EVGA 750i FTW Overclock Thread

[d3v0]

Pizzaman, remember how we were talking about FSB straps and such?

I am starting testing now. Here is a taste:

As you can see so far (read the spreadsheet), the fsb going from 1333 to 1640 has not required a bump in voltage from the base (lowest) bios values on the FSB, NB, and SB.

Now, in an effort to fully bust your theory (no offense, this is how science works...we try to prove ourselves wrong!) I will continue until I can no longer POST at max voltages.

Further testing will entail moving each and every voltage individually once the system becomes unstable (probably 1700+) in an effort to determine which best provides stability when talking about the FSB only.

[PizzaMan]

Originally Posted by d3v0

Pizzaman, remember how we were talking about FSB straps and such?

I am starting testing now. Here is a taste:

As you can see so far (read the spreadsheet), the fsb going from 1333 to 1640 has not required a bump in voltage from the base (lowest) bios values on the FSB, NB, and SB.

Now, in an effort to fully bust your theory (no offense, this is how science works...we try to prove ourselves wrong!) I will continue until I can no longer POST at max voltages.

Further testing will entail moving each and every voltage individually once the system becomes unstable (probably 1700+) in an effort to determine which best provides stability when talking about the FSB only.

Very NICE!!!

You have already proven my 680i theory wrong. Exactly how it should be done.

The 680i had a large strap that covered the 800 and 1333 FSBs. So far you have proven that the 1333 and 1600 FSB are held in the same strap on the 750i(and most likely 780i). That or nVidia put more strap latencies in place so the NB and SB aren't needed for a while. Could you also test the area between 800 and 1333?

Very nice work!! +rep

[d3v0]

I absolutely will, but first I am shooting for 2300FSB

First instability at base voltages found @ 1800FSB. Lockup after 4 minutes with full NB load, rebooted and bumped up the FSB voltage by .05v. If it fails I will try NB, then SB, then FSB/NB, then FSB/SB, then NB/SB, then FSB/NB/SB ALL by one notch..then bump them more...

oh lawd

[PizzaMan]

Originally Posted by d3v0

I absolutely will, but first I am shooting for 2300FSB

First instability at base voltages found @ 1800FSB. Lockup after 4 minutes with full NB load, rebooted and bumped up the FSB voltage by .05v. If it fails I will try NB, then SB, then FSB/NB, then FSB/SB, then NB/SB, then bump them more...

oh lawd

If you feel like it you should do bandwidth and latency benchs in the mix. This would help point out strap changes. If you get better bandwidth/latency at a slightly lower FSB then you have prbly entered a new strap in the higher of the two FSBs. The lower FSB being of tighter latency thus giving better performance. Yet another thoery about the straps.

[d3v0]

The problem is my clock speeds are always different, as are my ram speeds. The clock speeds vary 300mhz or so, and the ram speeds vary 30mhz or so.

I think the results would be muddy at best.

[d3v0]

So far I've hit a sheer wall @ 2075FSB. I think the sweet spot is going to be 1960 though, since the problems start @ 2000.

Firefox crashed basically like clockwork at every FSB at or above 2000. I reinstalled it earlier today. IE7 ran like a charm though, even up to 2075FSB.

Attached is the final spreadsheet I ended up with. Woo, 40+ reboots!

[10acjed]

So your not needing any extra voltage to the CPU FSB til after 1600?

I may lower mine a bit since Im at 1600 right now w/ 1.4v, I will try 1.35v

[lordnoak]

He's also running a E8400 and decreasing the multiplier. Just FYI 10acjed.

[10acjed]

Yeah, first thing I thought of was the dual core part... But its worth a shot to get lower voltages. I know I am 13hr OCCT stable right now at my settings, so dropping .05v on the FSB and testing will tell me alot.

[PizzaMan]

Originally Posted by d3v0

So far I've hit a sheer wall @ 2075FSB. I think the sweet spot is going to be 1960 though, since the problems start @ 2000.

Firefox crashed basically like clockwork at every FSB at or above 2000. I reinstalled it earlier today. IE7 ran like a charm though, even up to 2075FSB.

Attached is the final spreadsheet I ended up with. Woo, 40+ reboots!

WOW!!

Well, from the looks of things. NF200 really doesn't help much until you really push the board. Wonder if you can get 2000 stable with a little NF200 voltage?