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

[jjk454ss]

If you are using Vista64, there is a hotfix that needs to be applied before you can use more than 3GB, otherwise it will BSOD on startup. I do beleive you also have to update the BIOS as well because I have applied the patch and now the BSOD I get says the bios is not acpi compliant.

http://support.microsoft.com/kb/929777

Strafe, are you using this board? Does anyone know if I do have to install this hotfix? I don't remember doing this with my ASUS P5B Deluxe.

[wedfighter]

tested with various CPU
just a quick test only...

E6300 @ 500x7


E6300 @ 515x7


E6320 @ 529 x 7


E6320 @ 536 x 7


E2140 @ 425 x 8


all done with intel stock cooler

wanna try again with sub-zero

[C-N]

I popped my Q6600 G0 in the DFI last night & after playing with voltages & GTL's for a few hours only just managed to get 495 x 7 to boot into windows a handful of times yet 490 appeared stable.

I know the Q6600 is good for 495fsb with my IP35 & a couple of times I have even booted 500fsb with it.

This morning I have spent 5 hours playing about with the Q6600 & LP P35 & for the life of me cant get anything over 350fsb prime stable on blend although the CPU will run small fft's for hours.

I started to believe I must have a dodgy component so I have swapped out my PSU, Video card & RAM yet still I cant get 350fsb prime stable. I then put my E6400 in as I thought maybe the board has gone funny & bang 500 x 7 with just NB, RAM & CPU voltage increases.

My E6400 has also just passed 1 hour of prime blend @ 450 x 8 so I am pretty sure that rules the board out.

I have also flashed back to 810 from 823 but for me It appears the DFI just cant run the quads like the others yet.

Has anyone else tried a quad yet on this board & if so how are you getting on with it.

CN

[wedfighter]

whre u get 810 BIOS ?
i checked the DFI main website and no 810 BIOS there
only 727, 815 and 823 BIOS

just flashed to 823 BIOS and get C1 randomly when tried to post
after some clear cmos i can have my mobo POST, OC was looks like similar with old BIOS (727)
i'll flash with 815 then..

[C-N]

whre u get 810 BIOS ?
i checked the DFI main website and no 810 BIOS there
only 727, 815 and 823 BIOS

just flashed to 823 BIOS and get C1 randomly when tried to post
after some clear cmos i can have my mobo POST, OC was looks like similar with old BIOS (727)
i'll flash with 815 then..

My board came with 810 already on it so when I flashed to 823 I saved it.

810 is on DFI's BETA site http://us.dfi.com.tw/Support/Downloa...FLAG=B&SITE=US

I had a few C1 problems with 823 & my E6600 when I pushed too hard too quickly but now I just use EZ clear if I mess the CMOS/BIOS up real bad.

A very very nice little trick EVA showed me yesterday is if your rig fails to boot after increasing the FSB you can just power down, hold down the home key & power back up, the board will then switch off & then let go the home button, the rig bill boot up leaving all but your last FSB setting as it was.

I have also just put my Q6600 back in my IP35 Pro & in 5 minutes flat it boots 500 x 7.

CN

[wedfighter]

@ C-N
thx u very much

[RLM]

I have also flashed back to 810 from 823 but for me It appears the DFI just cant run the quads like the others yet.

Has anyone else tried a quad yet on this board & if so how are you getting on with it.

I've had the same finding as you. small FTT's no problems, blend fails. C1 errors on boot etc.

The conclusion I've come to is that the problem is with memory, more than cpu. Basically it's a struggle to get it stable.

I'm still working on this, but first thing is to try setting some of the subtimings and settings manually, rather than just relying on AUTO. Check out memset and work with those settings in bios loosening up a few.

Maybe look at the strap/divider you are using?

2 settings that tend to be the biggest problem for me is EA and EDT. If these are set to FAST, then blend fails. Set to NORMAL, seems to make a big difference stability wise.

Maybe AUTO sets these to FAST on occasion?

As I say I'm still working with thses settings, voltages, GTL refs etc. I do think that a new bios is very much needed and memory is key.

RLM

ps. Great tip with the Home key. Saw that on a Japanese site, but didn't understand obviously.

[C-N]

I've had the same finding as you. small FTT's no problems, blend fails. C1 errors on boot etc.

The conclusion I've come to is that the problem is with memory, more than cpu. Basically it's a struggle to get it stable.

I'm still working on this, but first thing is to try setting some of the subtimings and settings manually, rather than just relying on AUTO. Check out memset and work with those settings in bios loosening up a few.

Maybe look at the strap/divider you are using?

2 settings that tend to be the biggest problem for me is EA and EDT. If these are set to FAST, then blend fails. Set to NORMAL, seems to make a big difference stability wise.

Maybe AUTO sets these to FAST on occasion?

As I say I'm still working with theses settings, voltages, GTL refs etc. I do think that a new bios is very much needed and memory is key.

RLM

ps. Great tip with the Home key. Saw that on a Japanese site, but didn't understand obviously.

Ok well that sort of puts my mind at rest that I haven't got a duff board if you are having similar issues.

I am also fairly certain its a bios/memory related problem as both the RAM & CPU run fine on all my other boards & are running fine in my IP35 right now as I type.

I have tried different RAM, BIOS's & played extensively with my own timings & eva2000's timings from his i4memory site & although I have been very impressed with the way the board handles ram & the speed of some of my Super PI runs I still cant get prime blend to run more than a minute with the Quad. I can however run either full auto or tweaked timings for hours with a dual core.

I will have another go with a quad later, try your suggestions & let you know how I get on.

Ye that home key thing is really something big thanks to eva

CN

[luihed]

I just breeze throught the thread so I might have missed it but are you guys getting higher fsb ( with same cpu ) than your last board?... this is the deal maker for me...

[I34z1k]

What's the story with the home key?

[C-N]

I just breeze throught the thread so I might have missed it but are you guys getting higher fsb ( with same cpu ) than your last board?... this is the deal maker for me...

I am not really that bothered about max fsb as I will only be running air anyway. The DFI certainly makes fsb easier than any of my other boards & it is more than capable of maxing out the dual core CPU's I have its just so far I cant get the DFI to hit the same fsb with a quad & even when I get close its not prime blend stable.

I personally have had my E6600 & 540 & 550fsb on an Abit IP35 Pro & an ASUS P5K Deluxe using uguru & setfsb but that definitely wasn't stable & it definitely wouldn't boot straight from BIOS at those settings where as the DFI will do 525-530 straight from bios on air & was stable enough to run 3D & Super PI etc.

CN

[C-N]

What's the story with the home key?

Here is what I said 2-3 posts ago.

CN

A very very nice little trick EVA showed me yesterday is if your rig fails to boot after increasing the FSB you can just power down, hold down the home key & power back up, the board will then switch off & then let go the home button, the rig bill boot up leaving all but your last FSB setting as it was.

[I34z1k]

Sorry :P I searched for "home" on this page and the last.

Thanks

[RLM]

C-N,

Interesting comments. I haven't tried with a duo core, so wasn't aware of the difference.

How does your vdimm read with the duo and the quad. Is there any difference between the two, including droop at load?

There's so many parameters it's hard to pin point a specific one. Wondering if due to the quad's draw, whether there's a voltage issue in anyway. For instance I've only really topped out at 1.65v on the NB, maybe it needs more

Anyway here is a 20 min blend test with the quad. Memory is Geil 8500 cas 4s. I think that the Gskill HZ's actually perform a bit more reliably on this board.

Some notes running up to this. Running the ram 1060mhz 4-4-4-12 (266/667 4:5) wasn't particularly stable and blend crashed the system. As I say I think maybe my geils are part of the problem there, but it's also a 1066 strap if I'm right??

With 1016mhz 4-4-4-12 (333/800 5:6) at auto performance level of 6, blend failed after a minute. I loosened the performance level to 7, before running the following blend test.... Note Halted rather than failed.

Q6600 GO 423x9 (3.8 ghz) Mem 1016mhz 4-4-4-12 Prime Blend Test

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

Exist Setup Shutdown: Mode 2
CLOCK VC0 divider: AUTO
CPU Clock Ratio Unlock: Enabled
CPU Clock Ratio: 9x
- Target CPU Clock: 3807 Mhz
CPU Clock: 423
Boot Up Clock: AUTO
DRAM Speed: 333/800 (5:6)
- Target DRAM Speed: 1016
PCIE Clock: 100mhz

Voltage Settings
CPU VID Control: 1.55000
CPU VID Special Add: AUTO
DRAM Voltage Control: 2.49v (Allowing for droop -0.05 idle a further -0.04 load)
SB 1.05V Voltage: 1.15v
SB Core/CPU PLL Voltage: 1.95v
NB Core Voltage: 1.57v
CPU VTT Voltage: 1.55v
Vcore Droop Control: Disabled
Clockgen Voltage Control: 3.75v
GTL+ Buffers Strength: Strong
Host Slew Rate: Weak
GTL REF Voltage Control: Disable
x CPU GTL1/3 REF Volt: 105
x CPU GTL 0/2 REF Volt: 105
x North Bridge GTL REF Volt: 110

DRAM Timing
- Enhance Data transmitting: NORMAL
- Enhance Addressing: NORMAL
- DIMM 1 CLK fine delay: Current 2 (empty slot)
- DIMM 2 CLK fine delay: Current 6
- DIMM 3 CLK fine delay: Current 2 (empty slot)
- DIMM 4 CLK fine delay: Current 7

CAS Latency Time (tCL): 4
RAS# to CAS# Delay (tRCD): 4
RAS# Precharge (tRP): 4
Precharge Delay (tRAS): 12
All Precharge to Act: AUTO (MSet Reads 4)
REF to ACT Delay (tRFC): AUTO (MSet Reads 42)
Performance Level: 7 (Auto sets to Perf to 6)
Read delay phase adjust: AUTO
MCH ODT Latency: AUTO
Write to PRE Delay (tWR): 14 (Mset reads 13)
Rank Write to Read (tWTR): 13 (Mset reads 12)
ACT to ACT Delay (tRRD): 3
Read to Write Delay (tRDWR): AUTO (Mset reads 8)
Ranks Write to Write (tWRWR): AUTO (tends to default I believe at 6)
Ranks Read to Read (tRDRD): AUTO (tends to default I believe at 6)
Ranks Write to Read (tWRRD): AUTO (tends to default I believe at 5)
Read CAS# Precharge (tRTP): 4
ALL PRE to Refresh: AUTO (memset reads 2)

PCIE Slot Config: 1X 1X
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
SATA Spread Spectrum: Disabled

[Dihon]

With BIOS 810 I also have problem with C1 & booting after crash, but 823 is better.
No more C1 issue but booting problem still sometimes occurs.
Also it has a problem with 5/8 divider.
So far this is what I get :

[C-N]

C-N, How does your vdimm read with the duo and the quad. Is there any difference between the two, including droop at load?

With the droop control disabled which is enabled it reads slightly over volts (according to CPU-Z) at idle from what is set in bios with both dual & quad core. Under load I have not yet seen 0.02v droop measuring with CPU-Z. Even with the droop control enabled - disabled the droop is minimal & nothing that bothers me coming from an IP35 & I know you know what I mean

There's so many parameters it's hard to pin point a specific one. Wondering if due to the quad's draw, whether there's a voltage issue in anyway. For instance I've only really topped out at 1.65v on the NB, maybe it needs more

1.65 is the highest I have tried too but it stopped having any effect around 1.55 ish volts when I was shooting for 495fsb so I don't think its to blame here.

Anyway here is a 20 min blend test with the quad. Memory is Geil 8500 cas 4s. I think that the Gskill HZ's actually perform a bit more reliably on this board.

Some notes running up to this. Running the ram 1060mhz 4-4-4-12 (266/667 4:5) wasn't particularly stable and blend crashed the system. As I say I think maybe my geils are part of the problem there, but it's also a 1066 strap if I'm right??

With 1016mhz 4-4-4-12 (333/800 5:6) at auto performance level of 6, blend failed after a minute. I loosened the performance level to 7, before running the following blend test.... Note Halted rather than failed.

My rig wont run prime blend past 400fsb on the 6, 7, 8 or 9 multiplier on either the 333/667, 333/800 or the 266/667 memory dividers I am stuck

CN

[RLM]

With the droop control disabled which is enabled it reads slightly over volts (according to CPU-Z) at idle from what is set in bios with both dual & quad core. Under load I have not yet seen 0.02v droop measuring with CPU-Z. Even with the droop control enabled - disabled the droop is minimal & nothing that bothers me coming from an IP35 & I know you know what I mean

No mate, I meant the memory voltage, not vcore

RLM

[C-N]

With BIOS 810 I also have problem with C1 & booting after crash, but 823 is better.
No more C1 issue but booting problem still sometimes occurs.
Also it has a problem with 5/8 divider.
So far this is what I get :

Thanks for the reply Dihon, I was told to watch out for the 5/8 divider but I thought that was just with 823.

I have tried all the available BIOS's & they all behave just the same for me.

I haven't had a wipe out crash that EZ Clear or the home key couldn't fix all day so I suspect it was just me being a bit heavy handed with it the first night I had it.

I can run PI & 3DMARK just fine its prime blend that wont keep going over 400fsb.

CN

[RLM]

With BIOS 810 I also have problem with C1 & booting after crash, but 823 is better.
No more C1 issue but booting problem still sometimes occurs.
Also it has a problem with 5/8 divider.
So far this is what I get :

Decent chip you've got there Dihon . Will it prime at those settings, and what stepping is it?

The only thing I will say is that running 1:1 like you've got there, does seem a lot easier stability wise. I can run 3-3-3-9 at around about that speed.

It's when trying to hit 1000+ on one of the other dividers, that problems seem to kick in. Especially on 4-4-4.

RLM

[C-N]

No mate, I meant the memory voltage, not vcore

RLM

Sorry bud I didn't read it right, I am starting to loose it as I have been playing with this board almost non stop for 24 hours.

I get 0.9v difference between what I set in BIOS & what Smartguardian reads with the quad which looks just the same as you.

I wouldn't believe Smartguardian anyway its been way off for as long as I can remember but. I will get a DMM out a measure a slot when I start pushing my voltages up but at the moment my OCZ's are at 2.1v in BIOS @ 800mhz CL5 so I have plenty of room to play with atm.

I didn't pay attention when I had a dual core in so don't have a comparison.

CN

[RLM]

Sorry bud I didn't read it right, I am starting to loose it as I have been playing with this board almost non stop for 24 hours.

No worries mate, I know what you mean, these DFI's are full on.

RLM

[RLM]

I wouldn't believe Smartguardian anyway its been way off for as long as I can remember

Just taken some readings with a multimeter. Thanks for pointing me in the right direction.

Smartguardian seems to be more accurate than bios according to my multimeter.

Bios set to 2.49v vdimm

Readings in windows xp taken from a free slot.

1 memory stick in reads 2.46v
2 memory sticks in read 2.44v

Reading taken while in bios

2 memory sticks in read 2.44v

So according to my multimeter -0.05v from what is set in bios.

I did a blend run, while taking a reading. SMG reports a droop of about -0.04v. Multimeter reports no droop at all.

Not sure how these things work, and whether that's accurate for the slots the memory sticks are occupying at the time.

RLM

[C-N]

Panic over I have finally got to the bottom of my problem of not being able to run over prime blend over 400fsb on this board with a Q6600.

Please don't all kick me at once it was the dammed PSU playing up.

I had actually swapped out the PSU trying to eliminate it but I must have been too far off with my settings to notice the difference between a crash due to a PSU fault & an unstable crash

My old boss used to say things like this were character building & you know what I think I finally got what he was on about now although this has given me a Jackal & Hyde character as last night I was ready to put a hammer through this board fuming with rage yet now with the weight off my shoulders & the board priming at 450fsb I feel like kissing it.

CN

[C-N]

She's trucking now, it feels like a different board.

The PWM & NB are running cool as ice passively cooled with relatively low voltage.

Voltages in BIOS.

CPU VID Control: 1.45v
CPU VID Special Add: AUTO
DRAM Voltage Control: 2.230
SB 1.05V Voltage: 1.070v
SB Core/CPU PLL Voltage: 1.55v
NB Core Voltage: 1.50v
CPU VTT Voltage: 1.60v
Vcore Droop Control: Disabled
Clockgen Voltage Control: 3.45v
GTL+ Buffers Strength: Strong
Host Slew Rate: Weak
GTL REF Voltage Control: Disable
x CPU GTL1/3 REF Volt: 110
x CPU GTL 0/2 REF Volt: 110
x North Bridge GTL REF Volt: 110

CN

[Dihon]

Decent chip you've got there Dihon . Will it prime at those settings, and what stepping is it?

The only thing I will say is that running 1:1 like you've got there, does seem a lot easier stability wise. I can run 3-3-3-9 at around about that speed.

It's when trying to hit 1000+ on one of the other dividers, that problems seem to kick in. Especially on 4-4-4.

RLM

No, I can't run Prime at those setting.
My Q6600 is L724 31Jul07.

Yeah, 1:1 is the most stable configuration you can get.

Seems that, what is necessary for this board is a BIOS that can make it run with 500+ FSB on Kentsfield chip.
Cause I can't find one that can do it.