I7 920 D0 & EVGA X58 SLI wont do 200fsb...?

[Bartman39]

It will run 20X191fsb and with turbo it has no problem running 4.0Ghz with 1.31 volts on the core but no matter what I do it will not run 20X200 with turbo disabled (will run briefly in Win7 but crashes pretty easy)...? I dropped the memory down to 1600mhz (2/8 divider) even though I was running the 2/10 divider for 1916mhz with the 20X191fsb... Yes did adjust everything in the bios to match the newer settings but it just wont do it...? Possibly this MB just wont cut it...?

or other idea`s on what to do...?

 

[Ben90]

Check Uncore? Sometimes it sticks the multiplier even when you lower the RAM divider.

Other than that, test each part individually to find out what can't handle that speed. If you can't attribute the crashes to any other part, then yes I would say its the mobo.

Also, doesn't your mobo allow a x21 multiplier on the 920? Or only when turboing?

 

[Elganja]

I don't have a i7 920 D0 but I have had similar processors and I have the same motherboard
-I had a i7 920 C0 that I ran at 19x200 w/ a 2:8 divider
-I currently have a i7 930 Do that I run at 20x200 w/ a 2:8 divider. You can see my exact setup in my sig (doesn't mean it will work for you)

So in short, the x58 SLI mobo isn't what's holding you back. You can try loosening your memory timings some or increasing your QPI or IOH voltages.

Can you post what settings you are using?

 

[Bartman39]

Ok did some more checking and testing and found...

No adjustment for a higher multiplier just Turbo mode (even tried the E-LEET Tuning utility)

Uncore is working fine... Everything is set manualy

I did find that it will run 19X200fsb (no Turbo) and even lowered the Vcore to stock voltage and kept the memory at 1600mhz but lowered the timings to 7-7-7-24 and its rock solid...? So why can I run 20X191 and with Turbo it runs 4.0Ghz (with 1.30 Vcore) and even run the memory at 1916mhz at 9-9-9-24...?

Doesnt make alot of sense...? Only voltage change was the Vcore which I kinda like even though its 200Mhz slower but man the memory kicks @zz...

 

[bryanW1995]

overclocking is a real pita sometimes. I've thought I had stable oc's several times on my i7, only find out later that I, well, didn't. have you tried adjusting other voltages besides just vcore? have you read the i7 overclocking guide?

keep in mind that just because one person has an evga mobo doing 200 qpi it doesn't mean that you will be able to.

check out the fritz chess benchmark thread. on page 4 or 5 rubycon gets into an interesting conversation about OC settings. it's helped me quite a bit. she also recommended that I run linx 0.6.4 for stability testing, that caught errors that I was missing with occt.

 

[Bartman39]

Thanks for the reply and yes have read the OC`ing guide here plus a few others which are basicly the same... I have adjusted voltages accordingly but its just one of those things I guess...? Been doing this quite a while but these new I7`s are a different animal for sure... Pretty happy with what I have now anyhow so may not even consider trying for that little bit more but have 4.0Ghz setup stored in one of the save settings section... May look a tad more into the E-LEET utility programe though...?

 

[bryanW1995]

nice OC on the truck btw. those are a little bit more expensive to replace...

 

[Shmee]

I seem to remember a lot of i7's do not like overclocking on the even multipliers as much. This is why having a locked turbo for 21x is very useful, which is available on some boards, such as my gigabyte ud5.

I even know my 980x seems to be doing better on an odd multi. Also, my old 920 D0 would do 21x200 for 4.2 with a locked turbo; I think both boards should be capable of 200 BCLK. Just check you vcore, vtt, and uncore settings.

 

[Rubycon]

Yes they definitely "prefer" odd multipliers. So called 'fsb holes/traps' exist as well.

 

[MyLeftNut]

If you're overclocking on even multipliers, adjusting all the different clock delays will help a great deal.

 

[Bartman39]

I remember hearing that somewhere so yep I opted for 19X210fsb and it does it no problem and still with 1.31 volts Vcore and might try tweaking it for less...? Still have the memory timings at 7-7-7-24 too... Oops... Was running memtest and started getting errors so 8-8-8-24 it is... Oh well...

thanks like it even better... :biggrin:


Yea the Dmax is a really expensive OC`ing rig but hey its so cool to have something small and light colored give you crap and you can leave it in a cloud of black soot... Best run has been 14.3 @96mph not bad for a 7200lb flying block (bigger than a brick)

 

[Shmee]

also, while 200 fsb shouldn't be hard for the board, 210 might be up there, and would probably need an increase in VTT, not vcore for stability.

 

[Gillbot]

Yes they definitely "prefer" odd multipliers. So called 'fsb holes/traps' exist as well.

agreed, I've seen many cpu not like an even multi or fsb, but if you jump up to the next it works just fine. I had a few C2D that did the same thing.

 

[Elganja]

Yes they definitely "prefer" odd multipliers. So called 'fsb holes/traps' exist as well.

agreed, I've seen many cpu not like an even multi or fsb, but if you jump up to the next it works just fine. I had a few C2D that did the same thing.

why is that? I found the same to be true w/ my i7 920 C0 but never understood why that was the case (although with my i7 930, 20x multi works just fine)

 

[Gillbot]

why is that? I found the same to be true w/ my i7 920 C0 but never understood why that was the case (although with my i7 930, 20x multi works just fine)

I have never seen an answer as to "why" but if you dig around you'll see loads of threads on the matter.

http://forums.anandtech.com/showpost.php?p=28636958&postcount=3
http://forums.anandtech.com/showpost.php?p=29203588&postcount=6
http://forums.anandtech.com/showthread.php?t=2062632&highlight=fsb+holes

There are loads of threads on the matter and lots of input/examples.