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

[carmatic]

Here are my settings for 440x8 266/667 tRD6, 24/7 stable:

CPU Feature
- Thermal Management Control: Enabled
- PPM(EIST) Mode: Enabled
- Limit CPUID MaxVal: Disabled
- CIE Function: Auto
- Excecute Disable Bit: Enabled
- Virtualization Technology: Enabled
- Core Multi-Processing: Enabled

Exit Setup Shutdown: Mode 2
Shutdown after AC Loss: Disabled
CLOCK VCO divider: Auto
CPU Clock Ratio Unlock: Enabled
CPU Clock Ratio: 8
- Target CPU Clock:
CPU Clock: 440
Boot Up Clock:
DRAM Speed: 266/667
- Target DRAM Speed:
PCIE Clock: 100mhz
PCIE Slot Config: 1X 1X

CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
SATA Spread Spectrum: Disabled

Voltage Settings
CPU VID Control: Auto (1.2325)
CPU VID Special Add Limit: Enabled
CPU VID Special Add: Auto
DRAM Voltage Control: 2.0V
SB Core/CPU PLL Voltage: 1.510V
NB Core Voltage: 1.504V
CPU VTT Voltage: 1.210V
Vcore Droop Control: Enabled
Clockgen Voltage Control: 3.45V
GTL+ Buffers Strength: Strong
Host Slew Rate: Weak
MCH RON Offset Value: 00
MCH RTT Offset Value: 00
MCH Slew Rate Offset Value: 00
MCH VREF 1 Value: 00
MCH VREF 2 Value: 00
MCH VREF 3 Value: 80
GTL REF Voltage Control: Enabled
x CPU GTL1/3 REF Volt: 85
x CPU GTL 0/2 REF Volt: 83
x North Bridge GTL REF Volt: 96

DRAM Timing
- Enhance Data transmitting: Fast
- Enhance Addressing: Fast
- T2 Dispatch: Enabled

Clock Setting Fine Delay
DRAM CLK Driving Strength: Level 6
DRAM Data Driving Strength: Level 8
Ch1 DLL Default Skew Model: Mode 0
Ch2 DLL Default Skew Model: Mode 0

Fine Delay Step Degree: 5ps

Ch1 Clock Crossing Setting: Auto
- DIMM 1 Clock fine delay: Current
- DIMM 2 Clock fine delay: Current
- Ch 1 Command fine delay: Current
- Ch 1 Control fine delay: Current


Ch2 Clock Crossing Setting: Auto
- DIMM 3 Clock fine delay: Current
- DIMM 4 Clock fine delay: Current
- Ch 2 Command fine delay: Current
- Ch 2 Control fine delay: Current

Ch1Ch2 CommonClock Setting: Auto

Ch1 RDCAS GNT-Chip Delay: Auto
Ch1 WRCAS GNT-Chip Delay: Auto
Ch1 Command to CS Delay: Auto

Ch2 RDCAS GNT-Chip Delay: Auto
Ch2 WRCAS GNT-Chip Delay: Auto
Ch2 Command to CS Delay: Auto

CAS Latency Time (tCL):5
RAS# to CAS# Delay (tRCD):5
RAS# Precharge (tRP):5
Precharge Delay (tRAS):15
All Precharge to Act: 6
REF to ACT Delay (tRFC): 54
Performance LVL (Read Delay) (tRD): 6

Read delay phase adjust: Enter

Ch1 Read delay phase (4~0)
- Channel 1 Phase 0 Pull-In: Auto
- Channel 1 Phase 1 Pull-In: Auto
- Channel 1 Phase 2 Pull-In: Auto
- Channel 1 Phase 3 Pull-In: Auto
- Channel 1 Phase 4 Pull-In: Auto

Ch2 Read delay phase (4~0)
- Channel 2 Phase 0 Pull-In: Auto
- Channel 2 Phase 1 Pull-In: Auto
- Channel 2 Phase 2 Pull-In: Auto
- Channel 2 Phase 3 Pull-In: Auto
- Channel 2 Phase 4 Pull-In: Auto

MCH ODT Latency: Auto
Write to PRE Delay (tWR): 14
Rank Write to Read (tWTR): 11
ACT to ACT Delay (tRRD): 3
Read to Write Delay (tRDWR): 8
Ranks Write to Write (tWRWR): 6
Ranks Read to Read (tRDRD): 6
Ranks Write to Read (tWRRD): 5
Read CAS# Precharge (tRTP): 3
ALL PRE to Refresh: 6

Raising CPU PLL may lead to a very quick CPU degradation. Also keep your VTT < 1.35V for 24/7.

thanks for sharing these settings, what ram do you have? my pair of 4gb ocz reaperx pc8000 doesnt even boot at the next higher divider which makes it go at 1110mhz... i bought mine at the beginning of the year, ive read that the new versions have different ICs on them and from what i can tell they go much faster than the sticks i have

i will lower the sb voltage and see if it will still keep priming stable... are 45nm's more susceptible to damage from sb core voltage than 65nm's?

now that i know the gtl and vtt combination for my q9450, i think i can try to start lowering the vtt and bring the gtl settings down one by one... before this it was majorly unstable and any gtl settings ive tried only made it crash quicker or slower... but for now i consider my job to be done, that is to get my computer running at the speed it used to run during the winter, like the time before i had to take the q6600 out because its overheating, wont even encode mp3's without crashing, and now i am running prime at 55 degrees load instead of 75 degrees load

and one last question, maybe im just abit out of touch but i thought c1e and eist made overclocking impossible?

[zsamz_]

thanks for sharing these settings, what ram do you have? my pair of 4gb ocz reaperx pc8000 doesnt even boot at the next higher divider which makes it go at 1110mhz... i bought mine at the beginning of the year, ive read that the new versions have different ICs on them and from what i can tell they go much faster than the sticks i have

i will lower the sb voltage and see if it will still keep priming stable... are 45nm's more susceptible to damage from sb core voltage than 65nm's?

now that i know the gtl and vtt combination for my q9450, i think i can try to start lowering the vtt and bring the gtl settings down one by one... before this it was majorly unstable and any gtl settings ive tried only made it crash quicker or slower... but for now i consider my job to be done, that is to get my computer running at the speed it used to run during the winter, like the time before i had to take the q6600 out because its overheating, wont even encode mp3's without crashing, and now i am running prime at 55 degrees load instead of 75 degrees load

and one last question, maybe im just abit out of touch but i thought c1e and eist made overclocking impossible?

btw thats 714 bios i think
just clock your sticks slower

[carmatic]

btw thats 714 bios i think
just clock your sticks slower

do you mean you have the 714 bios? im not sure which bios version i have
'clock my sticks lower' , that means use a lower divider? or lower my fsb?

also, i think i have found a good way of dealing with the bowing of the board from not using a backplate when you have the cpu heatsink or waterblock on too tight... i had some silicon fan push-through mount things i got from one of the 'silent fans' ive bought and i stuffed them right under the edge of the motherboard where it was bowing the most, straight up from where the cpu socket is, and the pressure from these things is enough to push the board straight... i also stuffed a couple under where the pwm area is so that the heatsink over there will have more pressure with the pwm chips, because the bowing from the pressure of the cpu mounting can bend the pwm chips away from the heatsink

::edit:: dang, the first core failed prime after 1 hour with my southbridge voltage at 1.54v...now trying with a lower vtt and vcore setting, maybe my gtls were abit too low to be stable at the voltage i was using before this, and maybe a lower vtt would match up with them nicely... setting the host slew rate to 'strong' really did make a difference, wouldnt have known that if it werent for you guys

[zsamz_]

do you mean you have the 714 bios? im not sure which bios version i have
'clock my sticks lower' , that means use a lower divider? or lower my fsb?

also, i think i have found a good way of dealing with the bowing of the board from not using a backplate when you have the cpu heatsink or waterblock on too tight... i had some silicon fan push-through mount things i got from one of the 'silent fans' ive bought and i stuffed them right under the edge of the motherboard where it was bowing the most, straight up from where the cpu socket is, and the pressure from these things is enough to push the board straight... i also stuffed a couple under where the pwm area is so that the heatsink over there will have more pressure with the pwm chips, because the bowing from the pressure of the cpu mounting can bend the pwm chips away from the heatsink

::edit:: dang, the first core failed prime after 1 hour with my southbridge voltage at 1.54v...now trying with a lower vtt and vcore setting, maybe my gtls were abit too low to be stable at the voltage i was using before this, and maybe a lower vtt would match up with them nicely... setting the host slew rate to 'strong' really did make a difference, wouldnt have known that if it werent for you guys

i will try the 714 after i get my e8600
the bios setting wizard gave you are from bios 714
you should try 603 its verry stable

[merlin76]

BIOS versions 0714 and 0725 for me are alpha versions and I don't really understand how DFI could make 0725 a final one. Take a look over here: http://www.csd.dficlub.org/forum/showthread.php?t=7502
Perhaps they needed a final BIOS quickly due to the problems with 4870 X2 cards.

@ carmatic

BIOS 0603 works best for me and the settings I posted work with BIOS 0603, 0714 and 0725.

If your RAM isn't capable of doing DDR2-1100 then switch to 333/800 divider and set tRD to 7 or 8. You cannot take settings 1:1 but it should give you a hint what is needed.

[zsamz_]

BIOS versions 0714 and 0725 for me are alpha versions and I don't really understand how DFI could make 0725 a final one. Take a look over here: http://www.csd.dficlub.org/forum/showthread.php?t=7502
Perhaps they needed a final BIOS quickly due to the problems with 4870 X2 cards.

@ carmatic

BIOS 0603 works best for me and the settings I posted work with BIOS 0603, 0714 and 0725.

If your RAM isn't capable of doing DDR2-1100 then switch to 333/800 divider and set tRD to 7 or 8. You cannot take settings 1:1 but it should give you a hint what is needed.

same here merlin 603 is best for me2
n yes 333/800 divider is best

[jcool]

Guys you think the board will like 4x1GB of Cellshock 1000C4? Finest D9GKX, couldn't resist to order at the current price
I'd like to run 1100 C4 with reaonable volts, yummie

[zsamz_]

Guys you think the board will like 4x1GB of Cellshock 1000C4? Finest D9GKX, couldn't resist to order at the current price
I'd like to run 1100 C4 with reaonable volts, yummie

mine not too crazy for d9gkx 4x1 but loves d9gmh 4x1 i think it might like cellshock better
you should be able to do 1120mhz cas 4 with 2.24v-2.3v

[jcool]

Well I already got 1 pair of Cellshock D9GMH and ordered a 2nd pair too, so just in case it doesn't like GKX I now got 2 pairs of 800C4 with GMH
Had to buy some cause the Crucial Value with D9GMH I used to run along with the Cellshocks is faulty now, after almost 2 yers. Oh well

[merlin76]

@ jcool

You'll need to flash BIOS 0714 or 0725 and set proper DS/DDS and DLL mode to make them clock well on this board:

DS/DDS 6/8
DLL Mode 5/5

Take a look over here: http://csd.dficlub.org/forum/showpos...9&postcount=17

[C-N]

Guys you think the board will like 4x1GB of Cellshock 1000C4? Finest D9GKX, couldn't resist to order at the current price
I'd like to run 1100 C4 with reaonable volts, yummie

Praz has had a bit of a breakthrough recently with his Cellshoch GKX & the x48 board see from post 1959 on in the x48 thread.

http://www.xtremesystems.org/forums/...postcount=1959

CN

[carmatic]

BIOS versions 0714 and 0725 for me are alpha versions and I don't really understand how DFI could make 0725 a final one. Take a look over here: http://www.csd.dficlub.org/forum/showthread.php?t=7502
Perhaps they needed a final BIOS quickly due to the problems with 4870 X2 cards.

@ carmatic

BIOS 0603 works best for me and the settings I posted work with BIOS 0603, 0714 and 0725.

hmm i think i'll try 0603 soon
my vtt doesnt want to come down it stays at 1.35v , not very sexy at all

[Slump]

how can i connect the Optical S/PDIF Connector to my HT system ( Samsung HT-QX 100 ) ? Please ! THANKS

[carmatic]

how can i connect the Optical S/PDIF Connector to my HT system ( Samsung HT-QX 100 ) ? Please ! THANKS

this is the bernstein audio module correct? http://www.clunk.org.uk/forums/revie....html#post7401 it says here that there is no optical out support, you will need to get one of those coaxial --> optical adaptors

::edit:: like one of these http://www.tvcables.co.uk/cgi-bin/tv...ml?id=9YWPyq7M

[Praz]

how can i connect the Optical S/PDIF Connector to my HT system ( Samsung HT-QX 100 ) ? Please ! THANKS

The module does have optical in/out but would necessitate wiring it yourself. As already mentioned it would be easier to use an optical to coaxial converter.

[carmatic]

i think that im getting better at overclocking my computer... like, maybe there is a good analogy for the pre-gtl days was that you could choose any voltage you want , and the board will 'choose the best gtl setting for you', even tho maybe theres no such thing as gtl back in the amd64 and pentium4 days...
now you start by choosing your voltage just like you chose your fsb and multiplier, then keep trying different gtl settings until you go through a sweet spot (this takes days of priming), then change the voltage one notch and repeat the gtl sweep again, the new sweet spot should be within +/-5 gtl points of the sweet spot of the previous voltage... hopefully when you begin its at least stable enough to boot into windows and let you run prime, then as you start varying the gtl settings you should go from instant lockups when you click on 'ok', to lockups which happen after a few seconds, then a few minutes, then you should start getting prime errors instead of system locks and they should take longer and longer to show up, after the sweet spot has passed the prime errors would start showing up more frequently again so you can get a 'feel' for which gtl settings are good

the gtl doesnt just depend on the vtt voltage, it also depends on vcore and cpu pll voltage as well, this is what ive found over the last 3 days of 'gambling' on my computer... as most of you know, lowering vtt also means lowering gtl, but i also have a feeling that lowering vcore also means lowering gtl but to a lower extent, but all i know so far is that changing the cpu pll voltage means the gtl settings i was using isnt valid anymore
anyone have any comments on this?

so far, this is how ive been working... system lock within seconds of beginning prime = major change in gtl , 5 points up or down ... system lock within minutes of beginning prime = 2 points up or down ... when the system no longer locks and runs stable enough for prime to report problems, single core failure within seconds= further change by 2 gtl points, single core failure after a few minutes = further change by 1 gtl point ... both gtl's go up and down together, on my computer it happens that they both like to be on the same value, but i say that only start using 2 different gtl settings when you definitely notice a pattern of 1st/4th or 2nd/3rd core prime failures, if there is no pattern then change the northbridge gtl and see if that helps, also if you have the 1st and 3rd or 2nd and 4th cores failing at the same time within minutes of starting prime its also likely a northbridge gtl thing, if it does not help change it back and try a different vtt or vcore
because vcore can be changed in finer increments than vtt, it will be necessary to start adjusting (probably increasing) the vcore to get more stability after adjusting the gtl's, when you have swept through the vcore sweet spot and it still wont prime stable then its time to start over again with a higher vtt setting

[carmatic]

heres a little chronicle of my journey into priming...
cpu vid at 1.25625v, cpu vid special add at 103.30%, vtt at 1.345 .... core gtl settings at 94 , prime stable for half an hour ... gtl settings at 95, prime stable for just over 10 minutes, and the same happens 93, core temps maxing out at 57 degrees, mostly at 55 degrees, cpu temperature diode at 39 degrees
cpu vid at 1.25625v, cpu vid special add at 103.56%, vtt at 1.345... core gtl settings also at 94, prime stable for 40 minutes, core temperatures didnt change, cpu temperature diode at 40 degrees
cpu vid at 1.25625v, cpu vid special add at 103.56%, vtt at 1.340... core gtl settings to be stable are at 89, prime stable for an hour... at 90 it crashes within 5 minutes, at 88 its stable for 15 minutes or so... temperatures maxing out at 56 degrees, mostly at 53 degrees, cpu temperature at 37 degrees... changing northbridge gtl from 112 to 113 results in prime failure within 10 minutes, and at 111 it crashes within 4 minutes
cpu vid at 1.25625v, cpu vid special add at 103.56%, vtt at 1.335... core gtl settings at 84 , still priming now fingers crossed, at 85 it failed at 40 minutes... temperatures maxing out at 53 degrees, mostly at 50 degrees... cpu temperature diode at 35 degrees

running much cooler at 1.335 vtt, if only i had a temperature logger because im not sure if im remembering these temperatures correctly...

[zsamz_]

got my e8600 had to use special add% or no boot
need 1.38v prime stable 63%gtl
everest score seams a bit low i think i gotta take out my ut p35

[carmatic]

cpu vid at 1.25625v, cpu vid special add at 103.56%, vtt at 1.335, 84/84/112 gtl settings prime stable for 1 hour 8 minutes... they always seem to last only an hour!
now trying with cpu vid at 1.25625v, cpu vid special add at 103.73% or whatever the next one up is... is it just me or does the load temps go down the closer your vcore is to your vtt? like increasing vcore actually makes the temps go down?

got my e8600 had to use special add% or no boot
need 1.38v prime stable 63%gtl

is taht 1.38 vcore or 1.38 vtt? that sure feels kind of high for a 45nm dual core...
could it be that there is an issue between flashing your x38 with the x48 bios which makes the 8600 not boot without adding voltage??

::edit:: damn it, prime failed at 1 hour 12 minutes... its happening alot at around this length of time, like just over one hour... now im trying with the vid special add at the next notch up, see if it still fails at that time

[kasu code]

does anyone have the same problems with the 4870x2 that they did with the 3870x2?

GRRRR... I bought two 4870X2s cant get either to post at all. Put my GTX280 back in and it works fine. Someone at Rage3D was having the same problem on a DFI X38board. Think flashing my Bios to X48 would fix this or no?

[RGone]

GRRRR... I bought two 4870X2s cant get either to post at all. Put my GTX280 back in and it works fine. Someone at Rage3D was having the same problem on a DFI X38board. Think flashing my Bios to X48 would fix this or no?

It is the weekend in Taiwan but I know that DFI is working on the issue today. RGone...

[carmatic]

damnit, 1 hour 20 minutes!! why does it always happen around this time

[Mysterfix]

damnit, 1 hour 20 minutes!! why does it always happen around this time

I think your chip is just starving for voltage. Try lowering your overclock by a few mhz (20-50) and see how much further you get with prime or raise your voltage by .01-.02v and see how far that gets you. Or keep messing with the GTL settings and hope to get lucky.

[kasu code]

It is the weekend in Taiwan but I know that DFI is working on the issue today. RGone...

Good to hear. Ill give it a week before I get a new board.

[nfm]

@carmatic
bro, why is your NB GTL so high? Try 84 83 84 GTLs with your current VTT at 1.335v and lower that vNB to around ~1.4v. vPLL seems to high, and Host Slew rate should be set to weak unless you disable the GTL Ref. Droop Control enabled?