• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Hardest Overclock ever!!!

T

Tweakin

Platinum Member
It's taken me over 6 weeks to complete an OC on Hasewell, the longest and hardest OC I have ever had to do!

Setting the vcore limit and clocking the frequency up was the easy part. The real work came when trying to OC the cache while running an XMP profile. I found that the XMP would consistently produce better throughput and with a lower latency than any setting I could configure, but it came at a cost. The cost being the XMP profile requires additional vcore and IMC to be stable.

My chip is a dog. At 1.3 and change, I can get into windows at 4.4Ghz, but no stability. Depressing to say the least. 4.3Ghz was doable but not with any memory timings above 11-11-11 @ 1600 or running the cache north of 3.5Ghz with stability.

After countless hours of adjusting VRIN, VID, System Agent, memory, cache....the list goes on and on and is distinguished, I have finally accomplished what I was after...a 24/7/365 (to be seen) OC with great thermals, low noise and decent throughput and stability. I have just shut down AIDA64 after running 24/7 in the background for the last 3 days while I performed my normal email, surfing, music, office, etc.

This was a battle.
pprn.jpg
 
Last edited:
I know how you feel my 3570k is terrible for overclocking, think I would need at least 1.4v to get 4.5ghz, wouldn't mind to try it but with stress testing it would go to 105c and throttling I think.
 
If you haven't considered it yet you might try delidding. It's not going to turn a dog into a thoroughbred, but it'll unlock a good chunk of potential. Ivy Bridge especially is more thermally limited than voltage limited so dropping the temps unlocks new territory.
 
If I thought I could get something worthwhile out of the chip, I would. At 1.35 I was still only hitting 70's thermally, it just wouldn't pass any tests.
 
Loved the 2500K (4.7GHz) but gave it to my younger brother.
Replaced with a 3570K, which was a little "meh" OC-wise (4.4GHz).

Maybe it was my expectations after waiting so long to upgrade my home i7 920, but reaching "only" 4.55GHz on the 4770K was "meh" to me (only equivalent to a ~5.7GHz Nehalem).
 
I took me almost 6 months to get my i5 2500K 24/7 stable at 4.5GHz. My BIOS has so many overclocking features like I've never seen before. It's probably more now, eh?
 
FalseChristian said:
I took me almost 6 months to get my i5 2500K 24/7 stable at 4.5GHz. My BIOS has so many overclocking features like I've never seen before. It's probably more now, eh?
Click to expand...

" . . probably more now . . . " Sure. But some of the same features have been reorganized and renamed. For instance -- and I speak from ignorance so no epithets or curses please -- there is an "adaptive" overclocking item in these Haswell motherboards. But "adaptive" may mean the same thing as "extra voltage for Turbo mode" -- an item in my Z68 BIOS that I use extensively in place of "offset voltage" to get higher vCORE for higher clock speeds.

Generally, and now speaking from real experience, with some of that experience you'd like to get from 0 to 60 (or stock to overclock) in a month's testing and tweaking before adding all your regular software to the system. I don't like building a new system and OC'ing while I use it for "regular bidnis," so I keep the old one running and only sell, hand down or "reallocate" it once I know the new one is tip-top ship-shape.

On the other hand, with every new leap of technology (including "extra" BIOS settings), there is still another learning curve. Sometimes stress-testing misses something. I had (likely past-tense now) an instability problem that only occurred every week or ten-days, sometimes with a lull of one month. So figure "tip-top and ship-shape" could take longer. And it will also take longer as a new motherboard will have a flurry of BIOS revisions, and new features if you choose to use them (i.e., ISRT ssd-caching when the Z68 boards first appeared), will mean more tweaking and testing, more possible causes of instability just for immature drivers.

This is my second "go-around" for tuning my (old) i7-2600K, and I discovered a few things. Some BIOS features designed to enhance stability at higher clocks also add to operating temperatures. If you can achieve a stable clock at your best minimum voltage settings without "PLL Overvoltage" or "CPU current capability" >= 110%, you don't need those features. LLC has restrained, measured and limited application, but vDROOP is also a "useful thing."

And at least for the SB-K and IB-K cores, the default "PLL Voltage" setting is too high for good overclocks in the higher range. It can probably be trimmed by 0.15V for any clock setting, which means reduced temperatures. Some reviews and guides that appear reasonably authoritative suggest there is a "sweet spot" for PLL Voltage between 1.65 and 1.68V. But for some reason, the default of ~1.8V is higher than it needs to be.
 
Nicely said sir. for stress testing measures, I use prime95 then 3dmark vantage for 3 loops, plus gaming for hours will make me conclude that it is stable. I only had problems with my lynnfield when the clockspeeds that I want to achieve needs higher voltages and proper ram multipliers.

anyhow, it is showing signs of aging as I cannot run my memory at 2140mhz as I think the imc is already taking its toll so I operated at lower frequencies with my ram just to be able to get it to run smoothly with no BSODs. planning to get a haswell-E or the devil bla bla that intel is shitting about.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top