**Official** GIGABYTE Z87X-OC/OC Force Club

[Shiranui Gen-An]

Ah, you guys have the 2600C10. That explains things a little, though my 2400C9 have always seemed a lot weaker than others I've seen on the forum.

[stasio]

Have one set Team 2666C10 for sale .......grab it !

http://www.xtremesystems.org/forums/...ix-and-Samsung

[Shiranui Gen-An]

At least my 2400C9 can do 2800MHz 11-13-13-30-1T like your 2600C10. This is with all four sticks.

[stasio]

At least my 2400C9 can do 2800MHz 11-13-13-30-1T like your 2600C10. This is with all four sticks.

Yes.....nice one.
You need less Vcore that me.
I'll try with 4 stick tmr.

[stasio]

GIGABYTE Tweak Launcher 2.0 B13.0711.1.......

http://download.gigabyte.us/FileList...er_8series.exe

[Blackwhite]

Right guys, need an advise.
My 4670k can do 4.5 GHz with 1.25v - not bad, but THB with you I have only set voltage and multiplier, that's it.
Should I play with PWM or LLC settings, and switch off C3 and C6 for better Overclocking?
How do you overclock haswell?

[vailr]

A before-purchase question: can anyone provide a screen capture of their Gigabyte Z87 board that confirms the presence of the C2 Haswell chipset? Because CPU-Z only shows: "Chipset: Rev. 06" in several screen captures shown in this thread.
Reference:
http://www.legitreviews.com/news/15374/
http://www.ocaholic.ch/modules/news/...p?storyid=6578
These web pages indicate that the C2 stepping was only released July 15.
Can't find any update showing that Intel may have altered their timetable and that Gigabyte could have maybe sold at retail the C2 stepping boards already. Or not?

.....................
Edit: Southbridge - Lynx Point - Rev.04 = C1 stepping? 05 = C2 stepping?
.....................
Update: the latest CPU-Z v. 165.1 has the correct C1 nomenclature; v. 164.0 & older show this instead as: 04.

[AVulgarPicture]

My OC says Rev 06, Rev C1.

[sin0822]

There could also b some C2s already out there, if anything it would be quiet. Intel confirmed at Compute that the chipset revision doesn't actually fixx the USB 3.0 issue, microsoft hotfixes have done this, GBT has them on the download site for windows 8.

[Alex-Ro]

My OC says Rev 06, Rev C1.

Rev 06 stands for cpu revision,southbridge is either C1 or C2

[munim]

Does anyone else's board increase vcore by itself under load? Here's how I overclocked:
1. Set manual, static vcore and static multiplier (turbo, eist, c states all disabled).
2. Overclock and find the processor is stable at 4.3 GHz with 1.195 to 1.200 vcore.
3. Enable all the stuff like turbo, eist, c states, set turbo multiplier to 43, regular ratio to stock 35. Vcore set to normal (this would be 1.165) and set voltage offset to +0.035 to give 1.2 voltage under load.
4. Boot, see that no load turbo is correct at 1.2volts (for a minute after booting windows turbo is engaged while stuff is loading I think but load is not heavy). Then start OCCT and now under load the voltage is 1.297!

Anyone know what I have to do so that voltage stays at 1.2? I've tried toggling VRIN load line calibration but that does nothing (tried Auto, Turbo, Extreme).

I'm running with all power saving / turbo features on so that the processor can downclock and undervolt when not under load. This entails setting vcore to 'normal' and setting the voltage offset. This is all fine but then under load the vcore increases a further .9 to .1 voltages. I figured out the voltage I needed by running at static clocks and static manual vcore

[munim]

Does anyone else's board increase vcore by itself under load? Here's how I overclocked:
1. Set manual, static vcore and static multiplier (turbo, eist, c states all disabled).
2. Overclock and find the processor is stable at 4.3 GHz with 1.195 to 1.200 vcore.
3. Enable all the stuff like turbo, eist, c states, set turbo multiplier to 43, regular ratio to stock 35. Vcore set to normal (this would be 1.165) and set voltage offset to +0.035 to give 1.2 voltage under load.
4. Boot, see that no load turbo is correct at 1.2volts (for a minute after booting windows turbo is engaged while stuff is loading I think but load is not heavy). Then start OCCT and now under load the voltage is 1.297!

Anyone know what I have to do so that voltage stays at 1.2? I've tried toggling VRIN load line calibration but that does nothing (tried Auto, Turbo, Extreme).

I'm running with all power saving / turbo features on so that the processor can downclock and undervolt when not under load. This entails setting vcore to 'normal' and setting the voltage offset. This is all fine but then under load the vcore increases a further .9 to .1 voltages. I figured out the voltage I needed by running at static clocks and static manual vcore

Upon further research this appears to be by design. OCCT uses AVX and the processor is supposed to increase vcore by approx. .1v for that. Cool!

[sin0822]

yes if the benchmark you use has AVX then the CPu will apply more volts, some close to 0.1v some less, depends on the CPU. But normally you don't really run many applications which use AVX, do you? I didn't knwo there were many other than benchmarks.

[munim]

yes if the benchmark you use has AVX then the CPu will apply more volts, some close to 0.1v some less, depends on the CPU. But normally you don't really run many applications which use AVX, do you? I didn't knwo there were many other than benchmarks.

Right I'm just using the processor for gaming, absolutely nothing else. It's interesting then, though, to think of what a stable overclock is. Should stability be determined by AVX enabled stress tests or something else?

[sin0822]

yea I kind of agree it is kind of stupid. IMO test stability with your game, sometimes you'd be surprised the different types of stress games put on CPUs rather than stress testers, especially CPu intensive games like BF3.

[vailr]

Intel confirmed at Compute that the chipset revision doesn't actually fixx the USB 3.0 issue, microsoft hotfixes have done this, GBT has them on the download site for windows 8.

Source for the "Intel confirmed at Compute" information?
That doesn't make any sense (to me) at all. Why would they (Intel) go to all that trouble of re-engineering their Haswell chipset, if in the end it actually makes no real difference to the end user.
And: what about other operating systems? such as: OSX (Hackintosh), Ubuntu, Windows XP, etc. Either USB 3.0 devices work as they're supposed to, and in any chosen operating system, or they don't.

[plane501]

Hi,
playing around with Team 2666C11, not definitive settings.

2933 12 15 14 34 1T @ 1,656v
(tWR 26 in BIOS)

CPU 4500MHz
Uncore : 4000MHz

I haven't tighten all the timings, still learning how to tweak memory and i am looking for 24/7 stable OC.

Advices are welcome

Team 2666C11(8Gx2 Hynix MFR)@ 2800 11-13-13-30 1T 1.70v
core 4.5G ring 4.5G
super pi 32m 6m50.499

[TRD]

Nice.

[sin0822]

Source for the "Intel confirmed at Compute" information?
That doesn't make any sense (to me) at all. Why would they (Intel) go to all that trouble of re-engineering their Haswell chipset, if in the end it actually makes no real difference to the end user.
And: what about other operating systems? such as: OSX (Hackintosh), Ubuntu, Windows XP, etc. Either USB 3.0 devices work as they're supposed to, and in any chosen operating system, or they don't.

First of all when Intel introduced USB 3.0 with Z77 they exempted Vista and XP from having driver support, so there is no wy to get USb 3.0 from Intel working in those OSes(you can disable xHCI support and get them into USb 2.0 mode and then it will work), thank Intel for that.
With Z87 they wanted to drop all XP and Vista support, however you can install ME and INtel INF drivers on XP with the win7 version.

Sorry i meant, Intel sources confirmed at Computex 2013 that the C2 revision doesn't really fix it, this issue has been around a long time, just media figure they need news and bring it up before launch, check some of those windows 8 hotfixes(windows 8 has a built in xHACI USB 3.0 driver), and see if you see anyone complain about this USB 3.0 issues. Which is specifically about certain USB 3.0 devices needing to be unplugged and plugged back in after S3 sleep.
I mean have you seen Intel officially comment on the issue?

BTW check it out, cookie benching some samsungs at some crazy speeds and volts all on air:

[TRD]

How do you tweak RTL and IO-L ?

I have values like 46 and 48 to set for the RTL, but the cursor is -15 to + 15

IO-L values like 4 and 5, choice between -3 to +3



I'm lost.

[owikh84]

yea I kind of agree it is kind of stupid. IMO test stability with your game, sometimes you'd be surprised the different types of stress games put on CPUs rather than stress testers, especially CPu intensive games like BF3.

stress tester apps like LinX/IBT/Prime95 work well for CPU stability, provided you use the right version with AVX enabled.
games stresses more GPU than CPU.

* removed some stupid remarks

[sin0822]

How do you tweak RTL and IO-L ?

I have values like 46 and 48 to set for the RTL, but the cursor is -15 to + 15

IO-L values like 4 and 5, choice between -3 to +3



I'm lost.

Yea I asked the BIOS engineers to make these fixed values, and they have done that but the BIOS is not available yet, still in validation processes. I understand your frustration with that(because the RTL and IOL values change with divider/xmp opposed to stock values shown in gray), but the offsets are offset of the grayed out value that is in the column closest to the timing selection. So before you set your OC it might say like 38 for the RTL CHB D1/R1 or somthing like that and you set offset of +5 it wont be 43 on restart if you set a divider highre than that you are currently at. So let's say you set 28.00x and RTL gray value goes to 48, then +5 woudl actually be 53, even if you set this +5 before the grayed value was 38, for that reason/confusion in newer BIOSes RTL and IOL values ill be fixed

Also the newer BIOS that comes out will fi things like vDIMM working to max vdd selectable(aka over 2.1v easily), as cookie showed in that spi run above.

stress tester apps like LinX/IBT/Prime95 work well for CPU stability, provided you use the right version with AVX enabled.
games stresses more GPU than CPU.

* removed some stupid remarks

yea of course, but some games like BF3 stress the CPU cache pretty hard as well and like on Ivy bridge, while you might pas 12 hours of prime95 BF3 might BSOD.

[TRD]

Yea I asked the BIOS engineers to make these fixed values, and they have done that but the BIOS is not available yet, still in validation processes. I understand your frustration with that(because the RTL and IOL values change with divider/xmp opposed to stock values shown in gray), but the offsets are offset of the grayed out value that is in the column closest to the timing selection. So before you set your OC it might say like 38 for the RTL CHB D1/R1 or somthing like that and you set offset of +5 it wont be 43 on restart if you set a divider highre than that you are currently at. So let's say you set 28.00x and RTL gray value goes to 48, then +5 woudl actually be 53, even if you set this +5 before the grayed value was 38, for that reason/confusion in newer BIOSes RTL and IOL values ill be fixed

Also the newer BIOS that comes out will fi things like vDIMM working to max vdd selectable(aka over 2.1v easily), as cookie showed in that spi run above.

Great

Beyond that confusion ; i have noticed, variations of values in particular for the RTL CHA A D1 and IO-L CHA A D1

It fluctuates with clear CMOS, and memory trainings despite all rest of my settings are untouched.


I have spotted the first one sometimes is 52 or 51
and the second one 7 or 6

I have to double check which offset value i have to set depending of the memory training.

But soon, it will belong to the past just need a newer BIOS ^^

[nfszocker]

G.Skill PI @ 1270MHz 8-12-8-28-1T-96



G.Skill TridentX @ 1333MHz CL9

[plane501]

Team 8Gx2 2666c11@2800 11-13-13-30 1T tRDRD@tWRWR=4