The Official Opteron 939 Overclocking Thread

[dogsx2]

I read through the first posts and I think even the CABNE's need ~ 1.45 vcore for 3ghz stability.
...

Some do, some don't.

http://img285.imageshack.us/my.php?i...930occt2gn.jpg

3000@1.4v

[ozzimark]

yeah, mine is definitly one of the lesser ones unfortunately. it loves voltage though

in summary:
max memtest boot: 2944mhz
max memtest stable for 1 pass: 2862mhz

max 8m = 2724.6mhz
max screen = 2901.7mhz

i didn't have the patience for 32m, so here's the max 8m shot:

subtract 9mhz for max 32m

here's max screen:

i validated it too, but unfortunately i hit f7 before cpu-z updated, so the validation is only 2894mhz
http://valid.x86-secret.com/show_oc?id=43936

[dinos22]

i validated it too, but unfortunately i hit f7 before cpu-z updated,

so doesn't hitting F5 only give you a BMP image in CPUID folder rather than validation link file

[ozzimark]

so doesn't hitting F5 only give you a BMP image in CPUID folder rather than validation link file

yup, i hit f7 then f5, it updated between the two, thus why i have a valid as 2894mhz and a screen at 2901mhz


nisco, remind me.. what's the cooling on that?

[dinos22]

yup, i hit f7 then f5, it updated between the two, thus why i have a valid as 2894mhz and a screen at 2901mhz


nisco, remind me.. what's the cooling on that?

so does F7 give you the cvb file or whatever it's called

[NISCO PT]

nisco, remind me.. what's the cooling on that?

sorry
prometeia mach1 moded with r404a

[Lane-k]

This opteron seems realy fine... What about the same in dual-core version and higher level like the 152-154... will those make better of the FX57?
(I will upgrade sooon... so... when i see the result of the 146..wow)

[eva2000]

yeah, mine is definitly one of the lesser ones unfortunately. it loves voltage though

in summary:
max memtest boot: 2944mhz
max memtest stable for 1 pass: 2862mhz

max 8m = 2724.6mhz
max screen = 2901.7mhz

i didn't have the patience for 32m, so here's the max 8m shot:
http://www.xtremesystems.org/forums/...id=38720&stc=1
subtract 9mhz for max 32m

here's max screen:
http://www.xtremesystems.org/forums/...id=38721&stc=1
i validated it too, but unfortunately i hit f7 before cpu-z updated, so the validation is only 2894mhz
http://valid.x86-secret.com/show_oc?id=43936

thanks just the info i wanted...

with my Opteron 144 0536 GPMW

so far memtest looks like this all 1:1 mem/cpu

AUTO Vcore = 2826mhz
1.425v = 2835mhz
1.45v = 2880mhz - 6 full loops
1.475v = 2906mhz 6 loops so far (this is 9x323mhz 2.5-4-3-7 which is the known limit of the ram on FX-57 so interesting to see how much higher it can go cpu and mem wise from here)

1.475v = 2926mhz running memtest so far at 9x325mhz 2.5-4-4-8 so far okay

1.50v = 2944mhz 9x327mhz 2.5-4-4-8 at 2.9v vdimm = test #6 errors - cpu related... so guess this is the point where more vcore is needed heh - but will try 1.50v but up vdimm to 2.95v first to see if it's mem before raising vcore

1.50v = 2944mhz 9x327mhz 2.5-4-4-8 but with 2.95v vdimm passed test #6 no errors, so was vdimm not vcore - so far 2 full loops error free

1.50V = 2961mhz 9x329mhz 2.5-4-4-8 at 2.95v so far memtesting okay passed 1st full loop, but 2nd loop at test #8 memtest error, so next stop 1.525v vcore to see if it's vcore of mem/vdimm

1.525v = 2961mhz 9x329mhz 2.5-4-4-8 at 2.95v passed 1 standard loop and a half

1.525v = 2980mhz 9x331mhz 2.5-4-4-8 at 2.95v booted into memtest and running up to test #5/6 errors now so more vdimm or vcore ?

1.525v = 2980mhz 9x331mhz 2.5-4-4-8 at 2.98v vdimm same test #5/6 errors, so guess time to run 180 mem divider

1.525v = 2980mhz with 9x331mhz with 180 divider mem @298mhz 2.5-4-4-8 at 2.95v standard loop 1st pass error free running up to test #5 in 2nd loop fine so far - 3ghz looks possible you'd think

[Zeus]

What with this MaxTcase thing?
Higher temperature rated chips are best?

[mongoled]

What with this MaxTcase thing?
Higher temperature rated chips are best?

Yup mine is 67C also.

Just something which I have mentioned be4 something I think is relevant because alot of the talk here is abt wht vCore a particular chip needs to get a certain mhz.

Its no good saying

'woot, im using 1.45v for 3Ghz stable'

if in reallity your vCore is 1.52v. Why I say this? Well my Ultra-D overvolts between 0.04-0.07v depending on the setting chosen in bios. Another member who posted in this thread (soz cant remember name) also said their DFI overvolts by abt 0.07v, now these are quite large differences from bios reading and mulitmeter reading points.

If their is going to be any chance of correlating data to find if MaxTCase has any real meaning then 'real' vCore needs to be taken into account!!

[sierra_bound]

Opteron 148 CABYE 0528GPMW

Air-cooled with Zalman CNPS 9500

[xforce]

Opteron 148 CABYE 0528GPMW

Air-cooled with Zalman CNPS 9500

Where did u buy yours ?

[rise4310]

Yup mine is 67C also.

Just something which I have mentioned be4 something I think is relevant because alot of the talk here is abt wht vCore a particular chip needs to get a certain mhz.

Its no good saying

'woot, im using 1.45v for 3Ghz stable'

if in reallity your vCore is 1.52v. Why I say this? Well my Ultra-D overvolts between 0.04-0.07v depending on the setting chosen in bios. Another member who posted in this thread (soz cant remember name) also said their DFI overvolts by abt 0.07v, now these are quite large differences from bios reading and mulitmeter reading points.

If their is going to be any chance of correlating data to find if MaxTCase has any real meaning then 'real' vCore needs to be taken into account!!

amen to that. every dfi nf4 (sli-dr and ultra-d's) i've had/have reads .06-.07 lower in windows and .04 lower in bios than what i actually set.

i think ut would be more accurate if people would state what they actually are setting.

[ozzimark]

1.525v = 2980mhz with 9x331mhz with 180 divider mem @298mhz 2.5-4-4-8 at 2.95v standard loop 1st pass error free running up to test #5 in 2nd loop fine so far - 3ghz looks possible you'd think

hey! that's not stock voltage
i guess i'll try something just as in-depth. i'm not exactly sure where this is leading though