**Unified** DFI LanParty UT 790FX-M2R-M2RS / Phenom /Data Base Thread**

[Brother Esau]

I cannot see it being a issue seeing how it is a B3 Revision chip

[justapost]

mmh i dont think the first kit will fit. looks like the pwm section of the 680SLI board is designed differently..

Thanks, also too expensive for what i plan.

of course one can mount some small coolers that fit somewhere on that cooling system.. there was a suggestion in that thread also..

http://www.forumdeluxx.de/forum/show...&postcount=578

i think such will lower the temps a little without giving too much heat into the cooling system.. and one can use the heatpipe to cool sb, nb and pwms..

here car_bon shows how he made it, was a watercool cooler..

http://www.forumdeluxx.de/forum/show...postcount=1384

That mosfet cooler fits perfect did car_bon report how well it worked? I spended a whole evening reading this interesting thread, thanks for posting.

BTW: AMD Support does not want the second 9500 back.

[Final8ty]

My AMD 9950BE JAAHB AA 0280BPMW
15X Multi stock volts
see if it locks up later.
going to play some games.

[Oese]

Thanks, also too expensive for what i plan.

That mosfet cooler fits perfect did car_bon report how well it worked? I spended a whole evening reading this interesting thread, thanks for posting.

BTW: AMD Support does not want the second 9500 back.

i'm sorry, it seems car-bon disappeared from that thread.. But I think it should work a lot better then the original transpiper, as the contact surface is higher and the temperature is lower then that of the transpiper, i suppose even if you take into account the heatpipe that maximizes heat transport...

[Final8ty]

Well BF2142 crashed.
Upped the volts to 1.4 still crashed.

Not getting the CPU Clock int was not exc.. every time tho like before.
got into windows @3Ghz 1.26v.

I think it really is a SB600 problem & my one seems to be more crap than others.

[justapost]

Well BF2142 crashed.
Upped the volts to 1.4 still crashed.

Not getting the CPU Clock int was not exc.. every time tho like before.
got into windows @3Ghz 1.26v.

I think it really is a SB600 problem & my one seems to be more crap than others.

That's no good news.
My 9850BE runs fine at 2.9GHz 1.475V-1.6V but no way to get 3GHz stable. Hope my 9950be arrives soon (expected today) so I can compare.

The clockgenerator chip used is the 9LPRS477. That chips is also used on the M3A and the GBT780G. The clockfrequency is generated via an multiplier and two diverent divisors. I wrote a linux kernel module for that chip so I can modify all three now. Using different divisors made no difference in terms of stability.

Anyway, thanks for posting your results. Your on 64bit vista ?

EDIT: 1.6V + 100.26% seems to be somewhat stable at 3GHz, at least the system did not immediately power off after starting CoreDamage. However the cpu and core temps are above 60°C now, so i'll not further investigate here.

[Final8ty]

yes vista 64..

[justapost]

yes vista 64..

Hmm, best i could get with linux 64bit was 2.7GHz for all cores at 1.35V. I tried each core and best i got was 2.9/2.9/2.7/2.7GHz. That worked fine at any voltage above 1.35V but no way to get a single core multi one notch up.

Can you run 2.9GHz stable at stock?

[Final8ty]

Dont know yet as im still trying to get 3.0 stable by messing with other things

[Brother Esau]

yes vista 64..

Thats why you need so much voltage and its not stable. Try XP32 bit and get back to us.

[justapost]

Dont know yet as im still trying to get 3.0 stable by missing with other things

He He, good luck. Must say I expected 3GHz with a 64bit os as the worst case oc. DDR800 at ~1.8B seems alot more stable than 1066 at 2.2V here. Beside that i tried 1.4GHz HT speed, 195ns RFC's higher/lower chipset voltages, normal/weak driver mode, different PCIe speeds and gap's but nothing helped.

[Final8ty]

Thats why you need so much voltage and its not stable. Try XP32 bit and get back to us.

No need as we all know that a 64bit OS is more of a strain than a 32bit OS.
This 939 was not stable at the same volts when i put a 64bit OS on it.
http://www.xtremesystems.org/forums/...postcount=3466

[ea6gka]

ok chaps got my 9950 , bug no1 the multi isnt working...set it 14 and it just doesnt work

edited , just seen Final8ty has the same setup and his multis are working , which bios mate? i have the latest non beta

[Final8ty]

ok chaps got my 9950 , bug no1 the multi isnt working...set it 14 and it just doesnt work

edited , just seen Final8ty has the same setup and his multis are working , which bios mate? i have the latest non beta

The latest beta.

[ea6gka]

damn those betas

or do i also have to change the cpu FID to make it work?

[Final8ty]

damn those betas

or do i also have to change the cpu FID to make it work?

The multi will override the FID.
clear CMOS ?

But im using the FID now for my 3Ghz

[JWilson]

[justapost]

HD995ZFAJ4BGH JAAHB AA 0820APMW

just arrived.

For comparison 9850 power consumption and temps at stock:

Idle: 153,7W CPU 21°C CORES 33°C
Load: 243 W (everest julia)
Load: CPU 42°C CORES 45°C (coredamage 5mins)

9950BE at 2.5GHz 1.3V

Idle: 169W CPU 27°C CORES 26°C
Load: 267W (everest julia)
Load: CPU 52°C CORES 45°C

So the 9950BE requires ~15W more than the 9850BE here, temp reading seems more correct on the 9950BE.

[ea6gka]

HD 995ZFAJ4BGH JAAHB 0820GPMW here

[justapost]

WinXP SP2 32Bit Test

3GHz 1.35V prime95 stable (5 mins)
3.1GHz 1.35V prime95 failed after a few seconds
3.1GHz 1.375V prime95 failed after a few seconds on third core
3.1GHz 1.35V prime95 failed after a few seconds on third and fourt core
3.1GHz 1.4V prime95 failed after one minute on third core
3.1GHz 1.425V prime95 stable (10 mins) cpu temp has spices into 60°C area.

So far so good, next step is linux 64bit.

[ea6gka]

your bios settings are the same as the ones you post in the firsty pages justapost?

[justapost]

your bios settings are the same as the ones you post in the firsty pages justapost?

Hmm no, the Crucial Ballisitcs degenerated so i use G-Skill 6400 4-4-3-5 Mem now.
To be save i use 1.4GHz NT speed and run the memory at 5-5-5-15-195ns/1.85V.

I'm happy with my first linux testings. Machine seems stable at 3GHz 1.425V. I use the stresscpu2, build-linux-kernel and multicore tests from the phoronis test suite for stability testing. The first one does heat up the cpu alot so it's comparable to coredamage under windows. The other two tests are the ones which normaly fail first if the system is unstable.

UPDATE:

Linux 64bit (Debian lenny):
3.0GHz stable at 1.35V (kernel build caused kernel fault at 1.325V)
3.1GHz stable at 1.425V

That chip has the same limits here as under winxp 32bit.

[justapost]

Best suicide at 1.575V. The first core is not stable at voltages above 1.5V.
3.2GHz is not prime95 stable at 1.575V and fails booting windows at 1.6V.

My cpu temps reach 70°C at those voltages so maybe getting em below 60° might help here.

[Brother Esau]

Is that a screen shot from New 6/12/2008 Beta Bios for 790FX-M2R????

[JWilson]

No, it's from the mod bios in my siggy.