Haswell Overclock Thread

Thumpsta · 8 Jul 2013 at 20:47

If it's any help mine does 4.8 at 1.34v and 4.7 at 1.31v.
Yours does appear to be pretty bad. Good luck next time around.

teppic · 8 Jul 2013 at 20:54

killoc said:
U still using adaptive mode for overclocking? U should be using manual for now.

Personally interest, whats the batch number for ur chip? L3**B*** and made in?

It was crashing on manual at under 1.36v. It wasn't just when I ran a stress test, it seemed almost random.

Batch L312B326, made in Malaysia.

killoc · 8 Jul 2013 at 21:05

teppic said:
It was crashing on manual at under 1.36v. It wasn't just when I ran a stress test, it seemed almost random.

Batch L312B326, made in Malaysia.

You got a "rare" crap chip. Hopefully ur next one is a lot better. U got ETA?

teppic · 8 Jul 2013 at 21:15

killoc said:
You got a "rare" crap chip. Hopefully ur next one is a lot better. U got ETA?

It'll be next day delivery once I put the return request in.

I read on another site that a quick test for these is fix the voltage at 1.25v, everything else default, and try to boot at 4.6GHz. If it boots, you've got an average or better chip. Mine BSODs the second it tries to boot.

edit: I'm trying this to see where it's stable. So far:

4.6GHz - cannot boot
4.5GHz - boots, instant BSOD at any load
4.4GHz - BSOD after approx. 1 min of load
4.3GHz - initially stable, but hardware error reports
4.2GHz - seems stable

killoc · 8 Jul 2013 at 21:25

using the 3 step overclocking guide haswell? thats a good guide for beginners.

Best I got is 4.8ghz 1.20v booted and bench for 15mins b4 I shut down myself cause of heat. Still havent reach 5.0ghz (1.26v max), maybe try this weekend.

teppic · 8 Jul 2013 at 21:31

It's running AIDA64 at 4.3GHz at the moment, it's been going for 4-5 mins without crashing. It looks like this one isn't happy above 4.3GHz without very high voltage.

killoc · 8 Jul 2013 at 21:35

teppic said:
It's running AIDA64 at 4.3GHz at the moment, it's been going for 4-5 mins without crashing. It looks like this one isn't happy above 4.3GHz without very high voltage.

Yeah, just swap it, and double check to see if there any missing/bent pins in socket when u remove cpu. I'm done for the day. Your new chip should be better then what u gettin right now.

teppic · 12 Jul 2013 at 21:30

Got my replacement. I'm testing on my backup system (mini-ITX) at the moment so I can't compare fully.

So far it looks a bit better, still not above average though. I couldn't boot at 4.6GHz / 1.25v, but I've booted up at 4.6GHz / 1.30v.

killoc · 12 Jul 2013 at 21:53

teppic said:
Got my replacement. I'm testing on my backup system (mini-ITX) at the moment so I can't compare fully.

So far it looks a bit better, still not above average though. I couldn't boot at 4.6GHz / 1.25v, but I've booted up at 4.6GHz / 1.30v.

Bad batch from amazon? Or totally unlucky?

You going to hit 90c plus during stress test. Best keep below 1.25v on air.

teppic · 12 Jul 2013 at 22:03

It's a different batch number on this one. It could be because this is a mini-ITX board and won't be as good at overclocking as my main one, I'll try with that one tomorrow hopefully.

Flibbles · 12 Jul 2013 at 22:43

Just done my Haswell build with a 4670k - got her running at 4.6ghz with 1.25v. As I understand I've got an "above average" chip going by one source...

I'm currently running the Small FFT test in Prime with temps maxing out at an average of 75c - do you reckon I can get any more out of her? Running a Megahalems with 2x Apache fans. Tops out at 70c in Blend.

killoc · 12 Jul 2013 at 23:15

I usually use aida64 or occt with avx ticked in linpack cause prime95 dont have haswell new instruction mention else where.

If you planning to use adaptive mode fir c6 c7. Best to keep voktage around 1.15v cause adaptive can add additional voltage of up to 0.1v.

8 Pack · 12 Jul 2013 at 23:27

Guys on ASUS boards try using Turbo Vcore. This is a very light on system resources version of Auisuite and works great.

Flibbles · 13 Jul 2013 at 00:08

Currently using Aida to test at 1.22v, 4.6ghz. Only ten mins in so...

So pleased I haven't got a duffer this time compared to my i7 920 which just refused to go above 3.8, even with HT off...

killoc · 13 Jul 2013 at 01:22

My bro i7 930 refuse to go above 3.8ghz with asrock. Np with gigabyte g1 at 4ghz and lower voltage

teppic · 13 Jul 2013 at 01:23

I'm hoping for something similar with my Asus Gene. If this one still won't do 4.6GHz at <1.3v I'm going to be really disappointed.

petedread · 13 Jul 2013 at 12:19

Hi. This is my first post here. I've been following this thread for about a week though. I was hoping I could get some advice about OC'ing my 4770K, it's on a G1 sniper 5.
So far the best I can get stable in Aida 64 is 4.4ghz @2.25v. That's with 100.00mhz bus and 4100 uncore, HT enabled and pretty much everything else on auto.
What could I try, to get 4.5ghz on similar volts? At the moment my pc crashes about 2 mins in to Aida at 4.5.
Also, at 4.4, 2.25v my temps are not to bad with Aida running. They spike high, 88c but mostly around 74c.

teppic · 13 Jul 2013 at 12:30

2.25v? It'd explode. Do you mean 1.25v?

petedread · 13 Jul 2013 at 13:04

Indeed

lol. I mean 1.25v

teppic · 13 Jul 2013 at 13:19

That sounds a bit less scary

Stress testing at mid 70s is fine really. It'll be tricky to get it higher without increasing the voltage. You could try 1.26v and see if it's stable, it may just need a slight nudge.

Try increasing the cache/uncore frequency to 4.2-4.3GHz, and the cache voltage to 1.15v. You'll also want a high LLC setting (this depends on the board) as otherwise the voltage will fall at high load.