DFI UT X48-T3RS LanParty - Living Review

[kpmkieran]

hi clunk,

recently bought this board... a great board... so anyway... ive run into my first problem... the board doesnt seem to be able to run dual channel memory... ive tried my memory which is a set of OCZ 2GB DDR3 PC3-12800C7 1600MHz Platinum Enhanced Bandwidth (2x1GB) Dual Channel DDR3 (OCZ3P1600EB2GK)... paired in both the green and yellow slots... it stops on C1 and fails to post, however each dimm works perfectly on its own in any of the slots... can i assume the ram is fine?

have you any thoughts or suggestions? is this a problem you have come across before?

ive upgraded the bios... first to 710... and again to 821... without any change.

[Praz]

May be DLL models not quire right. I've had the same memory stable at 2000MHz. Almost has to be some setting.

[Clunk]

Originally Posted by kpmkieran

hi clunk,

recently bought this board... from supershanks infact (hey), a great deal... thanks again... so anyway... ive run into my first problem... the board doesnt seem to be able to run dual channel memory... ive tried my memory which is a set of OCZ 2GB DDR3 PC3-12800C7 1600MHz Platinum Enhanced Bandwidth (2x1GB) Dual Channel DDR3 (OCZ3P1600EB2GK)... paired in both the green and yellow slots... it stops on C1 and fails to post, however each dimm works perfectly on its own in any of the slots... can i assume the ram is fine?

have you any thoughts or suggestions? is this a problem you have come across before?

ive upgraded the bios... first to 710... and again to 821... without any change.

Welcome to the forums

As Praz says, it's probably just a setting or two.

You on the latest BIOS etc? If the board still has the shipping BIOS, it will have problems with your CAS 7 memory, you need to update it.

[kpmkieran]

the bios is 821... still only working with 1 dimm and not dual channel. tried adjusting the strap/fsb with no luck.

[SquidgyB]

New beta BIOS:

Download BIOS

Updated microcode for 45nm cpus...

I completely forgot to add as well: I had what I consider to be a minor breakthrough with my QX9650 at 400fsb. After 0821 it seems much more stable without any changes to GTL ref voltages... All I have to do now is bump CPU, NB and VTT (about 1.3, 1.45 and 1.3 respectively) to keep it stable(ish). It's a great improvement over what I could get previously, but memtest seemed to be flaking out so I went back to a multiplier overclock (3.6Ghz cpu, 333fsb with tight memory at 1333 6-6-6-18 1N etc) that's rock solid.

The board seems to be getting better and better for quads with each bios update. w00t, I think is the word I'm looking for

[WILSON]

Just d -loaded it now i think i will try for a higher oc this week. Alos has any1 used this stress test.
IntelBurnTest v1.6 download from Guru3D.com

[Praz]

Originally Posted by WILSON

Alos has any1 used this stress test.
IntelBurnTest v1.6 download from Guru3D.com

I've been using LinPack it seems like forever and this is just a nice frontend for it. It work really well for bringing voltages and settings close to stability in a short amount of time. Contrary to what people are posting though, it is not the final word on stability. After a successful run of IntelBurnTest follow up with your usual stability tests.

[c0ugar]

im having problems finishing the super-pi 1.5 test when i run it on 32mb and 16mb and 8mb it does not finish no matter what settings i use. over clocking ones, default super-pi never finishes it just <not responding>

[Praz]

Originally Posted by c0ugar

im having problems finishing the super-pi 1.5 test when i run it on 32mb and 16mb and 8mb it does not finish no matter what settings i use. over clocking ones, default super-pi never finishes it just <not responding>

If using Vista 64Bit need to use the HyperPi frontrend.

[SquidgyB]

I haven't used Super-Pi or IntelBurnTest yet as a stability test, I've always used Prime95, AtiTool and MemTest 2.01 to do the dirty work

I had fun last night, for the first time I've gone 8 hours on prime with no problems:



Last night I was just trying to get MemTest not to error out when running with really loose latencies, as you can see from MemSet in the pic. Every single time it would error on test #8, at between 60% and 80% through the test. I was basically going through the bios setting by setting, trying different phase settings and slew settings to see if any of them had an effect on stability, meanwhile, trying to find more info and reading up on different bios settings on my laptop.

Then I saw a post by Praz at some point, where he set DRAM CLK Driving Strength and DRAM Data Driving Strength to 1, so I tried that. test #8 went through fine, and it was late, so I set prime and atitool to run through the night, and it's the first time they've not had a problem with such an overclock.

I'll post accurate bios settings when I'm back home tonight, I set prime running overnight and didn't have time this morning before work to get the settings. In any case, CPU voltage is 1.328 in cpuz, NB was set to 1.5, VTT to 1.3. Pretty much everything is on auto including GTL ref (yay!), bar 9-9-9-28 on the memory, turning VDroop off, and obviously the FSB/strap settings. DLL Model is set to 3.

I'm confident most of these voltages will come down quite easily now that I've actually found a setting that appears to get the memory nice and stable, I'm just glad I've found something that appears to have made a big difference... It'll be fun from now on seeing how much further I can get the FSB to go, and how much lower the memory latencies can be pushed.

Oh happy days!