Overclocking P965 chipsets and unlocking their potential

[RJ]

I have to give props to SuperFly03 to start with. I was having error issues with my memory, and he turned me on to this article. I thought it should be here in OC'ing as well.

http://www.thetechrepository.com/showthread.php?t=30

I probably wouldn't have believed it possible if I hadn't seen it myself. Apparently this twist isn't "Asus only" as I've has similar success with my Abit board.

9x405 mhz...1620mhz data rate
7x405 mhz...2084mhz data rate

Dropping your multiplier increases your FSB. Go figure. 8O

 

[BUFF]

. Apparently this twist isn't "Asus only" as I've has similar success with my Abit board.

9x405 mhz...1620mhz data rate
7x405 mhz...2084mhz data rate

Dropping your multiplier increases your FSB. Go figure. 8O

very odd because the AB9 doesn't switch to the 1333 strap like the Asus :?
must read further when I have the chance.

 

[RJ]

. Apparently this twist isn't "Asus only" as I've has similar success with my Abit board.

9x405 mhz...1620mhz data rate
7x405 mhz...2084mhz data rate

Dropping your multiplier increases your FSB. Go figure. 8O

very odd because the AB9 doesn't switch to the 1333 strap like the Asus :?
must read further when I have the chance.

It doesn't say anything about a 1600 strap either, but apparently there seems to be a "smoking gun" that there is one. I had minor memory error issues @ 400 fsb, I switched to 405, and the issue went away. I suspect I engage the next fsb strap, which loosens the timing latencies in the NB. Something to keep in mind that I had to do is bump the CPU voltage to what appears to be a 4.05 ghz setting. It wouldn't run on the same voltage that 9x400 was okay on.

 

[BUFF]

. It doesn't say anything about a 1600 strap either, but apparently there seems to be a "smoking gun" that there is one. I had minor memory error issues @ 400 fsb, I switched to 405, and the issue went away. I suspect I engage the next fsb strap, which loosens the timing latencies in the NB.

400fsb is where most 965s normally kickover to the 1333 strap but the AB9 series doesn't use it. Sounds like something else (latency?) is changing though.

 

[RJ]

. It doesn't say anything about a 1600 strap either, but apparently there seems to be a "smoking gun" that there is one. I had minor memory error issues @ 400 fsb, I switched to 405, and the issue went away. I suspect I engage the next fsb strap, which loosens the timing latencies in the NB.

400fsb is where most 965s normally kickover to the 1333 strap but the AB9 series doesn't use it. Sounds like something else (latency?) is changing though.

This is from the article I link. It's near the bottom...

Clock generators recieve BSEL signals from the CPU so they set the correct FSB for the CPU its self, from the above table (compiled from various whitepapers and clockgen specs) we see 333(1333) and 400(1600) are listed. Looking at how the Asus P5B is performing could Asus just be switching in 400strap at 400fsb with a latency tweak? I am beginning to think they may be and we are testing to see if the 400 Strap actually works.

So how do you know that the AB9 has no 1333 strap? Above 400 mhz, you're engaging a 1600 strap. So how am I at 450 if there supposedly isn't even a 1333 strap? Why would I stabilize at 405 after mem errors @ 400?

Just cuz you can't see it don't mean it ain't there! 8)

 

[BUFF]

So how do you know that the AB9 has no 1333 strap? Above 400 mhz, you're engaging a 1600 strap. So how am I at 450 if there supposedly isn't even a 1333 strap? Why would I stabilize at 405 after mem errors @ 400?

Just cuz you can't see it don't mean it ain't there! 8)

because better, more knowlegeable people than me say that it doesn't.
Performance testing against comparable mobos apparently also shows the difference ~400 - where their performance drops as they switch straps supposedly that of the AB9 doesn't.
Also my QuadGT has manually selectable straps (don't know if the standard/Pro does) but 1066 is the highest.

 

[chrone]

will abit engineering raise the fsb strap even further to 1333mhz or 1600mhz in later BIOS update?

or should we wait for the new revision of the board which supports it?

this kind of thing really being a pain in the $ss

no more waiting please.. hehehe

 

[BUFF]

will abit engineering raise the fsb strap even further to 1333mhz or 1600mhz in later BIOS update?

I believe that they've been asked to look at it.
Thing is if it does 520+ fsb on the 1067 why do you need the looser, poorer performing 1333?
Even an E6300 will do over 3.6GHz with 520+ fsb & that's where the bulk of C2Ds top out anyway without significant cooling etc..

If abit got the hardware side right (& it looks like they did) then there won't be another board revision unless Intel change something again like the VRM requirements for another CPU as they did with C2D.

 

[RJ]

will abit engineering raise the fsb strap even further to 1333mhz or 1600mhz in later BIOS update?

or should we wait for the new revision of the board which supports it?

this kind of thing really being a pain in the $ss

no more waiting please.. hehehe

Considering it's already sitting there, I'd say it's only going to be a BIOS release to "unveil" the 1333 fsb strap. Considering I'm sitting on 2024 mhz data rate right now, 1333 would be a breeze.

 

[RJ]

So how do you know that the AB9 has no 1333 strap? Above 400 mhz, you're engaging a 1600 strap. So how am I at 450 if there supposedly isn't even a 1333 strap? Why would I stabilize at 405 after mem errors @ 400?

Just cuz you can't see it don't mean it ain't there! 8)

because better, more knowlegeable people than me say that it doesn't.
Performance testing against comparable mobos apparently also shows the difference ~400 - where their performance drops as they switch straps supposedly that of the AB9 doesn't.
Also my QuadGT has manually selectable straps (don't know if the standard/Pro does) but 1066 is the highest.
So there's no forcing a strap? How is that in the link that it shows on a 771, that there is in fact, a 400 strap as it shows the BSEL signals for it?
If there's no 1333/1600 strap, how is my NB fsb higher than most x975 chipsets can achieve? I haven't gone for mid to high 500s yet because there's no way my memory would handle that.

 

[chrone]

thanks for the info

 

[BUFF]

So there's no forcing a strap? How is that in the link that it shows on a 771, that there is in fact, a 400 strap as it shows the BSEL signals for it?

we aren't using 771s though & just because something may exist doesn't mean that it's actually utilised.

If there's no 1333/1600 strap, how is my NB fsb higher than most x975 chipsets can achieve? I haven't gone for mid to high 500s yet because there's no way my memory would handle that.

975X can do that though - I'm told that virtually all P5 WS will do 450-470 plus out of the box & I've seen screenies of unmodded AW9Ds up there too with vmodded over 500.
& of course 965 is a newer chipset so hopefully would benefit from past experience & improve upon existing designs (even though it's slower clock for clock than 975X :roll.

 

[RJ]

we aren't using 771s though & just because something may exist doesn't mean that it's actually utilised.

From the article...

From 2cpu.com this post by XeonTux we see a gem of info from a clockgen white paper, its for 771 CPU's but we all know they are REAL closely related

hmmmmm..... 8O

975X can do that though - I'm told that virtually all P5 WS will do 450-470 plus out of the box & I've seen screenies of unmodded AW9Ds up there too with vmodded over 500.
& of course 965 is a newer chipset so hopefully would benefit from past experience & improve upon existing designs (even though it's slower clock for clock than 975X :roll.

pffft! :?

So I guess ya missed the part about a 480 fsb setting actually achieving 640 mhz in the NB fsb? Did ya even READ the article?

 

[BUFF]

we aren't using 771s though & just because something may exist doesn't mean that it's actually utilised.

From the article...

From 2cpu.com this post by XeonTux we see a gem of info from a clockgen white paper, its for 771 CPU's but we all know they are REAL closely related

hmmmmm..... 8O
I can theorise that 2+2=5 but that doesn't make it true ...
& it still doesn't counter the "even though it may exist it may not be utilised argument"
We know that the AB9 & P5B are doing something differently because they perform differently (although not being a chipset or BIOS engineer I have to rely on others doing the work). I've been told by 2 separate, well known & unassociated journalists/reviewers etc. that the AB9 doesn't appear to use the 1333.

975X can do that though - I'm told that virtually all P5 WS will do 450-470 plus out of the box & I've seen screenies of unmodded AW9Ds up there too with vmodded over 500.
& of course 965 is a newer chipset so hopefully would benefit from past experience & improve upon existing designs (even though it's slower clock for clock than 975X :roll.

pffft! :?

So I guess ya missed the part about a 480 fsb setting actually achieving 640 mhz in the NB fsb? Did ya even READ the article?
Yes & I also read it back in October& November too along with Tony's original strap article.
& fsb on it's own isn't the be all & end all - you can look at Tony's analysis of the DFI RD600 mobo where even though the board will do over 500 fsb he says that there's no real point in running it over ~470 because performance starts to drop as the memory performance goes to pot & he advocates using CPU's with a larger multi to max. core clock whilst staying within the ~470 envelope.