Hardware Canucks

Hardware Canucks (https://www.hardwarecanucks.com/forum/)
-   CPU's and Motherboards (https://www.hardwarecanucks.com/forum/cpus-motherboards/)
-   -   Haswell owners and overclocks (https://www.hardwarecanucks.com/forum/cpus-motherboards/61634-haswell-owners-overclocks.html)

10e June 11, 2013 09:12 AM

Haswell owners and overclocks
 
Post your results here.

My results so far have not been satisfactory.

1.29 volts vCore (manually set) and 4.4 Ghz with some stability, though not enough to even run the Intel LinPack binaries with AVX for more than a few seconds without throttling.

My chip's an L311B203 batch, and while users elsewhere have said it's a good batch, but my experience is different, and not very impressive.

Here are links to my results/posts so far:

http://www.hardwarecanucks.com/forum/714174-post43.html

http://www.hardwarecanucks.com/forum/714197-post45.html

http://www.hardwarecanucks.com/forum/714323-post50.html

I just started a new thread to see what others are getting. I just can't believe that this chip is so voltage and power hungry.

Eldonko June 11, 2013 10:27 AM

4.9 on both a 4670K and 4770K ES C0 chips that I was told were average at best. High end H20 cooling. Going to pick up a retail shortly and see how it compares.

Sushi Warrior June 11, 2013 10:51 AM

Quote:

Originally Posted by Eldonko (Post 714347)
4.9 on both a 4670K and 4770K ES C0 chips that I was told were average at best. High end H20 cooling. Going to pick up a retail shortly and see how it compares.

I heard that a few OEM's (Asus in particular) were saying that ES samples did far better than retail chips. 4.7+ on ES, yet most retail maxed out at 4.2-4.4.

Eldonko June 11, 2013 11:33 AM

That's what some people have been claiming but Im not sure on it myself as Ive seen retails clock the same as the ES chips I have. They are both C0 in the end and I doubt there is that much difference. That's why Im going to go and buy my own retail chip to try for myself.

Spblue June 11, 2013 02:32 PM

Well, I've been testing my new system for six hours now, and bottom line, I'm disappointed. I can get 4.6Ghz stable, but that's with a high-end water loop and a very high vcore of 1.35v.

With this setup, testing with AIDA, it has run stable for 60 minutes so far... The temps are around 75C, but they can climb enough to trigger a 3% throttle for a second, once every 2 minutes or so.

At 4.8Ghz, the system will barely boot into the BIOS, and at 4.7 I can get into Windows but it isn't stable.

Ring frequency is 4.3Ghz, 1.5v

The CPU won't even boot with a strap of 166, and I didn't notice any improvement at 125. I keep hoping that I'll find some options that will give me a better overclock but so far I haven't found anything.

Eldonko June 11, 2013 02:37 PM

4.6 isnt exactly that bad.. Especially for 6 hours of testing. Ive put in at least 40 hours of testing into this board and chip alone.

Spblue June 11, 2013 02:42 PM

Do you have any suggestion as far as things to try? I'm really bad at manual work so delidding is not an option for me (just changing my Heatkiller waterbock from 1366 to 1150 scared me).

I've disabled Hyperthreading and all the low power features, meaning that the CPU always runs at 4.6Ghz and 1.35v even when it's doing nothing. I figure I can always re-enable those options later but it makes for an easier time finding the best overclock...

I'm just not sure what other setting I can try.

Eldonko June 11, 2013 02:51 PM

Reduce mem speed and uncore, find max cpu clock that has reasonable temps. Reduce cpu speed by around 50Mhz and drop voltage as much as possible. Work up mem clocks until you find max stable, repeat with uncore.

I would also use adaptive voltage instead of manual as it saves on temps and power when not fully loaded.

crowTrobot June 11, 2013 03:57 PM

Retail 4770K, 4.5GHz @ 1.265V on Z87 Pro with AIDA stability test. I haven't tested if it is stable with memory over 1866MHz yet, I couldn't get 4.4 stable with the same chip on the MSI G45 board if the memory wasn't on default JEDEC. Dissappointing but I love the low power consumption though.

10e June 12, 2013 07:10 AM

Quote:

Originally Posted by Eldonko (Post 714367)
That's what some people have been claiming but Im not sure on it myself as Ive seen retails clock the same as the ES chips I have. They are both C0 in the end and I doubt there is that much difference. That's why Im going to go and buy my own retail chip to try for myself.

That seems to be the case. My L311B203 sample was not stable at 4.4Ghz at 1.29 volts. After folding@home for 10 hours or so it hard crashed with the famous BSOD 124.

I re-tried those LinX and IBT binaries at stock. They are the problem (for me) as they die at stock settings. They don't crash but they fail within two minutes with an error, so I'm back to using Prime95 for stability testing.

4.5Ghz is a no-go for my sample. At 1.45 the CPU is still not stable even with cache ratio/un-core set at 39. At that point there is almost no reason to go this high on the core when the un-core is negating this extra 100mhz, so I'm back to testing at 4.4Ghz and 39-41 un-core to see how this plays out. The 1.45v vCore setting starts to get my de-lidded unit to the high 80s low 90s and that's beyond my comfort zone.

My CPU's maximum vcore at stock is a rather high 1.25 volts. Strange. My opinion at this time is that it's a hot, hungry retail sample.

Off the top of my head I don't remember my 3770K being this high at stock, as I think a .25 positive vCore off-set resulted in 1.38 - 1.40 max on that chip.

I'm really interested to see how your retail sample goes. Like you said in the other thread, I probably should have just re-sold the chip, but I'm pretty hard-headed that way, as I don't like giving others my problem heh.

I can still use it, just not at a very high speed


All times are GMT -7. The time now is 12:26 PM.