Overclocking i7 930

[Atomic_Sheep]

Try this.

CPU Multiplier: 19*200bclk OR 20*190bclk
vcore 1.34v
qpi/vtt 1.31v
dram 1.5 and 999 24 auto timings
leave the rest on auto

If that stable...then try to reduce vcore bit by bit.
When you get to know more...then start to play with other settings.
What CPU cooler do you have?..... i7-930 runs pretty hot above 4GHz.
What is your goal for this overclock?

Ok, will do.

My aim is as high as possible at reasonable temps. By reasonable I would prefer to have the CPU running at below 70 in FSX during summer (I'll probably have to downclock about 200mhz from whatever I get it up to right now for summer but we'll see). It's currently running at 60 or so, so I'm sure I can have it set to around 4.2 or similar, but unfortunately I don't want to run Prime95 at 4.2 @ >75c temps for long periods of time to test stability so I don't know if I'll be able to go that high. I'll be decently happy with around 4, @4.2, I'd be ecstatic.

Cooler is Megahalems Rev B.

UPDATE: Ran Prime95 for 1 hour and 15 minutes @ 1.215v QPI/VVT, 1.25v vcore & 200x19 and no crashes so that one off bluescreen was possibly just that... a one off.

I then started 200x20 at the setting recommended by SamSaveMax i.e. 1.34v for vcore (mine was slightly above 1.34 as there wasn't a 1.34v dead) and 1.31v for qpi/vvt (once again... slightly above as 1.31 exact didn't exist)... seemed to run fine although at 76 degrees which was unpleasant. Opened my window and switched the heater off to get the temps below my comfort zone of 75 and as close to 70 as possible... ran tests for around 20-30 mins at those settings then reduced both by 1 notch and repeated until I got down to 1.325v vcore and 1.255 qpi/vvt. This was stable in Prime95 for 40 odd minutes. Then reduced vcore 1 notch as I had already tried 1.325 and qpi/vvt at 1 notch above the sweet spot of 1.215v @3.8Ghz but crashed BSOD'd after 20 mins literally seconds before I was about to turn it off and reduce vcores once again. So looks like those are my final overclock settings. I don't think there's a point going to 4.2 as the temps are already in the 70s... low 70s with an open window and mid 70s with the heater on. I don't much want to go to above 80. So voila... I'm happy for now. Will do a longer Prime95 test at these settings as well as a memtest, which I haven't done up to now... but not tonight.

 

[MIAHALLEN]

BTW - those temps are very low....80C is nothing to worry about

 

[Atomic_Sheep]

Personally anything above 75 for long periods of time is out of my comfort zone although I have had it up to 85/86 for a short period of time out of pure curiosity before but it's not an area I want o visit again any time soon.

 

[MIAHALLEN]

That's fine....I was just letting you know it's safe.

 

[Atomic_Sheep]

Yerp, I understand.

 

[Awptex]

Wouldnt go past 60

 

[Atomic_Sheep]

AAAAH... got some BIOS error saying the system has experienced some blah blah due to over clocking (first time since the over clock)... reverting to original setting. Didn't save my over clocked profile!!! Tried over clocking again and have totally forgotten what half the stuff meant and although I wrote down most of the settings, I can't remember what else I tweaked. Major frustration. It's possible to save the settings once I do an over clock right?

 

[Atomic_Sheep]

Hi guys. I just reinstalled my system including a new updated BIOS and the new F6 BIOS for my mobo looks totally different to the old F5 one. I think they mainly rearranged the menu items but they may have included new options but I don't know... anyway... Basically, I input the exact same parameters that I used to have on my old overclock but the computer doesn't load windows, just gets to the DMI page and just before booting it resets and then it deteriorates and resets sooner with each restart and then it stops booting, then resets to last good settings etc etc. The one setting that I haven't been able to set manually because i don't know where it is anymore is the 24 in the 9 9 9 24 for the memory timings. I'm not sure whether this minor thing could be causing the problem but yer, I don't quite know what to do now. I switched turbo off... or at least I think I did... its got some extreme memory profile [X.M.P.], not sure whether it had it before but its disabled, erm yer... I don't even know what to say, as I said, I've got the exact same settings as I did before but it just doesn't boot. Any ideas? Where should I start troubleshooting the issue?

EDIT: looking at this thread with the screenshot that I attached earlier, looks like XMP was in the F5 BIOS so yer... was disabled before, is disabled now, so def not the issue... no idea what it does either anyway... just pointing that out for no reason in particular.

 

[Atomic_Sheep]

Well, I'm more familiar with the new BIOS layout, so I've found some flaws in my previous attempts at getting my old overclock working. Unfortunately, nothing seems to work, I'm almost 100% positive now that my settings are exactly the same as in the old BIOS but I just still keep getting reboots at the DMI page all the time. I tried reducing the multiplier to x19 so 3.8Ghz but kept all the voltages the same as for the 4Ghz overclock, but still getting the exact same behaviour. How could a different BIOS effect things so much?

 

[Draizien]

It would be helpful if you could post your BIOS settings.

 

[nzaneb]

What cooler are you using?
200 blck x 20 multi
Vcore 1.325V
VTT 1.325V
LLC enabled
DRAM manually set timings, speed, and voltage to spec
Disable EIST
Leave all other voltages on auto

This should be very close to what you need for 4ghz

 

[Atomic_Sheep]

Well as I said, I've got settings that used to work on the old BIOS.

QPI clock ratio - x36
bclock - 200
mem multiplier - x8

Intel turbo boost - disabled
performance enhance - standard
hyperthreading - enabled

memory timings:
cas latency 9
trcd 9
trp 9
tras 24
command rate 2

(same settings for channels a b and c)

vcore - 1.325
qpi/vtt - 1.255
ioh - auto
dram voltage 1.5

(memory voltages and timings are set to spec)

LLC - not sure what this is (c1 state?) - ive got 2 options... the first one is C1 which is what I have enabled at the moment. Can't remember what I had it as on my last overclock to be honest, but I don't think I screwed around with it too much.

the second option is c3, c5 or whatever the other ones are but they are diabled.

all other voltages and all other settings are set to auto.

Not sure what EIST is (will check it out).

My cooler is the same Megahalems Prolimatech that I've been using.

EDIT: just tried bumping up vcore and qpi/vtt... 1.33125 and 1.315 respectively and also reduced multiplier to 19, at elast it started booting up windows but would just die during the boot up but now with a BSOD. I then turned off hyper threading, turned c1 state off, still the same result. Do processors deteriorate like this over time? One day stable running 4Ghz and then bam, the next day can't even boot windows with a different BIOS and the same old overclock?

 

[nzaneb]

LLC = Load Line Calibration (you should have an option in the BIOS to enable it)
EIST = Enhanced Intel Speedstep Technology (aka Speedstep)
You can disable c-states as well

Turn the QPI clock ratio back to auto. I'm guessing that is Gigabyte's way of saying Uncore and that they're not actually referencing the QPI Link. Regardless, at a 36 multiplier, you're trying to push 7.2 ghz through the Uncore at 200 bclk. Unless Gigabyte has some sort of formula that then cuts that in half, you're twice as high as you should be. Just set it to auto for now, and see if that is your problem.

Chips will definitely deteriorate, but it will not be an instantaneous degradation, unless you're doing something extreme. At the voltages you have listed, and with a properly installed decent air cooler, you will have upgraded long before it starts to degrade... if at all.

 

[Atomic_Sheep]

Ok, thanks for the advice, I'll give it a go after my RAID volumes get rubilt for the upteenth time in 2 weeks... RAID 0 might not have any redundancy but by golly, I never had any issues with them... RAID 0+1 is total rubbish from where I'm sitting... almost daily rebuilds and RAID deteriorations... but that's just me thinking out loud.

 

[Atomic_Sheep]

Ok, well I tried EIST off, I tried setting QPI to auto, I also reduced the CPU Clock Drive to 700mv from 800 and PCI Express Clock Drive from 900 to 700 as they were when I first got the BIOS but for some reason they went up when I started overclocking... I never touched them... anyway... the results... even worse... started to crash even sooner and even more intermittently. I then increased the CPU Clock Drive and PCI Express Clock Drives to 800 and 900 respectively and predictably, still got the same result. I tried setting memory timings to auto, didn't like that much either... restarted pretty much immediately and proceeded to restart until stock standard settings were loaded but yer, I'm pretty sure it likes the x36 multi. Haven't looked into LLC yet. I think there's different options, I had a quick look but didn't change anything, can't remember what different settings are there but yer I'll check it out.

 

[nzaneb]

I'm sorry, but you have a setting "extremely" wrong, if it will not even boot at the voltages I listed. Please post pictures of your BIOS overclocking menu and submenu's (Including Voltages and Memory settings)
We will get this figured out. Looking down in a previous post, I see a picture of your BIOS, and it does list the QPI as QPI Link. Anything less than an Extreme Edition chip, or a Xeon, runs at 4.8 QPI. Provide the pictures and we'll fix it

Edit: Did you load defaults before you tried the setting I suggested?

 

[MIAHALLEN]

Sounds like memory problems....test with memtest86+ and see

 

[grizzlybluntz]

u can do 4ghz on air, i achieved it stable with a CM geminII cpu cooler at 40/65 idle/load.

Just switched to CoolIt Vantage system, 38/55 idle load at 4.20 ghz stable with validation
http://valid.canardpc.com/show_oc.php?id=1508573

 

[Atomic_Sheep]

Ok thanks guys, I will post some pictures up as soon as I can. On a different note, before updating the BIOS, I didn't revert it back to default settings, someone else on a different forum suggested that this must always be done, could this be an issue and why?

In relation to memory, should I run a memtest when overclocked or stock? I ran a memtest on stock speeds when I first got the comp with 6Gb because the system was intermittently not reading the full 6 gigs but it came up with no errors, the feel of the system is such that the memory is working as it did before but with different BIOS updates, the issues have become accounted for. If you are suggesting I run a memtest after overclocking then I don't think it will work simply because it is very likely to restart a short time after I start the mem test if I get that far in the boot process in the first place.

 

[Atomic_Sheep]

Well, to cut a long story short, I RMA'd my MB after it wouldn't post and I ended up running stock settings all these years. But I've always had the itch to finish what I started and now that I'm trying to extend the life of my computer, I decided it's time.

EDIT: Disregard everything below since I realised my uncore was less than 2ce the memory frequency. It's all good now, the process continues.

So here's where I'm at in my current attempt:

I finished by bclock overclock... got it to 200 @ 1.175v VTT. Ambient temps in the room during OCCT tests were around 25c and the max temp spiked to 74 but was averaging 70. Anyway, now I'm up to the spot where I think everything went pear shaped last time. Memory.

I changed the timings to 9 9 9 24 40 as per CPU-Z XMP and as per specs of my memory sticks. I put DDR voltage to 1.5v. It boots if the multiplier is at 6, however the moment I change the multiplier to 8 to get it up to 1600, it doesn't boot. Do I now start cranking up the voltage until it boots/until I reach the VTT to DRAM delta of 0.5v? I can't find any info related to this.