples and controlling cores independently. AMD did a great job with this, I would love to have 4 cores enabled and leave 3 cores parked at idle and one core maxed out that I can set affinity to.

Thanks for the unlocked anniversary edition, this will bring endless joy to many more overclockers than before.

http://fugger.ipage.com/32mair.PNG

Thanks

Nice. I need one.

Good on air/water...average on LN2 :/.

Very nice. Thanks Charles for the info ! :)

Nice info!!

Thanks!!

Thx for the test. Looks like DC is rather overhyped. I'll probably stay with my good 4770 instead of wasting money for "new" gen.

Added stuff after I got some sleep.

4-stage cascade AND F1?
DJ got 6.5 on his..but..you know..it's ES

nice, seems very simillar as my 4770K from LAM. I hope, with Broadwell-K chips we will see lower coldbug or without coldbug...

Team msi removed the ihs at the Intel event. The tim and everything else looks the exact same as Haswell.

Ah.. Its new info for me...

Was the IHS removal successful?

I think that some of the MSI boards were coming with shims for the IHS removal people this time around :)

Quote:

---

Originally Posted by Splave

Team msi removed the ihs at the Intel event. The tim and everything else looks the exact same as Haswell.

---

Pretty hard to judge thermal paste by looks... But yeah, it's obviously not solder.

Yep success

@ Splave...1 in 20 ratio again or worse? :)

IDK man out of 130 ES maybe 10 5ghz 1.4v on air cooling 4c8t so far, hope they get better

Quote:

---

Originally Posted by zalbard

Pretty hard to judge thermal paste by looks... But yeah, it's obviously not solder.

---

Soldered chips will have a vent hole in the IHS, chips that aren't soldered don't have this hole. I'm guessing DC is using a ceramique-like paste instead of the normal generic stuff...better, but still not ideal.

so air / water results will be the same ?

Updated again, added air stuff.

darkage - Air and water would be much different.

i know its the English you know :-)
i was trying to ask if it will be the same results average as the last haswell revision for aircooling and watercooling because people started to hope for a 5ghz devil
sorry man sometimes it wont come out as i am thinking :-)

Once you get over the vcore increase the 4790K will end up being the faster 24/7 OC chip
5Ghz is possible on more chips with water cooling for sure.

I need one with low VCore @ 5.0Ghz on water.

your chip has to be one from that 5 GHz five chips form 130 pcs in intel tray, because in all reviews is maximum 4,7 GHz on air, and my chip is :banana::banana::banana::banana: with max around 4,6 GHz

Don't expect to get 5Ghz on air with a gimpy cooling solution.

Pushed my chip to 5Ghz for benches, dialing in memory more.

http://fugger.ipage.com/32mair.PNG

Quote:

---

Originally Posted by FUGGER

Don't expect to get 5Ghz on air with a gimpy cooling solution.

Pushed my chip to 5Ghz for benches, dialing in memory more.

http://fugger.ipage.com/32mair.PNG

---

Wow that's a lot of volts to get 5ghz. Tested mine also last time.
http://i.imgur.com/jtMHeUv.jpg

Fugger, how hot is this chip in Superpi with 1.5V?:)

I didn't have monitoring up, I would imagine close to 80c

I was stressing it a bit with 1.5v, could be done at 1.4v but I was clocking up to 5.4G on air with that at boot.

Updated again, added more defined vcore to temp steppings.

Quote:

---

Originally Posted by dhenzjhen

Wow that's a lot of volts to get 5ghz. Tested mine also last time.
http://i.imgur.com/jtMHeUv.jpg

---

Welcome to XS DJ:up:

Nice cpu!

Quote:

---

Originally Posted by Dumo

Welcome to XS DJ:up:

Nice cpu!

---

Thanks :)

Quote:

---

Originally Posted by FUGGER

Don't expect to get 5Ghz on air with a gimpy cooling solution.

Pushed my chip to 5Ghz for benches, dialing in memory more.

http://fugger.ipage.com/32mair.PNG

---

As far as I'm concerned, validations between air and water are pretty much the same +-100 MHz max.
24/7 stability, yes...32M, pretty much the same as validations, 1 core load is not going to heat things up like crazy on good air cooling / AIO

EDIT:
Big welcome to dhenzjhen, about time!

Thanks BeepBeep2, bout time to join and learn from the elite folks here :)

Did some 3d runs with my chip also

05 http://i.imgur.com/bnybnua.jpg
05 http://i.imgur.com/f6OP4qS.png
03 http://i.imgur.com/NegtH9t.png
Cine 11.5 (request from OCN guy to run cine at 5.6ghz) http://i.imgur.com/i5UrBNk.jpg

Interesting thread, thanks for all the information provided, Charles! :)

Thanks Oliver,

Welcome to XS and yes, very nice cpu. mine is unstable at 6.3Ghz

4790K ready to play :). First results at weekend with AIR (today Im working with baytrail vs Kabini review and at benchtable with i3-4330)

https://farm6.staticflickr.com/5154/...d15ca52d_c.jpg7

https://farm6.staticflickr.com/5559/...83f2c30f_c.jpg6

https://farm4.staticflickr.com/3915/...1958f3d0_c.jpg1

Nice , very helpful information. Charles if by chance you have a single stage phase unit around -35c load , can you please do a quick test for me and check max Mhz stable on any 3D benchmark ? 2 cores / 4 cores no HT, Thank you. OR Maybe someone who has tried can report this as well , maybe Allen , DJ or Michael who also play with SS and have a chip. Almost forgot , welcome to XS DJ :up: about time bro.

Quote:

---

Originally Posted by dhenzjhen

Thanks BeepBeep2, bout time to join and learn from the elite folks here :)

Did some 3d runs with my chip also

05 http://i.imgur.com/bnybnua.jpg
05 http://i.imgur.com/f6OP4qS.png
03 http://i.imgur.com/NegtH9t.png
Cine 11.5 (request from OCN guy to run cine at 5.6ghz) http://i.imgur.com/i5UrBNk.jpg

---

3 posts !!!!

Where have you been DJ !! ;)

hey Charles, This was a great read. I haven't been paying much attention to the new chips so its nice to see this. Lot's of useful information in there.
I appreciate all the time you put into this. take care

Thanks everyone, I have moved the 4790K over to water cooling for the next few weeks.

I was able to run the 5Ghz XTU, my cores reached 100c and there was a lot of throttle going on with 4c8t.

Thanks guys ^^

Looks like I may have found another candidate to go cold. 05 looks lil better compare to the first one
that did 1.395v.

32m http://i.imgur.com/8MczOKh.jpg
05 http://i.imgur.com/XP79f6q.jpg

Fantastic results there, 5G at 1.4 ish is pretty schweet for that chip. Thanks again

DJ, you need to run AIDA 64 more, you are out of control fast and should watch read/write/copy numbers for best results.

Your sub timings are off, tWR I think needs to be a 5 @ 2666, 2800 if you are watching AIDA 64 hurts performance on write and copy by >10%

Thanks ^^ but not really doing 5ghz spi32m runs, just boot and go on both 05 and 32m just to test the chip on air.

you have really GEM like allways :) My is rock stable at 1.3V only for 4,6 GHz for 4,75 GHz i need 1,4V, 5 GHz is not reachable at 1,7V ... :(
With High end Noctua air cooler

Quick tested yet, but stable seems up to 4700 MHz with 1.375 V. Hard benchmarks ending at 4800 MHz with higher voltages (1.44V). Gelid cooler+4000 RPM Delta fan :-)

Quote:

---

Originally Posted by dhenzjhen

Thanks guys ^^

Looks like I may have found another candidate to go cold. 05 looks lil better compare to the first one
that did 1.395v.

32m http://i.imgur.com/8MczOKh.jpg
05 http://i.imgur.com/XP79f6q.jpg

---

Tested last night and CB at -110 tried mucho vcore and high low VIN still didn't help :(

https://farm4.staticflickr.com/3867/...2327b359_z.jpgi7-4790K@4732MHz 4430cache Cinebench R11.5=10.36b

wprime at 4732 MHz
https://farm3.staticflickr.com/2906/...9d78bf18_z.jpgi7-4790K@4732MHz 4430cache wprime1024M=156,344s

cooling air cooler Gelid Tranquillo+Delta 4000 rpm. I cant get more (maybe +0.1MHz BCLK) :-/. I hoped for 4800 MHz, but it crashed again and again. I tried steps from 1.35V-1.45V for Vcore and combination of vccin from 1.75 to 1.97V, tried PLX and still nothing. Seems average 4790K at air. Charles has better, I believe :)

PC Per managed 4500mhz at 1.2volts and that's it. Most people (maybe not around here but in general) including myself aren't comfortable going above ~1.2V for 24/7 usage. Normal voltage for the 22nm process is what, ~1.05V? Temps might be lower but it doesn't reflect in better overclocking. Mr Fugger has a golden sample in his hand, is my guess...

looks like...There is my chip: http://www.xtremesystems.org/forums/...-Devils-Canyon

here is 5.4 on air ASRock z97 OCF :P

http://i.imgur.com/FB1SZ8A.jpg

wtf? voltage?

Quote:

---

Originally Posted by FlanK3r

wtf? voltage?

---

ancient Chinese secret

Hi Splave,

What is the maximum CPU frequency you can achieve running 3DMark06 at 4/8 on air with that board.

Please let me know, I am thinking about pre-ordering this board on the Egg.

Thanks!

5.1ghz XTU :) on air
http://i.imgur.com/r9FQt3s.jpg

1.55 volts, wow...

That is simply sweet Splave, nice chip. Seems the IMC are much better then many other chips. Assrock boards certainly help the situation as well.

I asked intel about the IMC, they said no change.

good to see you shredding it up fugger! enjoying the thread.

Nice thread Fugger!

Can't wait for the retail posts to start showing up.

BTW staying in Vegas on the strip this weekend for free thru my Hyatt card as they partner with MGM hotels :cool:

Anyone knows if the D0 stepping is for mobile only?

We got 4 ES CPUs for the team, 3 of them do 5GHZ 32M around 1.37-1.38Vcore on H80/100, one CPU has issues at 5Ghz 1.4Vcore... Seems these ESses are far better then the ones at the Computex competitions

Are they 2014 batches Leeghoofd?

I picked up a retail today, testing now.

Resetting and results in a few hours.

Thanks for the tests, interesting to see how the cold bugs happened. I had similar results.

Hello there. I have a question. I'm going to buy 4790K at NewEgg. Should I remove IHS and change thermal grease immediately after buying or should I try to get 5GHz on original one ? Will be using Corsair Hydro H100i with some hi-end thermal grease. BTW. What's 4790K IHS made of (material) ? Gonna be using some Arctic Silver 5 or maybe even Coollaboratory Liquid Pro.

Quote:

---

Originally Posted by BorniPL

Hello there. I have a question. I'm going to buy 4790K at NewEgg. Should I remove IHS and change thermal grease immediately after buying or should I try to get 5GHz on original one ? Will be using Corsair Hydro H100i with some hi-end thermal grease. BTW. What's 4790K IHS made of (material) ? Gonna be using some Arctic Silver 5 or maybe even Coollaboratory Liquid Pro.

---

Forget all about 5GHz daily stability on air or water... aim for +4.6Ghz, that would be a more realistic goal... unless you find a real gem CPU

I agree, for daily driver 4.6Ghz is better target.

Quote:

---

Originally Posted by Leeghoofd

Forget all about 5GHz daily stability on air or water... aim for +4.6Ghz, that would be a more realistic goal... unless you find a real gem CPU

---

Okay, but... 4,6GHz in normal mode, non-turbo ? ;>

I see some rich people buy 10 box finding best one and sell others.
(may be he sell them all)

You can see full scaling for G3258 L336D081 and 4790K L333C247 here.

Both scale up to ~-128 oC

Scaling on the 4690K is the same, my retail is better than the 4790K ES sample. I think the 4690K is priced a little bit too high for the reduced cache and lack of HT.

The retail 4790K I tested was a 6Ghz chip and the memory controller was far better than the CPU cores. I should have kept that chip.

The G3258 I tested was a dud, it would not break 5.7Ghz and cold bugged at -110c

Thanks Monstru

Hello all. Thought I would share my results with my 4790k. This is 24/7 OC on air.

http://i.imgur.com/xhCm7ZZ.png

http://imgur.com/0xHPmfI.png

Little disappointed. But at least I was able to keep the 4.5GHz I was running on my 2600K.

My pleasure Charles. I Hope it helps people better understand DC and G3258

4800mhz_1.30V/cpu + 4600mhz_1.31v/cache (1.95V vccin under load)

Super stable, aida64 stability test 7hours.
occt:cpu_large data set = 1hour stable
occt:linpack = 1hour stable
Cinebench loop,xtu bench, aida64 benches, pc and 3d marks..etc..
All this with 70-85C max(delided cpu). No crashes/freezes whatsoever...

Now, the problem is 4900Mhz. Cant get it stable even with 1.4V. Thats 0.1V increase but only 100mhz gain..
Tried vccin 2.02 under load and didnt help.... Can run aida64 stability test for hours, can do XTU for hours too, can do cinebench loops, wprime 1024M etc... But in occt:cpu it crashes to bsod in 5 mins.

Any ideas ?

That's funny, the first thing I thought was...if this is Fugger I bet he is using a binned review chip...But you came right out and let us know that, "This chip is binned for review sample". Hahaha, kudos to you Fugger.