also stasio. Haswell a bit different from ivy/sandy. u dont gain much from cpu speed on bandwidth.

but when u tightened the bandwidth ure going to need more vcore.

Yea,I know......i didn't mean speed.
Here is less speed and less frequency,but Read/Write /Copy is higher then Kevin....



OK..this is with 4670K,will found with 4770K.
Btw,
Kevin posted some pic from XS and there can see also difference.


....you have super low Vcore...haha

Edit:
with 4770K@ 4.3GHz HT.......Hynix CFR RAM 2666@11....@ 2800MHz 12-14-14-35-2T @ 1.65v

compare with Kevin in post #116





This post has been edited by stasio: Jun 18 2013, 11:24 AM

Like cstkl1 said. The bandwidths can be improved by reducing the 3rd timings mainly the tRDRD and tWRWR, but the latency will be reduced as well.

The lowest tRDRD and tRWRW can boot at 3000 at 1.65v is 6, the bandwifths will shoot up to 30k. Will try to run it by tonight hopefully will yield the same performance without needing voltage increment.

nah c-state C6,C7. Got it working. Its awesome haswell. The voltages on the cpu cores most of them drop like 0.000V on idle. Lowest i have seen 1.9watts.

stasio its the tRDRD and tRWRW. his rams are stable only this setting.

But normally if he ups it he has to up his vdimm. I suspect the 3k rams being sold will end up the same. Default for 2.4k ram and above all dont run tRDRD at 4.
hence y u see my 2133C7 all 33k. It comes at a cost of vdimm and vcore.

Ure asking whats the benefit. this settings shows difference on aida on multicore bandwidth mem cache test. but on single thread test the 3k still will beat the 2133C7 on bandwidth like maxmemm.

so its a matter of balance on ure voltages .. mem speed etc. So far i think personally a 2666 C10 is the fastest from what i have i seen.

So what i am trying to say his 4.6ghz 2.8k/3kmem prime .. Ure not going to see many ppl are able to do that.

This post has been edited by cstkl1: Jun 18 2013, 10:06 AM

GIGABYTE Tweak Launcher 2.0 B13.0608.1



Btw,
memory timing locked ,by Intel (same like Asus Mem TweakIt 2.00.01)

Just got my 4770k. will most likely start test on thursday.
owik, your 4.7ghz chip is msia or costa?

Glad that you have made a good choice.
My 4770K is Malaysian batch L312B460.
Let's torturing.

*Evil music coming* I have the same batch as you. good god.

Different chip different VID and so for OCbility. It's all about draw of luck.

btw, compression performance on Haswell is awesome. Now you can cut the time of unrarring by half.



This post has been edited by owikh84: Jun 18 2013, 11:26 PM

alright looks like we have a version of LinX working for Haswell on win7 SP1 (not Win8)

Credit to acebmxer @ XS

OMG, bad news guys.
LinX 0.6.4 (10.3.7) is actually stressing more than I thought.
Now 4.6GHz @ 1.44v will hit TJmax of 100C instantly.
Now i have to reduce overclock to 4.4GHz @ 1.375v in order to avoid from overheating.
So, from now onwards I will not use Prime95 27.9 for stability check anymore because it is not ready for Haswell thus not as stressful as LinX.

This post has been edited by owikh84: Jun 19 2013, 06:52 AM

Yea,I working with him last 2-3 days......

Btw,
Latest MSI ,AsRock.......BIOS:

http://forum.hwbot.org/showthread.php?t=75024

This post has been edited by stasio: Jun 19 2013, 01:39 PM

Thank you stasio.

4770K @ 4.5GHz 1.3v | ram 3000MHz 12-14-14-25-1T

Gigabyte Z87/Haswell Overclocking(OC) Guide

http://www.overclock.net/t/1401976/the-gig...ocking-oc-guide

That's on water?

As mentioned previous, on my air-cooled Venomous X, prime95 (non-AVX) @ 4.2Ghz @1.1v was already touching 70-75'c. On prime95 (AVX) was 80-82'c. I can't imagine if I start to use LinX!

Now I'm beginning to think whether there is a stress program that can test for stability without the temperature going through the roof. Can I say that since prime95 (non-AVX) @ 4.2Ghz @ 1.1v was stable for 3 hours, my rig would be "relatively" stable 99.99% of the time?

yupe thats under water.
Prime95 needs some updates for Haswell. Normally I would trust LinX or IBT only since it tell us the GFlops which shows the efficiency and accuracy of our overclocks. Eg.

4.5GHz @ 1.25v - 125 GFlops, 85C
4.5GHz @ 1.3v - 135 GFlops, 90C

The higher the GFlops, the more performance the overclock yield. So, 1.3v is picked as our preferred vCore.

That was on win7 SP1 with AVX instruction AND LinX with AVX libraries build 10.3.7.

On win7 non SP1 OR LinX non AVX, you would get:

4.5GHz @ 1.15v - 60 GFlops, 60C

lower GFlops and reduced performance of AVX supported apps and might sometimes get BSOD on routine works.

Thats why I said the hotter or the more vCore needed, the more stressful the app. Also, the more vdroop (lower load vCore) the more stressful.

Remind me of tRD aka performance level on lga775
One of the most important parameter that determine your bandwidth and voltages
Test of ram chip/pcb quality too

Owi,
change Latest BIOS......to:
http://forum.hwbot.org/showthread.php?t=75024

Massman just update OP.

Wow! 85'c on water for 4.5Ghz @ 1.25v

Ok, I will try Linx this Friday when I'm on leave ( ). Thx for the clear explanation.

Trying to get some info on AVX and gaming but looks like not much info at the moment. Aikes, there's even AVX2 probably to make my proc roast more!!!

I think that LinX 10.3.7 is kinda old as far as I could recall on 3770K + Z77 it needed less vCore than 11.0.1.005 but yield the same Gflops.

Hint is on the top pane of LinX.

AVX edition: less vCore needed
Simply linpack: more vCore

but generates the same GFlops.

so there is no perfect app to stress Haswell yet atm

Anyway, to choose whatever stress app is really up to you.

I even saw an OCN forummer advised fellows not to use AVX based stability tool because it doesnt represent daily routine work lol.

Another forumer from XS told me that if you could finish the Fritz Chess benchmark then your overclock should run fine on any game without stability issue.