**Official DFI LanParty UT P35-T2R Review/Overclock/Guide Thread**

[C-N]

C1E and VDROOP enabled. VCORE was set using VID Special Add leaving VID Control on Auto. Adjusting VCORE this way allows the voltage to decrease along with frequency while idle.

Very interesting will this work with 65nm's too?

CN

[Praz]

Very interesting will this work with 65nm's too?

CN

It's worked on all processors I've used. Both 65nm and 45nm on the P35 and X38 boards. Definitely have to have a stable overclock though.

[mad_skills]

It's worked on all processors I've used. Both 65nm and 45nm on the P35 and X38 boards. Definitely have to have a stable overclock though.

Nice work Praz, tnx.

[C-N]

It's worked on all processors I've used. Both 65nm and 45nm on the P35 and X38 boards. Definitely have to have a stable overclock though.

Excellent, I'll give that a go later



CN

[Praz]

Setting the voltages up this way removes the only valid argument for disabling VDROOP. The system never sees the additional voltage at idle that results from VDROOP. And from an electronic point of view the benefits of VDROOP are realized. Most boards don't allow this capability. Once default VCORE is changed the voltage no longer scales with frequency.

[~aoe~]

My LT P35-T2R just arrived.

I'm still looking for an E8400 and, seeing as none of my regular suppliers have any in stock, this may take a couple of weeks. In the meantime, I'll take the time to read through this thread. I've already had a quick skim through and it's clear there's a lot of valuable info here, so a big thanks to all those who have contributed so far .

As I'm a 'DFI virgin', it's quite possible I may have some difficulties - just so you're all prepared .

[vortex240]

Praz, few questions if you can

Why have the vdrop enabled? I thought that makes the voltage less stable.

Also you wrote that using a straight 7 allows for the NB to run at the stock voltage. Whats the highest fsb this can work at, and would this apply to 65nm allandale e2180?

If so do I just leave all the phase settings to disable and set the performance level to 7? I asume this works for both the 266/677 and 333/800 straps.

thanks for the help
Wojtek

[Mav451]

Well he's left some of his power saving features on. I don't. I know with my OC, the vdroop (even tho small) is enough to add an element of instability. However, I am manually setting my vcore.

[Praz]

[QUOTE=vortex240;2799021]Praz, few questions if you can

Why have the vdrop enabled? I thought that makes the voltage less stable.

Having VDROOP for a 24/7 configuration gets a bad rap for a couple of reasons. One is the increased voltage at idle. As I have already posted, with this board it's a non-issue. The other is stability. Most users see instability because they don't know the true value of VCORE under load. Software is notorious for rounding errors and taking jumps when displaying the voltage. Quite often, because of the benefits VDROOP brings, a slightly less loaded voltage is obtainable. This difference isn't much but it's a difference none the less.

Also you wrote that using a straight 7 allows for the NB to run at the stock voltage. Whats the highest fsb this can work at, and would this apply to 65nm allandale e2180?

Yes it would apply. The highest FSB is going to be dependent on how much voltage a person is willing to push through the NB. At the speeds I posted it would be very easy to show much better performance. I tried to balance performance to voltage for 24/7 use.

If so do I just leave all the phase settings to disable and set the performance level to 7? I asume this works for both the 266/677 and 333/800 straps.

thanks for the help
Wojtek

If you want a guaranteed even Performance Level setting this is best way to obtain it. We'll use a PL of 7 as an example. Set the Performance Level to 8 and enable all the phases for both channels. This will pull it down to 7. The reason for doing this is there is no disable option for Phase Delay. When a phase is left on Auto the BIOS may, under certain circumstances, pull it in. If the NB requires more voltage for this you will then see instability or even a no-boot condition.

[cantankerous]

Setting the voltages up this way removes the only valid argument for disabling VDROOP. The system never sees the additional voltage at idle that results from VDROOP. And from an electronic point of view the benefits of VDROOP are realized. Most boards don't allow this capability. Once default VCORE is changed the voltage no longer scales with frequency.

Thanks for this info Praz, it is much appreciated as usual.

I was always wondering why I was no longer getting a voltage drop at idle, even though my clocks were dropping once I started to overclock the board. It makes perfect sense now however, that before I started overclocking, I had the vcore options set to AUTO which allowed this to happen. Once I started to overclock, I had manually set my vcore which disabled the voltage drop from occuring. I have now mimicked your posting, however have decided to leave vdroop disabled while setting vcore to AUTO, and simply adjusting vcore via the special % value. I found having vdroop enabled caused too much fluctuation in the readings (even for software), and found I needed a much higher % value to keep my cpu stable under load. Once I disabled Vdroop again I was able to get identical under load to what I had with vcore manually set, but now had the added benefit of still having the voltage drop under idle along with the clocks. My cpu temperature average has already dropped 1-2c when compared to before I did this. Judging from my UPS, my system draws 7w less this way too. So far so good. I am going to run some OCCT/Prime with my system set up this way to be sure it truly is still 100% stable like it was before these changes.

Wondering. Is it harder on the board/mosfets running the systems vcore simply from the special % value? I know it may be a stupid question, but I just like to know what I am getting myself into in case something creeps up that didn't before this change was made. I'm just wondering if it is more stressful on the board/components to run things this way, if not the same or even easier?

Thanks again Praz.

[C-N]

Well I got round to playing with the C1E function & for me its happy days

C1E : Enabled
Droop control : Enabled
CPU VID Control : Auto
CPU VID Special Control Add : plus whatever % it needs

Works great scaling CPU speed & voltage up & down between idle & load.

The interesting thing is C1E also still scales the multiplier & voltage up & down with either Droop control enabled or disabled so you can still take advantage of the more stable droop disabled function if need be & retain C1E's function too.

Win Win cant be bad ay.

CN

[Praz]

There were some boards in the past, specifically in the NF4 days, where VID Special Add was best not used. But all seems well now. And with constant voltage monitoring I have seen no issues at all. One thing to note. Changing Special Add just one setting will not always be indicated by monitoring software. The change in voltage is less then the software will display.

Having VDROOP enabled will not cause stability issues. However the fine tuning required with it enabled may be more then some users will want to spend. As C-N has already wrote, disabled or enabled frequency and voltage will still change. Also it goes without saying that for benching turn off this stuff off.

[ixtapalapaquetl]

Changing Special Add just one setting will not always be indicated by monitoring software. The change in voltage is less then the software will display.

Very true. Fluke 179 measurements (assuming the pic below shows correct measurement points - only ones I found on web):

1.40625 BIOS, 1.441 real
1.40625*100.xx%, 1.456 real (forgot to write it down, but one immediately below 100.94%)
1.40625*100.94%, 1.460 real
1.40625*101.18%, 1.462 real

Also, despite CPUZ reports, there is hardly ANY change from load to idle.

[If anyone knows a better measurement point, please let me know! This one is courtesy of symphy]:

[Praz]

You can also measure along the CPU side of the two inductor packs. Makes it a little easier.

[ixtapalapaquetl]

You can also measure along the CPU side of the two inductor packs. Makes it a little easier.

Much easier. Thanks so much!

[runmc]

E8400 02/14 BIOS
333/800 Strap/Divider
2x1GB OCZ PC2 9200 Reaper
XFX8800 GTX

C1E and VDROOP enabled. VCORE was set using VID Special Add leaving VID Control on Auto. Adjusting VCORE this way allows the voltage to decrease along with frequency while idle. Measured voltage at idle - 1.276. Measured under load - 1.330. Frequency at idle - 2670MHz. Frequency under load - 4005MHz. NB voltage can be reduced to stock if tRD is set to a straight 7 (no phases pulled in).

Thanks Praz -

[Krzychu]

[QUOTE=Praz;2799130]

Praz, few questions if you can


Having VDROOP for a 24/7 configuration gets a bad rap for a couple of reasons. One is the increased voltage at idle. As I have already posted, with this board it's a non-issue. The other is stability. Most users see instability because they don't know the true value of VCORE under load. Software is notorious for rounding errors and taking jumps when displaying the voltage. Quite often, because of the benefits VDROOP brings, a slightly less loaded voltage is obtainable. This difference isn't much but it's a difference none the less.


Yes it would apply. The highest FSB is going to be dependent on how much voltage a person is willing to push through the NB. At the speeds I posted it would be very easy to show much better performance. I tried to balance performance to voltage for 24/7 use.


If you want a guaranteed even Performance Level setting this is best way to obtain it. We'll use a PL of 7 as an example. Set the Performance Level to 8 and enable all the phases for both channels. This will pull it down to 7. The reason for doing this is there is no disable option for Phase Delay. When a phase is left on Auto the BIOS may, under certain circumstances, pull it in. If the NB requires more voltage for this you will then see instability or even a no-boot condition.

Thanks for some fresh air in this thread!

[cantankerous]

There were some boards in the past, specifically in the NF4 days, where VID Special Add was best not used. But all seems well now. And with constant voltage monitoring I have seen no issues at all. One thing to note. Changing Special Add just one setting will not always be indicated by monitoring software. The change in voltage is less then the software will display.

Having VDROOP enabled will not cause stability issues. However the fine tuning required with it enabled may be more then some users will want to spend. As C-N has already wrote, disabled or enabled frequency and voltage will still change. Also it goes without saying that for benching turn off this stuff off.

I remember those NF4 days with my Ultra-D which is why I asked about the use of special voltage only and the harm/stress it does to components. I trust your word very much, so if you say it is 100% ok to use strictly this special % to add vcore then that is what I am going to do. I am loving the idle drop I am now getting.

[DemonEyez]

I'm having a annoying startup problem on my LANParty DK P35 T2RS

When my system has been powered off for several hours or longer, the first bootup will not succeed.
I'll have to use the Eazy Clear option to reset my startup settings.
Then after I've been in the bios and have reloaded my settings, it reboots and works flawlessly. When I shut down, then restart my system an hour later it also starts up as usual.

My E8400 is rock solid on 4.2GHz 8.5x490 @ 1.36v .

CPU Feature Page
Thermal Management Control................Enabled
PPM (EIST) Mode...........................Enabled
CIE Function..............................Auto

Main BIOS Page
Exist Setup Shutdown......................Mode 2
Shutdown After AC Loss....................Disabled
O. C. Fail Retry Counter..................0
CLOCK VC0 Divider.........................Auto
CPU Clock Ratio...........................9x
CPU Clock.................................490 MHz
Boot Up Clock.............................Auto
DRAM Speed................................333/800

Voltage Setting Page
CPU VID Special Add.......................+ 200mv
DRAM Voltage Control......................2.25V
SB 1.05V Voltage..........................1.070V
SB Core/CPU PLL Voltage...................1.55V
NB Core Voltage...........................1.45V
CPU VTT Voltage...........................1.20V
Clockgen Voltage Control..................3.45V
GTL REF Voltage Control...................Disabled
CPU GTL1/3 REF Volt.......................0.63x
CPU GTL 0/2 REF Volt......................0.63x
North Bridge GTL REF Volt ...............0.67x

These are my 24/7 settings. When the system is up and running,there's no problem(3Dmark, SuperPi and Orthos all run fine), only during a startup after the system has been powered off for several hours or longer. So is this a coldboot issue and how can I solve it?

[cantankerous]

Are you on the latest 2/14 bios and experiencing this or an older bios? Try and update to the latest as it really is geared more towards the new 45nm cpu's.

[Boyu]

I'm having a annoying startup problem on my LANParty DK P35 T2RS

When my system has been powered off for several hours or longer, the first bootup will not succeed.
I'll have to use the Eazy Clear option to reset my startup settings.
Then after I've been in the bios and have reloaded my settings, it reboots and works flawlessly. When I shut down, then restart my system an hour later it also starts up as usual.

My E8400 is rock solid on 4.2GHz 8.5x490 @ 1.36v .

CPU Feature Page
Thermal Management Control................Enabled
PPM (EIST) Mode...........................Enabled
CIE Function..............................Auto

Main BIOS Page
Exist Setup Shutdown......................Mode 2
Shutdown After AC Loss....................Disabled
O. C. Fail Retry Counter..................0
CLOCK VC0 Divider.........................Auto
CPU Clock Ratio...........................9x
CPU Clock.................................490 MHz
Boot Up Clock.............................Auto
DRAM Speed................................333/800

Voltage Setting Page
CPU VID Special Add.......................+ 200mv
DRAM Voltage Control......................2.50V
SB 1.05V Voltage..........................1.070V
SB Core/CPU PLL Voltage...................1.55V
NB Core Voltage...........................1.45V
CPU VTT Voltage...........................1.20V
Clockgen Voltage Control..................3.45V
GTL REF Voltage Control...................Disabled
CPU GTL1/3 REF Volt.......................0.63x
CPU GTL 0/2 REF Volt......................0.63x
North Bridge GTL REF Volt ...............0.67x

These are my 24/7 settings. When the system is up and running,there's no problem(3Dmark, SuperPi and Orthos all run fine), only during a startup after the system has been powered off for several hours or longer. So is this a coldboot issue and how can I solve it?

try using different memory divider. last time my system had similar problem like your n it solved by change memory divider

[Praz]

try using different memory divider. last time my system had similar problem like your n it solved by change memory divider

Especially since the memory is needing 2.50V to run there.

[mad_skills]

Seems like all DFI LanParty P35 mobos work best with 266/667 and 333/800 Ratios. This is the case with me as well...folks on Anand Tech say that these give the best results considered the following:
* Raw CPU MHz per volt
* Memory bandwidth
* Memory/chipset latency
* Memory copy speeds
* Memory write speeds

And also that raw MHz and loosened latencies works better than smaller MHz with tightened latencies..

[MacKaquE]

Hi everybody, it's been a long time I didn't gave a msg in there

Well .. I have my Q6600 @ 3.6GHZ since 4-5 months now on my P35-T2R and I must say I have fun with this setup .. really happy with it !

Last week problem started tho. I began to get some BSOD with diff msg and they were all realted to RAM mostly. So I deceided to do some memtest and I had no error after 4 hours of memtest so I was like it's probably not the memory. But I still deceided to put some cheap kingston value 2*1gb DDR800 CL5 in my system (I run 9 * 400) and my mem is 1:1 so cheap DDR2 800 was fine with me. had no BSOD at all for 2 days and I had BSOD with my old memory (crucial ballistix tracer PC8500) every 5-6 hours. So I deceided to flash my bios and put back the crucial in it to made some bios change, I read a lot and tried everything and with my crucial, I always have BSOD. in memtest = perfect, in windows, BSOD (I also tried to put my pc back to stock speed and same problem btw ..). I deceided then to try my memory on a new build (stock speed and all). And this pc work for 2 days now without any problem at all.

So what you guys think about this ?

How come my memory is like not compatible anymore with my DFI but with an ASUS M2N-E crap board it seems to work without any problem. Also how come memtest gives me no error ?

Like I said with some cheap memory from kingston my OC is back to stable mode but this ram is slow, need to put CAS 5 and all

So if anyone can explain me how this possible I think I gave all the information, if anyone need more infos let me know I'll give more.

I think I'll just get another kit of memory. Get me some nice deal lately that works good with this board would be nice also

Thanks for everybody who will participate in this weird things

[DemonEyez]

Are you on the latest 2/14 bios and experiencing this or an older bios? Try and update to the latest as it really is geared more towards the new 45nm cpu's.

I'm running the latest official bios 12/24, not the latest beta. I'll try it this weekend. But if 'm correct the only thing changed in this new beta are the GTL table's.

Especially since the memory is needing 2.50V to run there.

Oops, I made a typo there. Memory is running with 2.25v, not 2.50v
I want to keep that on the low side because of the reports of dying Micron DGMH chips running 2.3v 24/7

Seems like all DFI LanParty P35 mobos work best with 266/667 and 333/800 Ratios. This is the case with me as well...folks on Anand Tech say that these give the best results considered the following:
* Raw CPU MHz per volt
* Memory bandwidth
* Memory/chipset latency
* Memory copy speeds
* Memory write speeds

And also that raw MHz and loosened latencies works better than smaller MHz with tightened latencies..

I agree, I'm getting my best results running 333/800 ratio. For my Tracer to run at 590MHz 5-4-4-9 and 2.25v I needed to loosen tRD with the 'fine clock delay settings'. With auto settings I even couldn't run them at 590MHz 5-5-5-15 and 2.4v


Today my system booted correctly from coldboot. I powered down last night, but did not flip the switch on the back off PSU to completely cutt off the power.
Lo and behold, it booted correctly this morning after been powered off for several hours. I’m beginning to think the problem is related somehow to the standby voltage. Any idea’s?