Offler

100 % stable at last:

FSB 400 Mhz
CPU: 3400 Mhz
RAM: 1600 MHz cl 7-7-7 Dual channel

CPU: 1.275v
FSB VTT: 1.21v
RAM: 1.78v

voltages for NB and SB were of course on default values to see whether the board itself is or is not stable.

Important things to take care of:

a) When the CPU is not fully stable the OCCT will report strange voltages in graphs. For example my CPU had 1.9v for a second, or 12v rail reported only 5 volts. All voltages were unstable.

If values reported were true and not only data error due CPU issues, it will of course impact whole computer, causing errors on places where there is no problem. Also this can solve why those boards are so hard to OC.

Since the 3.3v rail was also impacted, it can surely answer why the TV tuner is so unstable - just because it is on PCI slot - pretty example - i was believin that the TV tuner does not work well, but it was a mistake.

b) The CPU unstability could be masked behind normal performance of system, and good running benchmarks. But LinX and/or OCCT will help find any problems.

c) I can only recomend to you OCZ memory for this board. But be careful. X48 likes 1,9v for 1600Mhz. It is good to buy the ones certified for Intel - they run on 1.9v, but you can choose even platinum series with 1,65v, althought the resulted voltage must be somewhere in between. For me the VDIMM voltage is such result.

d) Win 7 x64 is much better for stability tests. The 64bit apps are more like to crash during the testing like 32bit ones. This will show you the way which configurations are stable or really stable.

After last test (setting CPU to 2800Mhz and default voltage fo Core and VTT and stability measure) i can finally proceed to the OC... It took me about 9 months to see where has this board its strong and weak sides. As always Mr. Wu did a great job, but only few OCers will be able to benefit from the full potential which this board offers. It is really hard to control this board to that level when you can identify the problem.

In i few weeks i shall finally post some results which will tell more about real OC potential of this board and Core 2 Quad CPUs. Just fear a bit that it will be still somewhere arnound 440 Fsb, but now i am sure that the limit was the CPU and its heating abilities.

eva2000

Nice info, yeah 64bit apps test more addressable memory space

Offler

Most important:

If you wish to run memory on full power of this board (400/1600) check this:

http://www.ocztechnologyforum.com/fo...LP-UT-X48-T3RS

I used DRAM driving strenght and Default skew models from this page - 1/1 7/7 and they work perfectly. Also i had to add a bit voltage to CPU.

Range from 1,225-1,275v is bootable and will work but only after hard tests by OCCT you can be completely sure about stability. Right now the 1,28v is the voltage and I hope it will not consume more than now.

Offler

What if the source of hissing is the beeper?

eva2000

Honestly i wouldn't worry about the hissing

Offler

Yeah. I need a shrink

I checked Battery voltage. At 333 Mhz and default voltages for everything system measured 3,3v. When CPU Core has too few volts the battery voltage will be lower and thus system will be unstable.

At least i have 100% guaranteed indicator of cpu stability.

eva2000

Interesting first time hearing about that

Offler

Tested bit more. Works very very well, but the voltage may change in different ambient temperatures.

For example day before yesterday i had 25 degrees in my room and the battery had 3.3v measured by OCCT at 333Mhz (defaults for everything).

Yesterday i had 20 degrees and voltage on battery was 3.28v on default, while system became unstable when the voltage was 3.26v.

At all I can compare voltage on defaults with voltage on desired FSB frequency.

As a side effect I was able to start the system with PCI Spread spectrum 0.30% and whole computer worked very well. Another sign of better stability. Previosly this option crippled the system on FSB 400. ATA was not working, SCSI was very slow during bootup, network was gone...

Very sensitive board...

Edit: The board booted by accident. There is 25% chance that the board will boot normally with PCI Spread spectrum.There are two ways how to fix this:
1. Turn off the IDE Jmicron chip.
2. Set Clockgen voltage to 3.6v

Since my Corsair HX650W is in RMA i had to use Arctic cooling PSU 550w with 80+ certification. Certainly this PSU cannot feed the CPU (or at least i have to raise the CPU VTT and core voltages), so all tests have to wait until i force the shop to accept the RMA.

At all there are only few possible reasons of such behaviour.
1. The Jmicron chip does not have enough power.
2. The Jmicron chip is damaged, or is bad choice for this board, or the construction on this place of board is bad so the chip does not work well.
3. The Clockgen does not have enough power.

Since the clockgen works well for CPU, ram and others even with CPU spread spectrum ON, this can clear why the board might be unstable on certain situations. If the jmicron chip fails to communicate it is same like to remove a PCI card from non-hotplug PCI slot - instant BSOD or reboot. Also i dont know where it gains power - the 3.3v rail provide only 3.16v at most.