Asus M3A78-T + 9950BE overclocking... wow its all new again

[Jimmer411]

So far Im really liking the Phenom.



Anyone else have experience with ganged and unganged and its (possible?) effect on ram stability? So far unganged is far and above more stable than ganged.

[merkk1]

Hi i run ganged all the time my hightes posted front FBS was 245 with my DDR-2 800 A-DATA 5-5-5-18 2t ram was running about 980 . I use a 3.003 ghz. mulit x13 FBS 231 ram running at 924 from my 24/7 clock . My 9850 be not as good a chip as that 9950 you got . My max posted has been 3.2 ghz .

[merkk1]

O that with the HT and HTlink running at full speed defalut v and the FBS running at 231 on the 24/7 clock .

[HondaGuy]

Here is Unganged...



Now Ganged

[QuietIce]

Hi i run ganged all the time my hightes posted front FBS was 245 with my DDR-2 800 A-DATA 5-5-5-18 2t ram was running about 980 . I use a 3.003 ghz. mulit x13 FBS 231 ram running at 924 from my 24/7 clock . My 9850 be not as good a chip as that 9950 you got . My max posted has been 3.2 ghz .

Nice OC!
I had as yet to see numbers like that for K10 - 231 MHz 1:1 for the RAM is awesome!

I hope my luck is as good as yours.

Here is Unganged...

Now Ganged

So ganged allows you 0.5x higher clocks? That is odd as I've had better luck in unganged mode on the 9850BE ...

[HondaGuy]

I've been able to get 3.5 out of this Phenom on air, which was Unganged. I can get better scores while on Ganged

QuietIce
So ganged allows you 0.5x higher clocks? That is odd as I've had better luck in unganged mode on the 9850BE ...

I was just playing around with the Multi. I also changed my NB@2530

[QuietIce]

...
And AMD stopped using FSB 4 years ago so please stop using this term

AMEN!!!

[Jimmer411]

Started messing with HTT speed trying to get 3.2ghz+ and so far so good at HTT246. Ive gotten in windows just fine at 250HTT as well. The thing is that when I try 260x11.5 it fails to boot and lists clock speed at 3.38ghz, its driving me nuts because it seems to randomly just decide to use its own multiplier.

Anyone else have issues with this so called random multiplier?


So far it seems that my 9950 favors multis below 16x. But Im still testing things. Im still able to get into windows at 3.4ghz with the 17x multi, tho not stable as it randomly crashes. 16.5x is stable for about 5 mins of dual orthos where 17x would crash as soon as I hit start on both. 16x was rock stable for a few hours before I decided to end it. Now Im working on HTT overclocking while keeping the NB clocks as high as possible

Heres some validation goodness. http://valid.x86-secret.com/show_oc.php?id=415091


Edit: Im not sure why my validation shot does not list voltage, but Vcore is 1.55v

[i found nemo]

you should try 16x for a day and tweak from there.

[kazgirl]

Started messing with HTT speed trying to get 3.2ghz+ and so far so good at HTT246. Ive gotten in windows just fine at 250HTT as well. The thing is that when I try 260x11.5 it fails to boot and lists clock speed at 3.38ghz, its driving me nuts because it seems to randomly just decide to use its own multiplier.

Anyone else have issues with this so called random multiplier?

Hi I'm sure you know already, but about the random multiplier, you have got C'n'Q and C1E disabled hanv't you....also make sure that you have Vista/XP power settings to - Performance (Vista) - Home Desktop (XP) - to make sure that its not trying to use ACPI states....

Otherwise I'm not sure what could be causing the random multiplier, BIOS problem maybe.... theres a new BIOS available for this mobo I think its v.0304 at the ASUS website.

Looking forward to seeing you results I'm waiting for the M3A79-T at the moment.

[Jimmer411]

Ive got C1E disabled. Im going to look into the ACPI states right now, thanks.


Now that I read that its best to keep NB below or at 2600 I think I found whats causing it. 246x11 would have put me over 2700 on the northbridge. Lowering the NB multi caused the problem to go away. Ive got a chart that I wrote out on paper which pretty much has what multi to use for the NB at 2600, 2400, 2200 and 2000 speeds for any given HTT so that I can remain no higher than the targeted NB speeds. Ive also got a chart for which HTT speeds I need to attain 3.2ghz from each multi going from 10x to 16x just so I know which HTT to test to find my max. I plan on doing the same thing for 3.3ghz and possibly 3.4 so that I can find my max. Makes things a bit easier to have something to refrence without going to a calculator.

Once I get office installed Ill make a few excell sheets people can use for their own refrence.

[daltona]

When i use cpu-z 1.47 and look under mainboard, it says that my M3A78-T is with chipset AMD 780G ATI SB750, howcome its not 790GX and SB750. Why is that, and what about other boards.

[gOtVoltage]

Ive got C1E disabled. Im going to look into the ACPI states right now, thanks.


Now that I read that its best to keep NB below or at 2600 I think I found whats causing it. 246x11 would have put me over 2700 on the northbridge. Lowering the NB multi caused the problem to go away. Ive got a chart that I wrote out on paper which pretty much has what multi to use for the NB at 2600, 2400, 2200 and 2000 speeds for any given HTT so that I can remain no higher than the targeted NB speeds. Ive also got a chart for which HTT speeds I need to attain 3.2ghz from each multi going from 10x to 16x just so I know which HTT to test to find my max. I plan on doing the same thing for 3.3ghz and possibly 3.4 so that I can find my max. Makes things a bit easier to have something to refrence without going to a calculator.

Once I get office installed Ill make a few excell sheets people can use for their own refrence.

2700nb seems to be the Max for 9x50's.. Ive tested several CPU's( 2500 to 2680 is optimal) Depending on cooling and Volts run on Mobo . This ive found with most of the CPU freqencies ive used while running Ganged and Unganged (Both6400/8500 ram)..

Sometimes the Bios is the biggest factor with this.. a good Bios = great Oc period.


Note:
Ive found that the HTT likes 230/250 max Most Ram freqencies ive tested . Even though I can push the HT to 5200 it is Usually = to the Max NB you run for your given Oc with out exceeding 5200 .


My conclusion,,
Stable HT above 5200 and NB above 2700 might not be possialbe until 45nm and DDR3 Mobos which is desinged for HT5200+. Then we can push that limit further i believe.
More so i believe the Mobos and Bioses ATm are responsible for the limits of 9x50's. I can push my CPU to 3.55 with 2000HT 2000NB but Mobo holds it back( ,,Particulerly the PWMS cook with the Voltage it needs to Do it.

Your chart would be very valuble and save lots of time tweaking . Especially fo those people that have no unlocked multi going UP

P.S. i think CPU-Z need more updates for GX/750

[Jimmer411]

Mine shows the same things. Technically its still a 780G board, just with an updated southbridge which I imagine is why its called the 780GX rather than 880G.

Whats the max HTT your board will run? Also do you know what the SB Spread Spectrum does? There is a warning about not disabling it aswell as enabling it and another spread spectrum along with it. Ive ignored every other warning up till now since I started overclocking but Im not sure why I dont ignore this one lol.

I think that I found my max HTT. 250+ the board will not boot at 3.2ghz, but 249x13 http://valid.x86-secret.com/show_oc.php?id=415130.

[gOtVoltage]

Usually Spread spectrums are DISABLED for Overclocking and can affect the Power States/voltages used for CPU/satas/HDs when Enabled.. 750ATI chipsets may be different now to where it need to be Enabled for SB only..

Only Way to know is to try it out,, Wana be the Giunie piG

[Jimmer411]

Once I finish running dual orthos for stability, Ill be the pig

[kazgirl]

Hi Jimmer..

I've got am Excel Spreadsheet that I use for OC'ng with Phenom you might find it useful or not, but I thought I'd post it up.

Looking forward to seeing your results.

Phenom_OC_Calc.zip

[TheBlueChanell]

Hi Jimmer..

I've got am Excel Spreadsheet that I use for OC'ng with Phenom you might find it useful or not, but I thought I'd post it up.

Looking forward to seeing your results.

Phenom_OC_Calc.zip

Nice. This may really come in handy when I get my Phenom setup up and running. I need to sell this P5Q Deluxe/E8400.

[QuietIce]

Mine shows the same things. Technically its still a 780G board, just with an updated southbridge which I imagine is why its called the 780GX rather than 880G.

Technically it's the same but the graphics clock is kicked up a notch or two and there is the added "Sideport" memory.


BTW - You tried turning down the NB and HT Link multipliers for 260x11.5 ...?

[Particle]

The 790GX NB is as QuietIce describes--it's got a more powerful GPU solution. I'd not call it it the same thing as the 780G unless you would call a GeForce 8400 and 8800 the same just with a couple tweaks on the same basic core. That's just a personal opinion on conventions though.

---

If you get higher clocks in unganged, by all means stay in that mode. While Honda's Everest results may seem to favor ganged, please remember that that is a synthetic benchmark. In my own testing, I found that unganged without exception outperformed ganged in all 22 configurations of my gaming/3DM tests.

[oldpueblo]

The graphics core is really just a faster clocked 3200 though, so I'd say 780G +SB750 +Faster onboard is pretty accurate. Why re-spin something great and not just add to it?

[Jimmer411]

Yes I did try turning down the NB multi but was still unable to boot at 260 HTT. Right now Im running dual orthos at 246x13 with SB Clock Spread Spectrum disabled. I attempted to run dual orthos while at work at 249 and it crashed tho Im not sure at what time it happend. Same goes for 246x13 overnight. Im hoping that stupid setting is the culprit because Ive been able to play COD4 and Stalker just fine at these settings.

What do you guys use for testing stability in Vista 32/64? Could there be a chance that orthos just isnt the way to go with vista?

[Particle]

If Orthos is crashing, you aren't stable even if you haven't yet experienced a system crash outside of Orthos. Same goes for Prime95. Personally, I like to use Prime95 25.6 beta (MT enabled).

[gOtVoltage]

Orthos should be able to just fine from what ive seen and read for Vista.

Phenoms are very sensitive to High HTT/fsb Above 240/250..( Volts on the HTchipset to 1.28-1.31v are safe to use
24/7 with stock NB/SB heatsinks)

Ive found that some memory dividers will work better than others because of freqencies the phenoms run at..

The Higher the HTT the lower the Highest Oc will be ..

@288 i can only run Max CPU @ 3340mhz no matter the Volts or NB/HTT/HT settings.

@ 260 i can run 3450mhz STABLE but not Stable @ 3.5ghz NB2680
@ 250 and lower i can run 3450+mhz with Max NB 2690

Depending on the memory divider and BIOS will determin how how it can go..Even using HIGHvolts i can NOT go above 3340 @ 288+ with high NBfreqencies

There are so many variables with HTFreqency . Seems CPUxmulti and RamXmulti are a main key for stability.

Using DDR2-8500 1066ram allows a better Oc in my opinion unless you have real good 6400 that can run 533+..

@ 266 i run my Ram @800 = 1066 5-5-5-15-2T

Some memory xmultis will work better than others with same htt/ht setting.. i can use 533 but not 400 . I can alsouse 800 and 1066 but not 667 dividers when i try any HTT/HT combination. Memory dividers will chage as the ram and HTT go higher some just work better at all speeds vs others..

But to go Higher with HTT only causes you to lower your Overall OC and stability by not keeping things closer to 2000HT and HTT

My Max (Not Prime or orthose stable)..More of a Suicide runThe Boards PWM's fail @this speed from heat. i can only do this run with HTT200



34.8ghz/Nb2680 .. that goes out the window once i try to raise HHT over 268..but if i keep it below 260 i can do 3.5ghz with a little less on the NB with better performance.



Overall CPU is Higher but with lower HTT on this run...Performance is better @3.5/nb2600 vs 3.84/nb2680



Last this is what i am Running ATM 24/7...



BIG NOTE: these runs are with HIGHVOLTS (CPU and NB above 1.55v "Do not try this with out XtremeCooling"

High HTT isnt all that and i can get the same performance using lower HTT = Stability @3.45ghz vs 3.5ghz HTT260 and be UNstable.

P.S.
Didnt mean put so much,, Sometimes its easier to just see why ,,than for some one to type numbers

gOtVolTage

[Jimmer411]

Very nice, but now I get to throw all instabilty testing Ive done out the window. Vista auto updates at 3am everyday so my overnight orthos may have been stable after all. I caught it today doing a countdown to restart. It had killed one process of orthos, the others was 11hr stable and running. grrrr