EVGA 750i FTW Overclock Thread

[d3v0]

Had to reduce again, 1540FSB :/

[PizzaMan]

Originally Posted by d3v0

Had to reduce again, 1540FSB :/

Have you scanned your GTL lanes yet?

[d3v0]

I read that thread, how does one "scan" a GTL lane?

All I get are lockups, never a core failure.

[PizzaMan]

By scan, I mean to move up a notch at a time and do sterss tests until they fail. You have to start with completely stable settings first though.

The lockups sound more like the board then the CPU though. What SPP voltage are you giving it?

[d3v0]

SPP right now is 1.5v.

As far as stable settings are concerned...what do you mean? If its stable, then why mess with the GTL? Do you mean stock? Or do you mean starting with all lanes at +00mv at my current overclock and stress testing? the amount of combinations is sickening.

[PizzaMan]

You want to start with it completely stable as high as you can get it stable. Let's say 1500QDR is as high as you can go with all lanes at 0. The idea is to raise the GTl until it fails test. Start with lane 0. I started by jumping to 100. If it passes, try 110. If it fails, try 90. Lets say for example 120 is the highest it will go and be stable. You take the mean between 0 and 120, which is 60. 60 would be what you want for lane 0. Then move to lane 1,2,and finally 3. Yes, it can take a while, but after which you should be able to run the chip with lower vFSB and normally makes everything easier to stablize.

[kurosu]

Nvm.

[d3v0]

Originally Posted by PizzaMan

You want to start with it completely stable as high as you can get it stable. Let's say 1500QDR is as high as you can go with all lanes at 0. The idea is to raise the GTl until it fails test. Start with lane 0. I started by jumping to 100. If it passes, try 110. If it fails, try 90. Lets say for example 120 is the highest it will go and be stable. You take the mean between 0 and 120, which is 60. 60 would be what you want for lane 0. Then move to lane 1,2,and finally 3. Yes, it can take a while, but after which you should be able to run the chip with lower vFSB and normally makes everything easier to stablize.

So I move GTL 1 back to zero when starting test GTL 2 then correct?

I will be starting at 1450 FSB. After the GTLs are tweaked, I will try again for 1600

Also, are you making a 1hr OCCT linpack your "stability" test? I cannot fathom testing each at 8+ hrs P95 hehe

[PizzaMan]

Originally Posted by d3v0

So I move GTL 1 back to zero when starting test GTL 2 then correct?

I will be starting at 1450 FSB. After the GTLs are tweaked, I will try again for 1600

Also, are you making a 1hr OCCT linpack your "stability" test? I cannot fathom testing each at 8+ hrs P95 hehe

I would leave lane 0 at its new setting not 0. Yes, I normally run shorter tests for the GTL tweaking. When I did mine I used 10 loops of IBT, which takes about 10 mins. This was before I figured out that OCCT seemed to be more sensitive to the error that is being caused. I'm not saying OCCT will be the best for you. You will have to do your own testing for that.

[motqalden]

my q6600 is still running happily @ 3.4 ghz.
vcore 1.4875 bios
1.35 nb (gt lanes mostly +80)
1.5 sb
1750fsb @ x8 multi
linked and synced ram @ 850mhz 5-5-5-12


I recently added 2 more sticks of gskill ddr2 1066 rated. to go with my gskill ddr2 1000's.

oc still super stable with all slots populated. thank you evga!

3.6 stable.... @ 1.55 vtt I won't go their. for 24/7. but I am happy. max core temps 65 after 30 runs IBT.@3.4