Jetway MA3-79GDG Combo

[justapost]

http://www.abload.de/img/firstie9n.jpg (http://www.abload.de/image.php?img=firstie9n.jpg)

http://www.abload.de/img/secondjl0r.jpg (http://www.abload.de/image.php?img=secondjl0r.jpg)

My new toy arrived yesterday. I use an 5050e cooled with an Big Typhoon together with 2x1GB CB Tracers, an 74GB WD Raport WD740ADFD an cheap LG DVD drive and the Gigabyte Odin 550W PSU.

The board should be able to handle AM2/AM2+ and AM3 chips and 2xDDR2 or 2xDDR3 dimms.

Will add more pics once I get my hands on a better camera.

To overvolt the cpu this board uses a so called CPU Vcore 7-shift feature, wich increases the voltage in percentages from 5% to 35% percent in steps of 5%.

CPU-Z only reads the CPU VID without that percentage and Everest 5 stable does not read the cpp vcore at all.
So the only tool to read the actual vcore is the MyGuard app coming on the tools and drivers cd.

http://www.abload.de/img/ss1jtpa.jpg (http://www.abload.de/image.php?img=ss1jtpa.jpg)

The screenshot shows my inital oc with 1.25V cpu vid and 20% voltage increase means the vcore should be 1.5V. The tool however shows just 1.45-1.46V here.

First I tried it with an winxp sp3 os which I had used on other boards before. The board powered off everytime I tried to enable the energy saving power profile, so I decided to use a fresh installed os instead but the problem remains. So with the initila bios CnQ does not work.

[justapost]

Found what pwm controller is used on the mobo. It's an Richtek RT8855, I asked for the datasheet and hope they send it to me.

http://www.richtek.com/product3.1.1.jsp?serno=13

The RT8855 is a 4/3/2/1-phase synchronous buck controller with two integrated MOSFET drivers for CPU power application and a single-phase buck with integrated MOSFET driver for North-Bridge (NB) chipset.The RT8855 uses differential inductor DCR current sense to achieve phase current balance and active voltage positioning.
Other features include adjustable operating frequency, power good indication, external error-amp compensation, over voltage protection, over current protection and enable/shutdown for various applications. The RT8855 comes to a small footprint with WQFN-48L 7x7 package.
(2008-01-10)

[justapost]

http://www.abload.de/img/243ku1z.jpg (http://www.abload.de/image.php?img=243ku1z.jpg)
This is the max freq I can reach with an 20% voltage increase over stock. I went on thean 25% and now the myguard tool reads 1.5V which I expected for the 20% increase.

http://www.abload.de/img/247hhof.jpg (http://www.abload.de/image.php?img=247hhof.jpg)
With 25% I could increase the ref HT to 247MHz resulting in an 3211MHz cpu frequency. Now testing 30% increase, myguard reads 1.55V in idle and 1.56-1.57V under load. Richtek mailed me the pwm controllers datasheet, I hope I find good points to verify the cpu and nb voltages sood. Those specs show the pwm chip has an i2c/smbus interface so it should be possible to change the vcore on the fly with an small rw-everything script or with i2ctools under linux.

http://www.abload.de/img/2547nu4.jpg (http://www.abload.de/image.php?img=2547nu4.jpg)
Max is 254MHz means 3302MHz with an 30% voltage increase. Sonce temps are below 60°C I'll try 35% increase now. Myguard reads 1.6V in idle and 1.61V-162V under load. To me it looks like it is an normal voltage offset and no percentage increase here. 35% mean 0.35V+.

http://www.abload.de/img/2579zx0.jpg (http://www.abload.de/image.php?img=2579zx0.jpg)
Chip has the same limit here as the other 5050e I tested few weeks back on the asrock board. There 256MHzx13 was the max stable oc but the voltage requirements where much lower there it only required 1.475V in the bios and cpu-z had showed 1.525V there.

http://www.abload.de/img/254-2zjgy.jpg (http://www.abload.de/image.php?img=254-2zjgy.jpg)
For comparison I threw in the other 5050e. With an 35% voltage plus the system powered off after a few minutes. Now with just an 30% increase the temps are alot higher than with the new one.

I'm done with the am2 dual cores for now, gonna test my 9950BE now.

[justapost]

http://www.abload.de/img/15x2005fya.jpg (http://www.abload.de/image.php?img=15x2005fya.jpg)
Seems ACC does not work with the 9950BE here, I need 1.435V to run 3GHz stable. No way to get even 3.1GHz stable on air.
Also the bios has no NB FID/VID fields. The bios however has an field to select an AOD compatibility mode but if I enable that mode there is no way to selevt vdimm anymore. Bios pics up only SPT but no EPP info's from the mem so the CB tracers run at 2x400MHz mode with 1.95V.
On the GFX side the only thing I can modify the the GPU clock, no sideport speed or voltage options are available.

[justapost]

Well even 3GHz did not pass an long p95 test. Decided to put the 940BE in an it ain't gettin better with k10.5 here. For 3.6GHz I must pick 1.5125V in the bios. That results in an 1.5V CPU VID. MyGuard reports 1.51V in idle and 1.52-1.53V under load. Temps reached ~60°C in the first hour of the p95 test. I also tested with an small refHT increase at ~3.63GHz and the system silently rebooted after a few minutes. Might try it with an XIG cooler and two scyte fan's this week. As a sidenote at least CnQ worked with the 9950BE and the 940BE. Glad I did not buy this board for overclocking purpose because it's not much fun. Most tools report totaly off values, even AOD does calculate an wrong refHT with AOD compatibility mode selected in the bios. I intentionlay planned to replace my old 690g board in my 24/7 office machine and reuse the BE2400 chip till an low power AM3 chip is available. But without CnQ working that plan has to be delayed till eighter a new bios is out which fixes this power off issue in CnQ with the 5050e.

http://www.abload.de/img/3600-158l7q.jpg (http://www.abload.de/image.php?img=3600-158l7q.jpg)