**Official DFI LanParty UT X48-T2R(and LT) Discussion

[kaskuli]

Bad news for Q9450 overclockers. From the Anandtech review:

"The other positive is that users who don't want to go for elusive levels of FSB will find that the board will run fine without any need of tuning GTL levels right up to 485FSB for dual-core 45nm processors and 420FSB for quad-core 45nm processors. The bitter aspect is that boards from ASUS such as the Rampage Formula are in direct competition to DFI's offering in this segment and can run ultra high FSB speeds and lower tRD levels with relative ease."

DFi not OC topdogs anymore. Probably all gonna go downhill from here... Delay like hell to bring out the mobo then you find it's performance falls way short of expectations... Sucks though, layout etc is top notch

[wilson]

Can any1 tell me the best way to flash my bios pls as ive never done it on a DFI . So is winflash or diamond flash the way to go ??
Cheers all...

[PnoT]

What kind of beep?

Try with one stick of RAM.

Also, do a battery out CMOS clear, it can often fix problems like that

I reset the CMOS and could get it to boot normally again but nothing i have done so far will get me booting at 500FSB. I selected another strap so my ram was running around 1000'ish, tweaked up the nb volts to 1.6, and added a little more cpu vid. I'll have to play around with it again tonight.

Can any1 tell me the best way to flash my bios pls as ive never done it on a DFI . So is winflash or diamond flash the way to go ??
Cheers all...

This is the method i use and it's a no brainer... http://csd.dficlub.org/forum/showthread.php?t=1136


Is there any word on that new BIOS yet?

[Tony]

Guys I feel I need to jump in here.

X48 is no P35 for when it comes to clocking memory (DDR2) on the whole, I have tested lttest.bin (new bios) and it does help with higher density modules BUT its no P35 for ease of getting the clocks you want.

Even with the re-spin for X48 I still feel its a DDR3 chipset, and a dual core CPU chipset also as the load quads place on the NB and FSB just ruins how these boards clock.

A prime example is 1600EB 4GB kits, they clock easy with dual cores, put a quad in the board and the issues start. Same with 4GB ddr2 kits, start clocking over 1050 or so and the issues start...especially with a quad core CPU.

Oskar has worked some wonders with the new bios you will see soon, I did hit 1200 with flex2 using it where I can not even do 1100 in rampage formula, but it was not as straight forward as it was in the LP35 LT or the Foxconn MARS I tested.

X48 has limits, try to remember this.

[karateo]

@Tony

I just bought two of this kit
http://www.ocztechnology.com/product...pc_4gb_edition

I hope i don't have any problems to run them at least at stock settings (1066)
currently using my E8400 at stock too.

should I up my nb voltage before installing them?
any advice/pointing me to more info would be perfect.

also,
i want to thank you for all these articles you have written (since P5b dlx)
as they have helped me a lot.


great work

[Tony]

you need the new bios which should be up soon, Oskar tuned the board to help overclocking PSC IC's like the ones we use on the 1066 reapers you quoted there.

I will try to post guides for these and the 1150 flex2 over at the ocz forum to help those looking for settings...BUT remember, im not interested in CPU overclocking so those who moan that they can not hit the CPU clocks with the ram where they want it will be asked to test with CPU's (quads especially) at near stock speeds due to the nature of the chipset.

[karateo]

im not interested in CPU overclocking so those who moan that they can not hit the CPU clocks with the ram where they want it will be asked to test with CPU's (quads especially) at near stock speeds due to the nature of the chipset.

totally agree with that
you are not responsible with motherboard/cpu capabilities

i suppose higher nb voltage and/or relaxed tRD will help for a nice 450 fsb and memory close to stock -> 1080mhz

so i need the newer bios to run even at stock settings everything?
i have exams next month and I don't want to risk a bios update (even dfi has never turned my down on that part)

[eva2000]

Guys I feel I need to jump in here.

X48 is no P35 for when it comes to clocking memory (DDR2) on the whole, I have tested lttest.bin (new bios) and it does help with higher density modules BUT its no P35 for ease of getting the clocks you want.

Even with the re-spin for X48 I still feel its a DDR3 chipset, and a dual core CPU chipset also as the load quads place on the NB and FSB just ruins how these boards clock.

A prime example is 1600EB 4GB kits, they clock easy with dual cores, put a quad in the board and the issues start. Same with 4GB ddr2 kits, start clocking over 1050 or so and the issues start...especially with a quad core CPU.

Oskar has worked some wonders with the new bios you will see soon, I did hit 1200 with flex2 using it where I can not even do 1100 in rampage formula, but it was not as straight forward as it was in the LP35 LT or the Foxconn MARS I tested.

X48 has limits, try to remember this.

how's that compared to DFI LT X38-T2R ? closer to P35 or LT X48 ?

[Afterburner]

Thanx for the info Tony, much appreciated....

I have dumped my Quad last week and awaiting a 8500....

A quick question Tony, will the new BIOS help Micron D9 chips in anyway ? (I will keep 2 out of my 4 sticks aside as I have no need for 4GB now anyway)

[Aldy402]

anybody running this board with a 4gb gskill set?

[Big Lar]

anybody running this board with a 4gb gskill set?

I am, and answered your question somewhat in the memory section. I have a template a few pages back in the other thread that works nicely for me.

Larry

[PnoT]

anybody running this board with a 4gb gskill set?

I'm running a set of 2x2 8500's right now and I'm making small headway on getting things stable. I finally hit my 4ghz mark tonight and it'll run 4+ hours until it bombs out in blend mode so I'm tweaking the GTLs a bit.

Has anyone else noticed how far off Anandtech's GTL refs are from the ones that Clunk has posted up? If you do the math things just don't add up and not close enough to one another. I know these charts are "starting points" but i'd think they would be a bit closer.

Also the anandtech chart is green / yellow (they mislabeled the gtl refs but oh well). The green and yellow look to be the diff values of 61% and what it would be at 63% right? If that's so then why are the real vtt measurements off for the same bios vtt? Someone clear this up for me as it's very late heh.

[axis]

Any updates on what the new bios brings, besides a UT board ?

Axis

[zsamz_]

I'm running a set of 2x2 8500's right now and I'm making small headway on getting things stable. I finally hit my 4ghz mark tonight and it'll run 4+ hours until it bombs out in blend mode so I'm tweaking the GTLs a bit.

Has anyone else noticed how far off Anandtech's GTL refs are from the ones that Clunk has posted up? If you do the math things just don't add up and not close enough to one another. I know these charts are "starting points" but i'd think they would be a bit closer.

Also the anandtech chart is green / yellow (they mislabeled the gtl refs but oh well). The green and yellow look to be the diff values of 61% and what it would be at 63% right? If that's so then why are the real vtt measurements off for the same bios vtt? Someone clear this up for me as it's very late heh.

anantechs glts are for p35....

[zsamz_]

Thanx for the info Tony, much appreciated....

I have dumped my Quad last week and awaiting a 8500....

A quick question Tony, will the new BIOS help Micron D9 chips in anyway ? (I will keep 2 out of my 4 sticks aside as I have no need for 4GB now anyway)

i sure hope so lol
only my ballistix 8500s lovin this board all my other sets givin me trouble

[cstkl1]

i sure hope so lol
only my ballistix 8500s lovin this board all my other sets givin me trouble

hmm kekek
my teams seem to be like married to it..

oh yeah ocz reaper X pc2-8000 2x2gb so far no issue with it..

[Aldy402]

thanks Big Lar, Pnot

[Praz]

The GTLREF chart at Anandtech is for the X48 board. There are a couple of points of confusion relating to the values furnished by DFI. I'll try to post more on this in a bit. But the bottom line is for 45nm processors the values at Anandtech wil give the user a good starting point.

[Tony]

Im running the test bios on the X38, just a little more NB voltage needed on the X38

It was tuned to clock 2GB modules better, which it does...im revisiting the clock skews to see how much they help also as Raja is telling me he can see decent FSB increases by some tweaking.

The issue with bios files as open as this though is you are forced to tweak, a lot of this stuff IMO should be by auto as most end users just have no idea.

[motherboardpro]

New Price $329.99 on the DFI LanParty UT X48 T2R

[PnoT]

anantechs glts are for p35....

Why would they post p35 GTL refs in an x48 article... that makes no sense at all to me.

The GTLREF chart at Anandtech is for the X48 board. There are a couple of points of confusion relating to the values furnished by DFI. I'll try to post more on this in a bit. But the bottom line is for 45nm processors the values at Anandtech wil give the user a good starting point.

I'm eager to see what's going on with these values. Thanks Praz!

*EDIT* I found this over on the DFI forums and it would explain why the values are different in both clunks and anands charts.

http://www.csd.dficlub.org/forum/showthread.php?t=6538

[Aldy402]

i thought the gtl chart at anandtech was geared towards ppl using micron d9s
am I wrong?

[PnoT]

i thought the gtl chart at anandtech was geared towards ppl using micron d9s
am I wrong?

GTLs are basically for core stability when applying high FSB values and not RAM. That's a really basic breakdown but you get what i mean.

[Praz]

Two things I noticed about the VTT/GTL voltages provided by DFI. The charts specify 67% as a set-point for GTLREF. These values correspond to a 65nm CPU. Yet the chart starts at 1.10V. 1.10V is the default VTT voltage for 45nm processors.

The other anomaly is the actual voltages vs set VTT themselves. Two that stand out are 1.35(1.310) and 1.553(1.508). Try as I may, I cannot reproduce these results. Using a 45nm CPU and selecting 1.350V I get the same measured voltage of 1.339V as the chart at Anandtech. All other measured voltages compare similarly to the values shown in the chart at Anandtech.

I'm not even able to venture a guess as to why there is such a difference in these measured voltages. The only variable that comes to mind is the reference point chosen to measure the voltage. I can't speak as to the equipment other use to measure the voltages but I normally use either a 350MHz or 500MHz bandwidth scope. Regardless where I measure at on the board I don't see more then a couple of thousands of a volt difference between the various points of measurement. This being the case is seems unlikely that the difference in voltages could be attributed to the reference point selected.

This GTL stuff isn't rocket science as far as tuning in the BIOS. What makes it difficult for most users is there is no known starting point of values. Those that have no means of measuring the default values need to find a chart that works for them and fine tune from there. Users who do have the capability of measuring the voltages need to remember that consistency is the key. Pick a point from the various photos available and use only that point as a source of measurement. The accuracy of the measuring equipment that the average user has at their disposal and the loading effect that equipment has on the circuit is going to induce more of an error then where you choose to measure at. As are the board to board tolerances. The same applies to the point chosen for ground. And don't use the case or something like the video card bracket as a ground point.

[Aldy402]

GTLs are basically for core stability when applying high FSB values and not RAM. That's a really basic breakdown but you get what i mean.

I know but they still relate(ratios) I thought he geared the gtl table according to the ram he used in conjuction wit the fsb speed/ram freq ratio
guess not, thanks for clarifying