790GX + Phenom 9850 = 3.4ghz on air

[Sylvanas]

Click

AMD has been touting it's SB750 as able to increase existing OC's by 200mhz at least over the current SB600+790FX boards. Looking good- this is also the chipset with supposedly onboard video aswell (in addition to 2 Crossfire slots 8x8 2.0 that support hybrid crossfire with the mobo). I would think we are in for a surprise on the OC front once Deneb hits in conjunction with the 790GX.

 

[error8]

Wow, nice to see that. I hope it's a general situation and not just one particular chip with that mobo. Let's hope AMD brings those overclocks higher with Deneb, so that the Intel monopoly brakes down.

 

[SlowSpyder]

Can anyone explain to me why the south bridge makes such a difference? I have a 9850BE, so I should just be able to adjust the multiplier while I keep the HT at a set speed... I guess I'm just missing why that's important for the overclock. Or do the boards with the SB750 have better power output or something?

 

[nyker96]

good to know AMD is finally (albert slowly) catching up in speed.

 

[heyheybooboo]

Originally posted by: SlowSpyder
Can anyone explain to me why the south bridge makes such a difference? I have a 9850BE, so I should just be able to adjust the multiplier while I keep the HT at a set speed... I guess I'm just missing why that's important for the overclock. Or do the boards with the SB750 have better power output or something?

It's a PLL - or Phase Lock Loop - issue. It primarily comes down to 'phase' timing. All system clocks need to function in phase regardless of clock: the cpu clock, the HTT/NB clock, PCIe clock, PCI clock, USB clock, etc.

Remember the old black and white sci-fi movies with the high-tech ocilliscopes in their labs? Think of those 'waves' ~~~~~~~

When you overclock, the starting- and end-points of those waves need to be in phase to maintain system stability. When those waves go out of phase the variance is called 'jitter'.

In order to maintain phase in an OC when you 'hit the wall' you bumps up the volts to reduce jitter. The MSI 790fx for example has a voltage adjustment for the sb600 PLL.

Pin assignments on a cpu include clock input-output. I imagine when OCing with the sb600 it just gets to a point where those clock signals are so far out of phase stability becomes very difficult.

With the 'new' AMD 700 series chipset mobos there are just limits when OCing. The phase timing of the sb600 (and to a lesser extent - the sb700) appears to be one of the primary issues. SB750 is a big step forward to addressing them.

 

[heyheybooboo]

oops!

 

[heyheybooboo]

ouch!

 

[byronm]

3rd times a charm!

 

[heyheybooboo]

Originally posted by: byronm
3rd times a charm!

Deja vu all over again ...
Deja vu all over again ...
Deja vu all over again ...

 

[aigomorla]

sorry...

not going to believe it until i see OCCT or Prime95.

 

[SlowSpyder]

Originally posted by: aigomorla
sorry...

not going to believe it until i see OCCT or Prime95.

You'd never consider any of my oc's stable then. I run P95 on all the cores for about a hour, then I play games on it. If it doesn't crash while I'm gaming I'm happy. What do people consider 'the' test for stability, P95 for 4 hours? 8? 24? I never do anything like that, just curious what people do.

 

[aigomorla]

Originally posted by: SlowSpyder

Originally posted by: aigomorla
sorry...

not going to believe it until i see OCCT or Prime95.

You'd never consider any of my oc's stable then. I run P95 on all the cores for about a hour, then I play games on it. If it doesn't crash while I'm gaming I'm happy. What do people consider 'the' test for stability, P95 for 4 hours? 8? 24? I never do anything like that, just curious what people do.

Well, i only trust things that i knows works.

Just cuz i say i dont believ it doesnt means its not possible. It just means i havent seen anything spectacular to make me say WOW or tell my friends that this is the truth.

 

[Schadenfroh]

What is the ETA on this thing?

 

[Sylvanas]

Originally posted by: Schadenfroh
What is the ETA on this thing?

Link

It seems there are no firm dates atm but from a few quick Google searches, it could be in the next month or so. Many manufacturers had boards for show at Computex so I would imagine we are not far off and that it's just a matter of perfecting BIOS revisions before shipping them out.

 

[zach0624]

wow, pretty impressive for phenom and 3.4 is great with any quad and they do this with most chips I can see phenom actually being a choice for a future build (well, probably deneb) especially with such great crossfire platforms. Maybe deneb will actually be good enough to get nehalem out the door for cheap by next spring. Nvidia certainly left the door open for AMD/ATI with the lacluster performance of the gt200 in the video card market.

 

[Lorne]

Sylvanas, How did you get 13X multi, I also have the K9A2 Plat but 12.5x is as high as it goes, Is that rev 2 MB or something to do with the "RD"790FX.