• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Haswell Overclock Thread

Yes I have.
In the classic mode the straps work fine, I can boot to 4.75GHZ with ease at 1.35v, probably even maintain stability.

I'll end up leaving it at 4.6GHZ at 1.25v though.
 
My stability is either a RAM or board problem, I'm 1066MHZ CL9 to maintain stability with my RAM :p
So I can do proper testing when it's sorted.

I await the day I can run 1600MHZ RAM with my Haswell set up.
 
It's just some rather generic Excelram 1600MHZ.
But I think it's either my board's a DOA for RAM, or a RAM incompatibility, switching the RAM out Saturday.
 
Excellent. Is that stable at 4.6GHz! Still not sure what the best stress testing method is with Haswell?

I do 30 min LinPack test in OCCT with 'AVX Capable Linpack' enabled, if it passes that then i move on to AIDA64 and run that for 3 hours, if it passes that i call that a stable OC
 
Well 4.7Ghz is now stable @ 1.260v (using CPU Offset).....

isN3XYf7OhcwO.jpg


inGk7PoVzwoYy.jpg
The way im looking at it is 1.260v will be fine for a 24/7 OC as im using CPU Offset so i doubt that there is going to EVER be a app that i use that is going to load the CPU to 100% and cause the CPU to draw the full 1.260 volts it does when stress testing, am i right in thinking that?.
OCCT seems to literally thrash the CPU to an inch of its life, whilst ADIA64 doesnt seem to be so stressful on the CPU, this is shown in the different max temps that the 2 programs give after a stress session??

4.8Ghz on the other hand, got up to 1.36v and OCCT blue screened in the first 20 seconds. It seemed that the closer i got to 1.36v the quicker OCCT failed, at 1.30v OCCT lasted longer!. Is there any reason for this?
 
I haven't had much chance to play around since transfering my 4670K over to a 350D and H100i last night. I did try the Asus test of changing the multiplier to 46 and vcore to 1.25v, everything else except memory is on auto/default . That boots into Windows 7 and is useable but bsods if I run Prime95 blend. Temps are about 26c idle and max I saw was 62c under load (H100i stock paste and fans).

Hopefully I will have more time to play around with it tonight.
 
Got my Asus Maximus Hero yesterday and finally installed my 4770K.

After doing some testing today I have decided that I will stick with a 4.7GHz @ 1.25v overclock for my 24/7 use until I fancy tweaking it and adding more volts to go further. Cinebench and Handbrake temps are around 70c's and games don't even hit 60c's. Only AIDA 64 manged to go mid 80C's, but not like I will ever be running anything that will do that :)

Below are some results in pictures of what I managed to do with it so far at at 4.8GHz and even a SuperPi test at 4.9GHz.

There seems to be big increase in temps from 4.7GHz to 4.8GHz and I think it would be better to delid before going further. Though I don't plan to delid until next year this time, which is when I will either delid this cpu, or my new broadwell cpu.

I will make a sig for my spec soon btw.

superpi48ghz.png



cinebench48ghz.png


4900mhz125v.png


hyperpi48ghz.png


aida6448ghz30mins.png
 
My chip uses 1.26V for 3.9GHz if you leave it on Auto, no wonder people are seeing high temps, quite happily runs at this speed at 1.06V it seems if you set it to manual.
 
Hmm Mine is running 4.5 @ 1.25 volts but seems to be hitting 77c. It idles around 30c.

I am running my first custom water cooling loop so was expecting better temps. I wonder if I need to re-seat my CPU cooler as my GTX 470s are running at 30c idle and 45c load which is a lot more acceptable.

I did the Asus test on my Gigabyte OC board, booted into Windows but crashed on Prime. So 4.6 might be possible.
 
I haven't tested stability with other programs yet, but I've spent some time getting a lower voltage established for my 4.6 clock (I started at 1.32V). This might give me some room to increase the clock to 4.7, maybe 4.8 ... the graph in the image clearly shows the gradual drop in temperatures as I lower the voltage by ~0.1V between each 5 minute CPU Stress Test. This screenshot was taken just before it finally crashed in the next test. :p

I had to reset my ram to auto settings - the tighter timings caused the XTU CPU Stress test to crash (without failure) before I could even start this process. Once I'm done checking these settings for stability I'll work on the ram timings again.

46ghzfindingvoltage2.png
 
Just got a newsletter with OcUK's overclocked 4770K bundles. Seems there is only 1 build that has 4.7GHz, the rest are mainly 4.4GHz.

Does seem I got lucky with my chip :D
 
Back
Top Bottom