Intel Haswell Overclocking Clubhouse.

[DOM]

I wanna say 2.05 V? Those are some pretty tight timings. Dumo never holds back.

Howszit going with your clocking? Did you get any tips to break 6 GHz?

still stuck at 6

No haven't got any tips but seen there's been another beta bios released on the 2nd but need more ln2 XD

 

[cadaveca]

still stuck at 6

No haven't got any tips but seen there's been another beta bios released on the 2nd but need more ln2 XD

I'll leave it on my shelf for the moment then.

Keep track of hours on that chip... I wonder how durable these are. I think not very.

So, I wonder if you then kill a chip and use tuning plan warranty, will Intel give you a "tough" chip, like seemed to happen with SNB-E and IVB?

 

[DOM]

I never used that before and the ones I did kill not ocing sucked balls lol

And that's to much work to keep track

 

[radrok]

So, I wonder if you then kill a chip and use tuning plan warranty, will Intel give you a "tough" chip, like seemed to happen with SNB-E and IVB?

Makes me wanna purchase that tuning plan and run my chip at 1.55v daily

Might even survive until IVB-E who knows lol.


Are the Haswells tough chips or they're squishy?

 

[PolRoger]

So this week I've been able to hook up some of my custom water cooling parts and I'm now using a thick (60mm) triple "420" rad and Swiftech Apogee XT rev.2 waterblock.

Before when I was using a Corsair H110 I noticed some difficulty running/testing 48x and I figured that the cooling wasn't really up to the task so I didn't spend much time on it and ran mostly 45x/46x/47x.

Well now that cooling isn't really an issue... It seems that my chip may be "gimped" at 48x?!!

I can boot and do some benching but when trying to run a long term "crunching" Rosetta 100% load overclock... It always seems to result in random BSOD? It seems that adjusting the various voltages... Vcore, Ring, CPU input CPU SA, etc. never resolves the issue. I think my chip has a weak component/limit... Perhaps cache or a particular core that just won't hold up at 4.8?

Current stable 47x/44x overclock:

 

[cadaveca]

did you try to increase VCC-in?

 

[dumo]

what kind of volts are you pushing on the samsungs?

2.0V boot and final

Happy 4th!

Bios L1.32C

At least it can do some bench @ 4X4GB @ 3333

 

[PolRoger]

did you try to increase VCC-in?

Yes... I adjusted pretty much all of the major voltages... VCC-in, Cache, SA, Vcore and Dram (I did leave CPU digital/analog I/O on auto). I also adjusted/tested various cache multi down to 39x... Tried different ram kits... Hynix and Samsung as well as different dram speeds down to DDR3-1600... 4x4gb and 2x4gb configurations. Tried using 47x and raising bclk up to 4.8... No luck achieving long term stability... :shadedshu

And yet drop back to 47x or below and the chip then seems to behave and overclock in a rather normal manner?

 

[dumo]

Perhaps cache or a particular core that just won't hold up at 4.8?

Current stable 47x/44x overclock:

My good chip can only do 4.7Ghz max 4C/8T 1.35Vcore/2.0 Vccin on Thermaltake 2.0 h2o fairly stable for bench like cinebench, wprime but not prime67.
I eliminated the heat problem with -20C cooler and the chip jump easily to @ 5Ghz core/uncore 4C/8T 1.3ish vcore and imc up to 3600

 

[PolRoger]

Does anyone have insight into the how CPU SA and CPU analog/digital I/O affects higher multi overclocking and or running higher memory speeds?

So far I've pretty much left both the analog/digital I/O on auto... ASUS rules seem to default I/O at +000 for analog and +100 for digital while previously I usually just adjusted the CPU SA voltage.

I've noticed that digital I/O corresponds to the VTT sensor on HWiNFO64 (+100=1.104v idle/1.112v load). Currently trying to tweak digital I/O to see if my chips IMC will benefit from adjustments to aide in higher multi clocks and or higher memory clocks.

Slight bump on digital I/O to +115...VTT now at 1.120v idle/1.128v load:

 

[Womper]

So this week I've been able to hook up some of my custom water cooling parts and I'm now using a thick (60mm) triple "420" rad and Swiftech Apogee XT rev.2 waterblock.

Before when I was using a Corsair H110 I noticed some difficulty running/testing 48x and I figured that the cooling wasn't really up to the task so I didn't spend much time on it and ran mostly 45x/46x/47x.

Well now that cooling isn't really an issue... It seems that my chip may be "gimped" at 48x?!!

I can boot and do some benching but when trying to run a long term "crunching" Rosetta 100% load overclock... It always seems to result in random BSOD? It seems that adjusting the various voltages... Vcore, Ring, CPU input CPU SA, etc. never resolves the issue. I think my chip has a weak component/limit... Perhaps cache or a particular core that just won't hold up at 4.8?

Current stable 47x/44x overclock:

Same thing here with 48x (4.8GHz). I thought my chip was dying because it started crashing more often, but it turns out it's just random. I played one game for 2 hours straight, and then started a different game and froze every 20-30 minutes even as I increased core voltage and dropped cache speed each time. No improvement- I had to drop back to 47x. Interestingly, 49x and 50x are bootable (not benchmark-able) at manual 1.39v and 1.41v respectively, so it's strange that 48x cannot achieve long term stability regardless of voltage. The freezes at 48x don't throw a bsod 124 like low voltage normally does, so it must be something else.

New question- How much does cache voltage increase under load when using offset voltage? Does it jump up by 0.1v under AVX load like vcore?

 

[DOM]

anyone know what code 78 means on MSI Mpower Max

 

[MxPhenom 216]

anyone know what code 78 means on MSI Mpower Max

Is that info not in the manual? I think some manuals have all the codes and diagnostic and trouble shooting steps.

 

[DOM]

checked didn't find it

 

[cadaveca]

checked didn't find it

I get that from bad cache settings....


or BCLK pushed too far.

Not on MPowerMax though, still haven't played much with that.


You got BIOS v132?

 

[DOM]

I get that from bad cache settings....


or BCLK pushed too far.

Not on MPowerMax though, still haven't played much with that.


You got BIOS v132?

Yeah 132

But even on the 2nd bios shows 00 so think its a gonner XD

 

[MxPhenom 216]

What do you guys prefer for stability testing for Haswell? I think I'm done with Prime95. Takes so long to get any info in terms of system stability.

 

[dumo]

Yeah 132

But even on the 2nd bios shows 00 so think its a gonner XD

Killed my 1st. retail chip @ launch day

It was stuck @ 78 which means cpu half dead with pci or pch lanes burnt

 

[dumo]

Found low v chip with Thermaltake 2.0





IMC not too bad

[/URL]

 

[HammerON]

Nice dumo

 

[radrok]

Wow that's an awesome 4770k, dumo.

What do you guys prefer for stability testing for Haswell? I think I'm done with Prime95. Takes so long to get any info in terms of system stability.

I'm still waiting for a board but I'll probably use LinX AVX enabled.

It does good for SB-E and I'm sure it does good for Haswell, if it passes 90% memory with all threads for atleast 1 hour then you won't "probably" find anything that makes your system crash with lower loads.

 

[Grnfinger]

Going to pick up a MVI Hero and 4770K tomorrow and join the club. Will post some results in the next few days.

 

[cadaveca]

What do you guys prefer for stability testing for Haswell? I think I'm done with Prime95. Takes so long to get any info in terms of system stability.

I start with wPrime 1024M. It's good to point out pretty weak OC's quickly without risking any sort of OS corruption or similar issues.

Then I run AIDA64's built-in stability test, with default settings, noting voltage. This cause the voltage boost that many reviewers report, but not the high temperatures. This is a pretty good stability test for most users needs and I think accurately represents high workloads that most users will see, and is higher than most games, too(but not all).

Then I run AIDA64's stability test with only the FPU option enabled. This is a load that some users will experience(only if strictly using AVX2 to the highest possible), and causes those high jumps in power and heat that many reviewers have reported.


That's JUST for the CPU cores. I use other tests for other parts of the chip.

 

[tom_mili]

Well, I am on my journey to 5 GHz mark today. All attemps have been in vain
Also cranking up the voltage up to 1.4 V didn't solve the stability problem. Thermal could be a problem for my AIO cooler, though.
I tried the 1.25 strap with 40x but still no luck, won't even boot into windows and the black screen of overclocking failure bla bla came and wanted me to enter the bios again.
I can't even boot into window when using the 1.25 strap when everything is set auto.
I think I must be doing something wrong using the strap.

My max stable overclock is 4.8 GHz @ 1.25V. I was so happy to know I still have a pretty good chip

http://img.techpowerup.org/130708/4.8 1.25 V.png

I am sorry Cadaveca for the VRING value, it was actually 1.05V stock. I was probably drunk or sleepy or even both when writing that
It was 3 AM if I remember correctly

 

[Kursah]

Well got back from camping yesterday afternoon and slapped on my brand new Noctua U14S. Sweet sweet cooler! I love it! At 4.2GHz and 1.200v I am idling at the upper 20's and lower 30's on this cooler and stormy Monday. I ran about 10 minutes of AIDA Stress and peaked at the upper 60's to very low 70's. But spent most of the load/stress time in the short test in the 50's and 60's though which made me pretty happy. That and stopping the test, instant drop to the 30's.

I did initially try for 4.5ghz at 1.200v, it would POST and then lockup with a 124 bsod just after booting into windows. Same with 4.4. So I figured I would try 4.2, and at 1.200v it seems dead stable and super cool. So SWEET I gained an easy 700MHz OC without trying very hard. But I want to see if I can hit that 4.4-4.5 mark and make it attainable with my air cooling. If not..oh well! It's way faster than my i5-760 oc'd to 4.0 was at even stock, now with a 200MHz lead, and more efficient operation, it screams.

Oh and one more thing I noticed you guys were posting early on in this thread:

First Boot Voltage:
CPU - 1.072v
Cache - 1.031v
CPU Input: 1.792v

Those were the ones I wrote down...I can get more and reset the BIOS if ya need.

I left memory at an easy 1600 speed, I had been running XMP 2133 and it POSTed fine at 4.5 just wouldn't get into windows so that was the first thing I backed off remembering they say to do the CPU first then add the memory speed to your OC to see if it stays stable. Here's a pic of my first stable Haswell OC. Not great, but I'm still happy! I'm not looking for WR's, I'm looking for a nice smooth running gaming rig that can pull off a nice OC on air.

It starts here :