DFI UT X48-T3RS LanParty - Living Review

[BlackDragon24]

Well I can finally get going with the 400/1600 divider now that I found this.

In the fine settings section, I can use 3,8,5,5 or 3,8,6,6 for my ddr3 1800 reaper kit.

I'm fine tuning right now....but it is HCI memtest stable at 9.5 x 450 ddr3 1800 8-8-8-27 1T tRD 8 400/1600

Problem is you get better overall memory bandwidth running ddr3 1600 speeds at 7-6-6-24 1T tRD 8.5 400/1333.

So I think the best setup to obtain ddr3 1800 speeds would be to run 540 fsb on the 400/1333. But that's gonna take some work for sure. Everyone seems to think the 400/1600 is the way to go....I find the 400/1333 with higher fsb and tighter timings with looser tRD seems to work best.

Still testing....

[Clunk]

Sorry mate, I just realised I said 400/800 back there, I meant to say 400/1600

[WILSON]

Well i seem to be stuck at my 4.2Ghz oc so im gonna try and get some more out of my ram. It is rated for 1333Mhz @ 9.9.9.24 so is running it at 1399Mhz @ 8.8.8.23 going to be asking to much and if it happens to be memtest stable is it a classed as a mild overclock reg ram??? I only ask as i dont really mess about with ram to much and don`t want to end up frying it .
Thanks Paul

[Clunk]

Only one way to find out

You won't fry it unless you give it excessive voltage. What voltage does it run at normally?

When you start playing with the memory, it's always best to save a stable set of BIOS settings in CMOS reloaded because you will no doubt end up clearing the CMOS quite a bit on the T3RS

[BlackDragon24]

This board is starting to mature as more info comes out.

Alright....now that I've consulted the DFI BIOS guidelines for how to set skew models, things have improved dramatically. Using 3,8,5,5 and 3,8,6,6 for the fine delay settings that seem to be so important on this board. This also finally allows me to run the 400/1600 divider, which allowed me to clock my memory up to ddr3 1800 8-8-8-27 1T tRD 8 (HCI memtest stable)

Unfortunately the memory bandwidth on the 400/1600 divider sucks. I get much better bandwidth on the 400/1333 divider with higher fsb, tighter timings, and lower tRD.

So here is my latest result...the farthest I've gotten. It is OCCT stable for 1 hour right now. Basic info:

E8600 8 x 530 1.300 vcore, 1.205 vtt, 1.555 NB, 1.894 RAM (BIOS settings)
Reaper 1800 kit at ddr3 1766 8-7-7-24 1T tRD 9

In order to run up to 530fsb, I has to increase the cpu0 clock skew to 200ps. 100ps was good up to 520, but once I hit 530 I needed 200ps just to get into windows. I have also raised my cpu vtt to 1.2 but I don't know if that has any effect yet.




Now here is the problem. I want go get to 8 x 540 but I'm having a heckuva time. I already know the chip can do 4.32 at 9 x 480 at this voltage, and I know that the mem can run at ddr3 1800 on the 400/1600 divider at this voltage. So it has to be a fsb stability thing.

So far I cannot get 8 x 540 stable. I've tried raising vtt to 1.25, nb voltage to 1.579, etc. I can get into windows but it crashes pretty intermittently. If I raise cpu0 clock skew to 300ps it won't even go into windows. Tried lowering timings to 8-8-8-27 with no luck.

Any tips on how to stabilize higher fsb OC's?

[Praz]

Originally Posted by BlackDragon24

Unfortunately the memory bandwidth on the 400/1600 divider sucks. I get much better bandwidth on the 400/1333 divider with higher fsb, tighter timings, and lower tRD.

That's because Everest type programs thrive on FSB.

[BlackDragon24]

Got a program in mind that might be a better measure of memory performance?

[sunderstan]

Hey guys I'm Stanley greeting from Hong Kong
I've bought a DFI UT-X48-T3RS last 2 weeks ago and found it quite difficult to tweak.
The reason is it's result is so unpredictable that I can't even figure out which setting is holding me back.
Things like "just saved a BIOS setting that can successfully boot just yesterday but NO POST today."
Therefore I would like to share some of my experiance on tweaking this motherboard and discuss together on what's actually going on...here's my config:

DFI LP UT-X48-T3RS BIOS 0829
Intel Core 2 Duo E8400 Q807A191 (Will replace a second handed E8600 Q820A628 soon)
Cricial Ballistix PC3-1333Mhz CL6 2GB Kit D9GTR
Delta ServerPower 650W 84% Eff.
Colamax HD 4850 X2 PCS (one of them is offline right now)
Western Digital 5001ABYS x1, 5002ABYS (Dual I/O) x1

One of the weird cases:
I was able to get 500FSB rock solid stable for 7/24 use, but with the same setting, just set back to 400FSB, result in NO POST.

Findings:
Set FSB Strap and clock skew model at the same time will result in NO POST.
With 333 Strap max stable FSB can get was around 410, random freeze will occur if go further no matter how high the voltages set.
Using 266 strap 266/800 aka 2:3 divider, 333FSB POST successful but stuck at debug code 50.
Max stable FSB overclock so far was 520FSB, with Orthos Large FFTs fail after nearly 8 hours.
550FSB success POST but very unstable no matter what voltage or tweak settings used, possibly CPU issue.
This CPU can successfully POST at 580 FSB @Asus P5Q-Deluxe but no go on loading windows instant BSOD.
OC Fail Retry is not functioning. Always need manually clear CMOS.
For any adjustments that will need to alter DRAM Skew the system will do a full reset. E.g. FSB frequency, tCAS, tRD(not sure), DLL Default Skew Model, DRAM Divider.
DIMM slot 3 and 4 is rather difficult in controlling skew settings for a success boot. Still investigating.

Annoying part:
No current DRAM Timings displayed in BIOS. Auto will set to optimal setting rather than SPD. By using 200/800 Divider, tRD was set to 4 and tCAS timing was set to 6-6-6-15, with tCAS set to 7 or above will result in NO POST.

Will add and post more if anything new is found.

[Clunk]

Originally Posted by BlackDragon24

Got a program in mind that might be a better measure of memory performance?

The best test of your memory performance is to run the programs that you use on a daily/weekly basis and see if you get any gains with them - If not, you can safely cut back and save electricity

I realise that's not the answer you were looking for, but I thought I would say it anyway

[Clunk]

Originally Posted by sunderstan

Hey guys I'm Stanley greeting from Hong Kong
I've bought a DFI UT-X48-T3RS last 2 weeks ago and found it quite difficult to tweak.
The reason is it's result is so unpredictable that I can't even figure out which setting is holding me back.
Things like "just saved a BIOS setting that can successfully boot just yesterday but NO POST today."
Therefore I would like to share some of my experiance on tweaking this motherboard and discuss together on what's actually going on...here's my config:

DFI LP UT-X48-T3RS BIOS 0829
Intel Core 2 Duo E8400 Q807A191 (Will replace a second handed E8600 Q820A628 soon)
Cricial Ballistix PC3-1333Mhz CL6 2GB Kit D9GTR
Delta ServerPower 650W 84% Eff.
Colamax HD 4850 X2 PCS (one of them is offline right now)
Western Digital 5001ABYS x1, 5002ABYS (Dual I/O) x1

One of the weird cases:
I was able to get 500FSB rock solid stable for 7/24 use, but with the same setting, just set back to 400FSB, result in NO POST.

Findings:
Set FSB Strap and clock skew model at the same time will result in NO POST.
With 333 Strap max stable FSB can get was around 410, random freeze will occur if go further no matter how high the voltages set.
Using 266 strap 266/800 aka 2:3 divider, 333FSB POST successful but stuck at debug code 50.
Max stable FSB overclock so far was 520FSB, with Orthos Large FFTs fail after nearly 8 hours.
550FSB success POST but very unstable no matter what voltage or tweak settings used, possibly CPU issue.
This CPU can successfully POST at 580 FSB @Asus P5Q-Deluxe but no go on loading windows instant BSOD.
OC Fail Retry is not functioning. Always need manually clear CMOS.
For any adjustments that will need to alter DRAM Skew the system will do a full reset. E.g. FSB frequency, tCAS, tRD(not sure), DLL Default Skew Model, DRAM Divider.
DIMM slot 3 and 4 is rather difficult in controlling skew settings for a success boot. Still investigating.

Annoying part:
No current DRAM Timings displayed in BIOS. Auto will set to optimal setting rather than SPD. By using 200/800 Divider, tRD was set to 4 and tCAS timing was set to 6-6-6-15, with tCAS set to 7 or above will result in NO POST.

Will add and post more if anything new is found.

Welcome to the forums Stanley

Sounds like you may have a corrupt CMOS - if that is the case, just flip the battery out for an hour or so, PSU unplugged...more than an hour is sometimes needed.

Another thing that can cause this kind of weirdness is using settings from CMOS reloaded that were saved using a previous BIOS - this was especially bad on the T3RS.

Another thing to check is that you don't have any bad memory. The behaviour you describe can also happen if you have a dead or dying stick - memtest86+ test 5 for an hour on each stick, make sure you cool the sticks while testing.