Official DFI LanParty UT X38-T2R(and LT) Discussion/Review/Overclock/Guide Thread

[lazy]

At this point I don't have a recommendation. Haven't had time to explore GTL values yet. I don't known at this point if the settings can even be carried over from the P35. Will probably get into it this weekend if I switch over to water. With air cooling I can't get to the MOSFETs to measure the voltages

alright then i look forward to your upcoming findings

[Brother Esau]

CellShock 8000 on this board acts the same as on the P35. After messing with the SPD it seems like a lost cause. I'm almost willing to bet money it's a PCB issue.

I feel the same way John Even in the AMD 790FX-M2R it requires insane Board and ram Voltage to Overclock the sticks in conjunction with the Processor

I have notified Robert and Sascha of my Findings with the Memory in the New DFI LanParty UT 790FX-M2R already and Sascha will be pushing to get some Samples for D.F.I to start testing with their Boards thats if MSC will listen

I have even Flashed to a different SPD.File that Sascha has given me with mo EPP Profile and its the same deal.

[SNiiPE_DoGG]

praz: do you have a picture of where I can measure GTL volts (contact points)? I want to make my own measurements.

[Praz]

I have even Flashed to a different SPD.File that Sascha has given me with mo EPP Profile and its the same deal.

SPD isn't going to help. That's why I'm leaning toward the PCB being the cause.

praz: do you have a picture of where I can measure GTL volts (contact points)? I want to make my own measurements.

It will be the weekend before I can get pictures. Look at the pics I have of the P35. The trace routing and MOSFET location is slightly different but close enough that you will be able to find them on the X38. To do the measurements you need to use a meter with 4 digit accuracy.

[Brother Esau]

Yah......If we could only get Cell Shock on the same page with D.F.I as OCZ Memory is that would rock

[fireice2]

Hope to be able to get one of these babies soon. Not enough minerals to get one to the Philippines though... shipping will cost me an arm and leg...

[OnDborder]

Ok, I got mine up and running.
Running a;
e6850 atm
2x1 crucial 6400
1 3870
Silverstone zeus 750
Tuniq tower
No os atm

My question is, when ever I adjust the vcore, I get a start up loop but no boot up. It's a good thing this isn't in a case,, yet.

As soon as I get this going, I'll throw on an os and my other 3870

[lazy]

its funny how everything you have mentioned was exactly what i was running in my setup before i changed it to a go

btw what can i use to cool the pwm area?

[OnDborder]

I have a fan blowing on the area.. It's not in the case. But when it is, it will be cooled off by a 60m fan attached to one of the waterblock posts/bolt. I'll be putting a q6600 in later.

[SNiiPE_DoGG]

the pwm gets much hotter when you add a q6600. i will be putting a 60mm on the outside zip tied to the I/O panel when i get it in the mail, hopefully it will help move all that hot air out.

[OnDborder]

My question is, when ever I adjust the vcore, I get a start up loop but no boot up. It's a good thing this isn't in a case,, yet.

As soon as I get this going, I'll throw on an os and my other 3870

Another "Duh" moment ..
0.44375 in the bios means 0.44375.. I thought it was like the last dfi I had.Where you add it on to the vcore.

[zanzabar]

u have to do the gtl when u adjust it

[pumbertot]

i shall ask again for general opinions, worth selling my DFI P35 LP for this board or nothing really to gain?not running crossfire.

[sofarfrome]

i shall ask again for general opinions, worth selling my DFI P35 LP for this board or nothing really to gain?not running crossfire.

I have had this board up and running now for 3 days with a WC Q6600 G0 and an X3210 G0. If I was going to give you advice I would say not to jump to it yet. So far this board will not OC a quad (except for the QX9650) any better than any of the other 4 boards I have tried. Yes, it is a tinkerer's delight but it still isn't in a class by itself when it comes to the Quad. It may be after a few bios updates but for right now I would have been just as happy with the P35 version for $90 less. Some are jumping on the Foxconn Mars bandwagon cuz someone has achieved 515FSB with a quad. My guess is that is a brute force situation where a ton of vcore was still needed to get that high. But then again, at least the board got that high. If you take a look around no one is talking 500FSB with a quad, at least not for 24/7 use. But everyone has a 3.6GHz OC with varying levels of vcore. This is so common that I would be quite disappointed with anything less. I can get both of my chips to the 3.7-3.8GHz range with reasonable vcore. I have yet to get either into windows at 4GHz no matter how much vcore I give it. There certainly could be something on the board I am not seeing causing this so I will need to do some more tinkering but I think right now it is time for a new chip or wait for the Q9550.

I have not tried a C2D yet (I no longer have one available) so I can't speak to its ability with those CPUs. But I am of the impression that 4GHz should be achieved with 1.5vcore or less to call a chip a good OCer. Either I do not have a chip capable of that or none of the boards are capable.

Also, I have not yet tried to OC the memory too much. I am running everything 1:1 so until I max out the CPU I won't bother with the memory.

[Praz]

If you take a look around no one is talking 500FSB with a quad, at least not for 24/7 use.

That's because finding a quad that will do 500+ FSB is as rare as a pink diamond. Most dual-cores that run run above 520 FSB are done with modded boards or CPU killing PLL voltages.

[sofarfrome]

That's because finding a quad that will do 500+ FSB is as rare as a pink diamond. Most dual-cores that run run above 520 FSB are done with modded boards or CPU killing PLL voltages.

I think that is why I am going to hold off on buying any more C2Qs and wait for the 45nm stuff to come out. (I can't justify $1K on a QX9650).

[Praz]

The 9550 should allow most to hit at least 4.0. I think the sweet spot for 24/7 use will be the 8500 though.

[sofarfrome]

Yeah, I will be torn between dual or quad core for sure. I do a lot of audio and video editing so the quad is nice but I am not on a timeline to get stuff done so having a smoking dual core is fine. I should have kept my L629B Xeon3060 just to play with. That thing was an easy 4GHz. Oh well...

[pumbertot]

thanks for that mate. ive got a QX9650 days away(on a plane from Canada).
but now im torn between this or giving 780i a go. or maybe i will get both,lol.

the addiction just cant be stopped....................................

I have had this board up and running now for 3 days with a WC Q6600 G0 and an X3210 G0. If I was going to give you advice I would say not to jump to it yet. So far this board will not OC a quad (except for the QX9650) any better than any of the other 4 boards I have tried. Yes, it is a tinkerer's delight but it still isn't in a class by itself when it comes to the Quad. It may be after a few bios updates but for right now I would have been just as happy with the P35 version for $90 less. Some are jumping on the Foxconn Mars bandwagon cuz someone has achieved 515FSB with a quad. My guess is that is a brute force situation where a ton of vcore was still needed to get that high. But then again, at least the board got that high. If you take a look around no one is talking 500FSB with a quad, at least not for 24/7 use. But everyone has a 3.6GHz OC with varying levels of vcore. This is so common that I would be quite disappointed with anything less. I can get both of my chips to the 3.7-3.8GHz range with reasonable vcore. I have yet to get either into windows at 4GHz no matter how much vcore I give it. There certainly could be something on the board I am not seeing causing this so I will need to do some more tinkering but I think right now it is time for a new chip or wait for the Q9550.

I have not tried a C2D yet (I no longer have one available) so I can't speak to its ability with those CPUs. But I am of the impression that 4GHz should be achieved with 1.5vcore or less to call a chip a good OCer. Either I do not have a chip capable of that or none of the boards are capable.

Also, I have not yet tried to OC the memory too much. I am running everything 1:1 so until I max out the CPU I won't bother with the memory.

[ic3m4n2005]

OK guys, since this is my last day at work for this year
i have time to test out my E2180 with this board.
A short test got me to the following situation, 3050 stable at stock voltages (1,232) but 3100 fails prime instantly.
Settings should be useable for 24/7 use. I'am on watercooling so where should i start ? Need some hints for board voltages, cpu vtt and maybe GTL voltages ( i think you can leave them at "disable" when using dual core, or am i wrong ?).

[SNiiPE_DoGG]

OK guys, since this is my last day at work for this year
i have time to test out my E2180 with this board.
A short test got me to the following situation, 3050 stable at stock voltages (1,232) but 3100 fails prime instantly.
Settings should be useable for 24/7 use. I'am on watercooling so where should i start ? Need some hints for board voltages, cpu vtt and maybe GTL voltages ( i think you can leave them at "disable" when using dual core, or am i wrong ?).

set the vtt to 1.6 for nice stable clocks, nb needs volts in general ie. if you are looking for more stability increase that first. then if the nb and cpu vtt isnt helping pop the cpu up a notch. thats my experience with the oc process on this board.

the general procedure for GTL is to find a stable clock (via cpu voltage) with gtl disabled, then lset the GTL to 67 percent using praz's p35 charts (I already linked a few pages back). once you do that you can fiddle with lowering your cpu volts and still have stability.

i learned most of this stuff (i hope its correct) through the UT P35 thread, so i would advise you check that out for more tips.

[Ace-a-Rue]

i shall ask again for general opinions, worth selling my DFI P35 LP for this board or nothing really to gain?not running crossfire.

I have had this board up and running now for 3 days with a WC Q6600 G0 and an X3210 G0. If I was going to give you advice I would say not to jump to it yet. So far this board will not OC a quad (except for the QX9650) any better than any of the other 4 boards I have tried. Yes, it is a tinkerer's delight but it still isn't in a class by itself when it comes to the Quad. It may be after a few bios updates but for right now I would have been just as happy with the P35 version for $90 less. Some are jumping on the Foxconn Mars bandwagon cuz someone has achieved 515FSB with a quad. My guess is that is a brute force situation where a ton of vcore was still needed to get that high. But then again, at least the board got that high. If you take a look around no one is talking 500FSB with a quad, at least not for 24/7 use. But everyone has a 3.6GHz OC with varying levels of vcore. This is so common that I would be quite disappointed with anything less. I can get both of my chips to the 3.7-3.8GHz range with reasonable vcore. I have yet to get either into windows at 4GHz no matter how much vcore I give it. There certainly could be something on the board I am not seeing causing this so I will need to do some more tinkering but I think right now it is time for a new chip or wait for the Q9550.

I have not tried a C2D yet (I no longer have one available) so I can't speak to its ability with those CPUs. But I am of the impression that 4GHz should be achieved with 1.5vcore or less to call a chip a good OCer. Either I do not have a chip capable of that or none of the boards are capable.

Also, I have not yet tried to OC the memory too much. I am running everything 1:1 so until I max out the CPU I won't bother with the memory.

i wholeheartedly agree with sofarfrome!...TRUE, it is a tweaker's board but you are really not going to do any better going from a P35 board to an X38 board...who knows, Foxconn could have cherry picked their boards to send out to the "big guns" in overclocking...i respect their results but for most of us, we won't achieve that, especially for a Quad cpu!

here's the Pro's for this board:

1) PWM HS and 8 phase digital power are excellent
2) love the cmos store banks (four)
3) tweak city on bios memory settings and voltages...BUT...that comes with a HUGE learning curve
4) hardly any vcore droop (0.02)
5) Areca raid card can work from PCIe slot #1 and VC in slot #2, and get full bandwidth 16x or in my case, 8x for my Areca card.
6) 6 USB ports on the rear I/O panel
7) like the added yellow bank with 2 SATA ports that run off the JMicron SATA/PATA controller...more flexibility.

CON'S:

1) can not mount oversized air cpu heatsink, like the TRUE, to face rear of case because of the oversized NB heatsink.
2) not enough fan headers to control fan speed...CPU fan header won't control fan speed with a 3 pin plug fan...had to move plug to NB header....the IP35 Pro puts this board to shame when it comes to fan headers and speed control.
3) smart guardian is just about useless...ok, maybe voltage readouts is good...compare that prog to Abit's "Uguru" technology and you will understand RIGHT away!
4) missing a rear mounted CMOS reset switch which is on Abit's Pro board and other Abit top line boards...i guess the 'Hotkey' is the answer to that but it seems arduous to get it done that way.
5) harder to overclock as you reach top limit of memory or cpu...the IP35 Pro is a breeze in comparison.

i will add more as i remember them

[Praz]

It is a tweaker's board. That is both it's strength and weakness. Once the components are pushed to the edge further gains become increasingly more involved. But with the settings at one's disposal usually the results will exceed other boards. If this is worthwhile for 24/7 use is a decision each user will have to make. But as Eva's memory results have shown, use the settings that most leave set on Auto and the results can be worthwhile.

The PWM fan speed issue is something I don't know if there is a feasible solution to or not. The controller from ITE supports PWM only. The other headers are converted by additional circuitry to function as 3 pin speed control. Maybe the same circuit with a digital switch activated be a BIOS setting? Will have to ask Oskar about that.

Personally I prefer pressing the Insert key to clear the BIOS rather then reaching behind the case. But I use a PS2 keyboard too. USB results can be iffy.

[eva2000]

Yeah clock fine delay is one powerful option for memory clocking! Works better and more predictable on DFI X38 than DFI P35 though

Weird thing is, even if modules are in dimm slot 2+4 instead of 1+3, the most beneficial clock fine delay setting to adjust is still for dimm 1 + 3 clock fine delay options (regardless where modules are in dimm slots) !

[Praz]

Weird thing is, even if modules are in dimm slot 2+4 instead of 1+3, the most beneficial clock fine delay setting to adjust is still for dimm 1 + 3 clock fine delay options (regardless where modules are in dimm slots) !

At least I know it's not just me.Trying to figure that one out myself.