The Phenom II List of Overclocks

[justapost]

ok then, i'll go for a 920 and try hardcore htt oc on a sb600 board to add to the discussion xD

2.8g NB is sufficient for 4ghz with the 920 without lowering the nb multi...

Abit hilarious in germany only one onlineshop has only the 920 available for preorder. I'll wait till I get an 940 in a store, have quite a few 15min's away.

:missing chip eating smiley on sofa:

[KeZzZu]

Looks like Sweden + Finland geting 0850BPMW in first shiping...if i´m lucky i will get mine on friday.

Btw nice NB clocks

OT /

Nice to see you mate, i wish that id have s939 mobo + cpu, so i could try to beat your bh-5 record :P my kit does 292 with 3.9v and highes i reached was 294 cpu-z stable... was on air cooling and +7c ambient temp if i remember right...

well soon back to work to get some money and new cpu & mobo, so i can throw this E8500 out of window.

[HondaGuy]

And how much is the NB clock

@ HondaGuy: Am I right this is already under water? Whats highest clock then you can achieve??

Oese: Yes your right , its under water..So far I have been able to hit 3.9 CPU so far, still in the testing mode

Amdcian: NB clocks were at 2600Mhz

Titan7171: I find running higher NB with a lower HT I get better results. Look at post #86 the first two test were with the HT@2600 same as NB.. So I droped it back down to stock and I received a better score by 392 points

Thanks Guys, hopefully you all get your Phenom II chips soon....Like to see what they can do on a M3A79-T Deluxe

[Oese]

thx...

[Oldguy932]

List has been updated.

[HondaGuy]

List has been updated.

thanks for the update, just checking it out,, my Memory is 4x1Gb

Everest stock with NB@2800- Ram Ganged


Everest CPU@3.7 NB@2800- Ram Ganged


Everest CPU@3.762 NB@2926-Ram Ganged

[Oese]

these are real numbers

[Oldguy932]

Thanks for consolidating all your screenshots. Its weird that you L3 latency went back up after making the nb faster from 2808 to 2926.

[Sgt.McRuff]

Im so wanting those NB speeds! I cant get mine stable at 2450.
All you did was add 1.45 volts to your north bridge? or is there some other settings you changed?

Edit: Also is NB water cooled? and what are its temps?

[HondaGuy]

NB is just stock cooling, Yes all I did was changed the Volts in the BIOS for the NB.......Phenom II 940 seem to Undervolt pretty good.

[Brother Esau]

NB is just stock cooling, Yes all I did was changed the Volts in the BIOS for the NB.......Phenom II 940 seem to Undervolt pretty good.

NICE

[TL1000S]

Wow!
That was great news/test HondaGuy!
PHII looking better and better...

[Oldguy932]

macci posted a pic of 3.3ghz nb while running a TRUE in dinos thread about ln2 needing a long time to evaporate. He had benchmarks with it but all the numbers were blotted out.

[Boschwanza]

Honda Guy

Can you make a quick run Sisoft Memory Bandwidth with 3.7 GHz and 2.9 GHZ NB and if you have a bit more time check results with rightmark memory analyzer using this settings

[HondaGuy]

Sisoft Memory Bandwidth

[jonspd]

ASUS M3A32-MVP seems to be ocing the cpu really nicely....

[HondaGuy]

WPrime:



WPrime CPU @3.814

[Caveman787]

I swore having your memory ganged was better for bandwidth scores and overall.

Why would he run sisoft unganged then?

Good score though.

[crazydiamond]

i know its supposed to offer the highest bandwidth

[Lightman]

I swore having your memory ganged was better for bandwidth scores and overall.

Why would he run sisoft unganged then?

Good score though.

UnGanged gives best performance in Everest Memory Copy test as well as in most real world apps. SiSoft is measuring NUMA bandwidth so each core is running this test simultaneously and gives best results in unganged mode as well, but ganged is not far off. Everest test memory only on single core ....

PS. In 3D section SF3D, Macci and Sampsa are showing some nice things. 3.4GHz NB anyone?? 2GHz HT under LN2 ...

[Caveman787]

The 3.2 nb was done under ln2? or air?

[Jaivan]

WPrime CPU @3.814

Thats odd I though you'd at least hit sub 10s with that clock speed.
I got 10.480 with my 9950 @ 3570Mhz.
Kinda disappointing .

[kazgirl]

Thats odd I though you'd at least hit sub 10s with that clock speed.
I got 10.480 with my 9950 @ 3570Mhz.
Kinda disappointing .

I think I'm right in saying though that, HondaGuy is using an SB600 M3A32-MVP mobo, so no SB750 benefits, and I think from most of the ss's that he runs with CPU Tweak disabled, which will also drop performance.

With CPU Tweak enabled however its possible that he wouldn't be able to get as high core clocks though, so its a swings and roundabouts thing, lower clocks increased core performance or higher clocks and lower core performance....

Might be a bit of a balancing act as to which one is better...????

[krjalone]

Quick tests:







Not even tweaked yet. NB is pretty low now. Watercooled.
Tweaking more when i get back home.

[rypt]

not exactly. there are no applications that currently run 100% cpu usage on every part of the cpu for at least 8 hours straight. most people can have computers that aren't even 1 hour prime stable and they have no problem with it at all. stable to me is if you can run every single application you use without ever having an error. not putting so much power through your mosfets that they might explode when in any other real situation it wouldn't happen.

Stability means that the system will not error even under extreme load. While you may never see the use that the stress tools put on the system, it is possible to get such high load (as the stress tools prove).

It says more about the *huge* amounts of current (Amps x V) that the poor vrms/Mosfets must take without melting/damaging the PCB/traces.

We do not want to burn our boards/test "board-clocks" do we?

Firstly "current" is not amps times voltage, I believe Amps x V is Power. Current is Amps.

But more importantly, if the board cannot hold the amount of stress you are putting on it, then you need to back down on the stress being put on it (by lowring the OC), simple as that.