DFI LP UT X58-T3EH8 - First bench, bios screens etc

[eva2000]

Max BCLK Super Pi 32M Benchable

Did a fresh hard drive install which is WinXP Pro SP3 & Windows 7 beta dual bootable before attempting max Super Pi 32M benchable Bclk frequency tests. This time with 3x1GB Samsung HCF0 based modules for triple channel

System specs

• Intel Core i7 920 3836A756
• Thermalright Ultra 120 Extreme + LGA1366 Bolt Thru kit + 2x 120x38mm Scythe UltraKaze 2000rpm 87cfm fans in push/pull
• DFI LP UT X58-T3EH8 R.A51 - 01/06 beta bios
• 128MB Gainward FX5200 PCI
• 2x1GB Corsair DDR3-2133C9DF Dominator + 1GB G.Skill PC3-12800HZ (3x Samsung HCF0)
• Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
• 80GB Hitachi 7K80 SATAII
• Pioneer DVD-RW
• 1KW Corsair HX1000 psu
• WinXP Pro SP3 full nLite updated hotfixes / Windows 7 beta dual boot

Triple Channel CPUZ Validation @229Bclk
Haven't pushed ICH 1.5 Voltage yet like I did for 230Bclk, so that will be next up for testing.



QPI Frequency = 4123Mhz!



Super Pi 32M
Ignore the times just testing with loose timings. Higher Bclk do require more vdimm that usual due to QPI ceiling. Without enough vdimm DFI post diagnostic LED = CF code for lack of vdimm voltage.

@226Bclk bios set - QPI Frequency = 4063Mhz



@226Bclk setFSB windows set - QPI Frequency = 4074Mhz



@227Bclk - QPI Frequency = 4088Mhz



@228Bclk - QPI Frequency = 4110Mhz
- had to bump vdimm, vcore and CPU VTT Special percentage added
- either a bad run, or lack of volts for either vdimm or CPU VTT or both led to slower 32M Pi time ?






01/06 Beta Bios settings used:

Genie BIOS Setting

CPU Feature
Set VR Current Limit Max: Disabled
Thermal Management Control: Disabled
EIST Function: Disabled
CxE Function: Disabled
Execute Disable Bit: Disabled
Virtualization Technology: Disabled
TDC Enable: Disabled
x TDC Limit: 0
TDP Enable: Disabled
x TDP Limit: 0

***** Logical Processor Setting *****
Intel HT Technology: Enabled
Active Processor Cores: All

DRAM Timing
Memory Control Setting: AUTO
Memory LowGap: 1024M

DRAM Command Rate: AUTO
CAS Latency Time (tCL): 9
RAS# to CAS# Delay (tRCD): 9
RAS# Precharge (tRP): 9
Precharge Delay (tRAS): 24
REF to ACT Delay (tRFC): 88
Write to PRE Delay (tWR): AUTO
Rank Write to Read (tWTR): 16
ACT to ACT Delay (tRRD): 6
Row Cycle Time (tRC): 31
Read CAS# Precharge (tRTP): 6
Four ACT WIN Time (tFAW): 31

Voltage Setting
O.C. Shut Down Free: Enable O.C.S.D.F
CPU VID Control: 1.350v * 1.3875v
CPU VID Special Add Limit: Enabled
CPU VID Special Add: Auto
Vcore Droop Control: Enabled
DRAM Bus Voltage: 1.875v & 1.905v
DRAM PWM Switch Frequency: Nominal
DRAM PWM Phase Control: 2 Phase Operation
CPU VTT Special Add: +0.0375v & +0.0500v
CPU VTT Voltage: 1.53v
VTT PWM Switch Frequency: Nominal
VTT PWM Phase Control: 2 Phase Operation
CPU PLL Voltage: 1.80v
IOH/ICH 1.1V Voltage: 1.25v
IOH Analog Voltage: 1.20v
ICH 1.5 Voltage: 1.5v
ICH 1.05V Voltage: 1.05v
DIMM 1/2 DQ/DQSTB Bus VREF: -00.0%
DIMM 3/4 DQ/DQSTB Bus VREF: -00.0%
DIMM 5/6 DQ/DQSTB Bus VREF: -00.0%
ADDR/CMD VREF Control: Disabled
x DIMM 1/2 ADDR/CMD Bus VREF: 110
x DIMM 3/4 ADDR/CMD Bus VREF: 110
x DIMM 5/6 ADDR/CMD Bus VREF: 110
CPU QPI Drive Strength: Normal
IOH QPI Drive Strength: Normal

Exit Setup Shut down: Mode 2
O.C. Fail Retry Counter: Enabled
O.C. Fail CMOS Reload: Disabled

PPM Function: Enabled
Turbo Mode Function: Disabled
1 core Max Turbo Ratio: 22x
2 core Max Turbo Ratio: 21x
3 core Max Turbo Ratio: 21x
4 core Max Turbo Ratio: 21x
CPU Non-Turbo Clock Ratio: 18x

* BCLK/UCLK/QPI Controller Settings *
QPI Control Settings: Enabled
QPI Link Fast Mode: Enabled
QPI Frequency: AUTO
CPU Base Clock (BCLK): 226 to 229 Mhz
Boot Up CPU Base Clock: AUTO
PCIE Clock: 103 Mhz
DRAM Frequency: 8x
UnCore Frequency: 16x (optimal to keep Uncore at 2x DRAM Frequency)

CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled

[jcool]

Nice going eva!

Which bios do you prefer, 12/31 or 01/06?
Currently I am having trouble upping the ram mult.. I got my 4,4Ghz 24/7 stable but this 940 usually won't even boot into windows if I so much as raise Dram/uncore to 8x/16x
That is using loads of VTT, different Vdimm settings, and really slack CAS9 timings that are memtest stable for hours (hell I passed memtest easily at 7-6-6-15 1600, that's not the issue).
Do you think the CPU simply requires more Vcore to run the mem higher? If so that would suck cause it means I'd have to step down from 4,4Ghz 24/7, or run 1,6V
Any other voltage having an impact here besides VTT and Vdimm? Maybe I haven't hit the right correlation between VTT and Vdimm? Although it seems stupid having to fine-tune the mobo to that extent only to run 8x mem divider at 192 BCLK.
Also I thought maybe there are variances depending on which CPU mult you use, so I tried 21 and 23x.. same disappointing results with both. At least my crap 920 could run mem like a champ...

Not even talking about the 10x as it is still unusable (needs 2,05V+ Vdimm).

[eva2000]

Thanks, I like 01/06 as previous bios my G.Skill PC3-12800HZ D9JNM and Samsung HCF0 modules didn't boot properly (HCF0) and weren't as stable (D9JNM). 01/06 allows my Samsung HCF0 to shine - see below

I haven't investigated the relationship between each mem multiplier in detail yet - but yes higher mem multi max ram clocking harder i.e. between 8x versus 10x versus 14x but look carefully at your subtimings on AUTO at each mem multi and at each mem frequency you test particular tWTR which on AUTO or even when set will have different values at different mem multis as well as at different mem frequencies if you check using cpu tweaker and/or memset latest versions


Max BCLK with 10x Memory Multiplier
- limited by memory max overclock frequency

Now to try max Bclk frequency with 10x memory multiplier - ultimately will be limited by how far my memory modules can overclock

System specs

• Intel Core i7 920 3836A756
• Thermalright Ultra 120 Extreme + LGA1366 Bolt Thru kit + 2x 120x38mm Scythe UltraKaze 2000rpm 87cfm fans in push/pull
• DFI LP UT X58-T3EH8 R.A51 - 01/06 beta bios
• 128MB Gainward FX5200 PCI
• 2x1GB Corsair DDR3-2133C9DF Dominator + 1GB G.Skill PC3-12800HZ (3x Samsung HCF0)
• Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
• 80GB Hitachi 7K80 SATAII
• Pioneer DVD-RW
• 1KW Corsair HX1000 psu
• WinXP Pro SP3 full nLite updated hotfixes / Windows 7 beta dual boot

Interesting that up to DDR3-2263Mhz system registered 3x1GB Samsung HCF0 unmatched modules as triple channel, but when I hit DDR3-2298Mhz, windows reported dual channel instead of triple channel ?? Not enough volts somewhere or maybe my 1GB G.Skill PC3-12800HZ module can't keep up with my Corsair DDR3-2133C9DF modules ??

Triple Channel CPUZ Validation @226Bclk
+
DDR3-2263Mhz at 1.995v bios vdimm





QPI Bus frequency = 4074Mhz



DDR3-2298Mhz at 2.025v vdimm bios
Triple channel showing up as Dual channel





Max Super Pi 32M @DDR3-2228Mhz 1.995v vdimm bios

[jcool]

I haven't investigated the relationship between each mem multiplier in detail yet - but yes higher mem multi max ram clocking harder i.e. between 8x versus 10x versus 14x but look carefully at your subtimings on AUTO at each mem multi and at each mem frequency you test particular tWTR which on AUTO or even when set will have different values at different mem multis as well as at different mem frequencies if you check using cpu tweaker and/or memset latest versions

Yeah sure, but read again, I can't even get the 8x to work stable at 192 BCLK, half of the time it freezes or bluescreens during boot already
Any ideas on that?

[eva2000]

Yeah sure, but read again, I can't even get the 8x to work stable at 192 BCLK, half of the time it freezes or bluescreens during boot already
Any ideas on that?

try backing off timings as if you have old non-triple channel D9GTR could be spd is too tight and with 3x2GB you probably need more volts than most modules...

Always use memtest86+ v2.11 to see where you before booting windows, as where memtest freezes or errors will indicator what you need to do to overcome those errors. Playing with volts etc and re-testing in memtest86+ to see if you overcome those errors

try

starting with loose subtimings

DRAM Command Rate: AUTO or 2T
CAS Latency Time (tCL): 9
RAS# to CAS# Delay (tRCD): 9
RAS# Precharge (tRP): 9
Precharge Delay (tRAS): 24 to 28
REF to ACT Delay (tRFC): 88 to 110
Write to PRE Delay (tWR): AUTO
Rank Write to Read (tWTR): 16 to 20 depending on ram frequency and timings and mem divider
ACT to ACT Delay (tRRD): 6 to 8
Row Cycle Time (tRC): 31
Read CAS# Precharge (tRTP): 6 to 8
Four ACT WIN Time (tFAW): 31

[Astratuner]

mmm i'm still not sure about wich motherboard i'm going to pick but a dfi is a contestant.


Now I've seen the new enermax revolution85+ psu's and I want one.

So knowing these things what should i do? Take the risk or switch from psu or else the other motherboard i've had in my mind?


Thanks

Anyone?

[Praz]

are ICH 1.05v and CPU PLL in right place cause if you look at PCB labels the seem to be switched ? I haven't measured them myself yet heh

The revised photo I posted last night should be correct. I received it from DFI, Taiwan.

[jcool]

try backing off timings as if you have old non-triple channel D9GTR could be spd is too tight and with 3x2GB you probably need more volts than most modules...

Always use memtest86+ v2.11 to see where you before booting windows, as where memtest freezes or errors will indicator what you need to do to overcome those errors. Playing with volts etc and re-testing in memtest86+ to see if you overcome those errors

try

starting with loose subtimings

Thanks for the ballpark figures, I already use memtest+ 2.11 ofc
Like I said, no errors anywhere (except when running the 10x mem divider at 200 BCLK below 2,05V Vdimm..)

I am feeling rather reluctant to go play around with memory settings though, as I got a black screen testing 190 BCLK and 8x ram mult (again), but this time it wouldn't come back up. No amount of CMOS clearing, disconnecting and reconnecting from/to the wall helped, had to remove battery for 2h before it would fire back up again.

And in that process I noticed my 940 has like the lowest cold bug ever.. below -40C it just won't come up, mobo starts with "88" and that's it.
Starting to believe this chip is also a crappy one, in its very own way (mIMC/coldbug/what else?)

Seems that with i7, you can't judge a CPU anymoe just by how high the Mhz will go, but rather how high it'll go at decent memspeed.

[burningrave101]

Seems that with i7, you can't judge a CPU anymoe just by how high the Mhz will go, but rather how high it'll go at decent memspeed.

Yup that's what happens when you have an IMC on-die. It kinda sucks because that sorta seems to be the problem with my i7 920. It will run 4Ghz at 21x190 but it's not wanting to do it a 19x211 regardless of me upping the voltage to the QPI to 1.45v+ and VDIMM 1.75v+. I've seen others however do much better on 19x211 and are able to lower their vcore.

It's been the same way with AMD since the launch of the Athlon 64 in terms of overclockability. When you overclock the cores your also overclock the MC and some MC's are weaker than others and can't handle high memory clocks when running at a high speed.

[Dami3n]

Im playing with ram, but i have doubts about relation between vtt and ram speed. What vtt should i use to 2000mhzs 9-9-9-24 1,65v(166x12)?

[jcool]

Well that depends on your BCLK and the quality of your chips IMC (and also on whether your mems can actually handle 2000 Mhz), but I guess you'll need at least 1,5V VTT, more like 1,55-1,6V to be stable.
Also don't forget the 10x divider is still bugged on the DFI, meaning it takes more Vdimm than it should. For example, I need 2,05V to pass memtest at 2000 CAS 8-8-8 with my 6GB CSX Diablos using the 10x divider, however if I run it at the same speed using the 12x or any other divider it only needs 1,95V to pass.
Of course none of these settings work in windows so far, so I'll need to fine-tune that freakin thing some more. Seems I got a crappy IMC anyway. AND a coldbugged CPU as well, argh

[DavidNJ]

Anyone?

You aren't clear what you have, what your budget is, or what you are planning to do.

If you have an R2E and a 500w PSU and plan triple SLI, you need a PSU. However, if you have a $400 MB, are adding $750+ in video cards, who cares about the cost of a PSU.

If this is a first Core i7 restricted budget build, the DFI would be a great board IMHO (my experience limited to the UT P45). However, the Corsair EX1000 PSU, often available with rebate for $200-250 would be more cost effective.

Another factor is if you are overclocking and your experience. Everyone looks at the max overclocks. I also look at the work that went into them, and how many people have trouble achieving low overclocks. That is, are there lots of posts about being unable to reach 200bclk or 4Ghz.

Even worse are lots of NewEgg posts about DOA boards. A problem with Gigabyte in general according to the NewEgg posts; on the popular EP45-UD3P, they had 5 consecutive different (lengthy) posts about DOA boards at the top of the list. It is easy to be lulled by often unused features or reports of 600Mhz FSBs.

The DFI board seems, by my observation, to require some expertise to OC because it doesn't always default as expected and may have some nuances at specific speeds and combinations. However, it also seems to have both the fewest reports of failing to make acceptable OCs and a large number of reports (relative to the total number of reports) of people making high OCs. Or even EVA2000's outrageously high OCs, matching some cascade phase change people while cooling with air.

[Dami3n]

Well that depends on your BCLK and the quality of your chips IMC (and also on whether your mems can actually handle 2000 Mhz), but I guess you'll need at least 1,5V VTT, more like 1,55-1,6V to be stable.
Also don't forget the 10x divider is still bugged on the DFI, meaning it takes more Vdimm than it should. For example, I need 2,05V to pass memtest at 2000 CAS 8-8-8 with my 6GB CSX Diablos using the 10x divider, however if I run it at the same speed using the 12x or any other divider it only needs 1,95V to pass.
Of course none of these settings work in windows so far, so I'll need to fine-tune that freakin thing some more. Seems I got a crappy IMC anyway. AND a coldbugged CPU as well, argh

Thanks jcool. What should be a safe vtt for 24/7 oc?

[Astratuner]

You aren't clear what you have, what your budget is, or what you are planning to do.

If you have an R2E and a 500w PSU and plan triple SLI, you need a PSU. However, if you have a $400 MB, are adding $750+ in video cards, who cares about the cost of a PSU.

If this is a first Core i7 restricted budget build, the DFI would be a great board IMHO (my experience limited to the UT P45). However, the Corsair EX1000 PSU, often available with rebate for $200-250 would be more cost effective.

Another factor is if you are overclocking and your experience. Everyone looks at the max overclocks. I also look at the work that went into them, and how many people have trouble achieving low overclocks. That is, are there lots of posts about being unable to reach 200bclk or 4Ghz.

Even worse are lots of NewEgg posts about DOA boards. A problem with Gigabyte in general according to the NewEgg posts; on the popular EP45-UD3P, they had 5 consecutive different (lengthy) posts about DOA boards at the top of the list. It is easy to be lulled by often unused features or reports of 600Mhz FSBs.

The DFI board seems, by my observation, to require some expertise to OC because it doesn't always default as expected and may have some nuances at specific speeds and combinations. However, it also seems to have both the fewest reports of failing to make acceptable OCs and a large number of reports (relative to the total number of reports) of people making high OCs. Or even EVA2000's outrageously high OCs, matching some cascade phase change people while cooling with air.

Hey DavidNJ,


I'm not talking about PSU costs.


Earlier in the thread they said the DFI + Enermax combinations isn't a good one?



That's the point where I asked help on


It's my first core i7 board + I'm new on Oc...


Maybe this is not meant to be for me

[The Nemesis]

How the heck do you get the vtt to go lower than 1.26v since it defaults at 1.21v and overvolts by .05v? I'm trying to settle in on a 24/7 low volt overclock but this thing overvolts everything! I think this board is gonna end up a testbed or in the closet. I need a 24/7 gaming board with a nice 4GHz oc decent volts and timings. Both my chips seem to bench 4GHz 1.35v or so but 8hrs prime takes 1.4v.

[jcool]

Thanks jcool. What should be a safe vtt for 24/7 oc?

I think as of yet, no one can truly answer that. Intel max spec is 1,375V on VTT I think, but in order to get a decent OC to be 24/7 stable, you gotta go way past that point. I am currently running around 1,48V but am on phase and still on lowest ram mult, running high memspeed with 6GB+ WILL need >1,5V, more like 1,6V to be stable.

If you're not comfortable with the possibility of your chip dying within a year or even a few months, I'd suggest staying at or below 1,45V.

@Nemesis: I always wonder how he can even boot at the low VTTs he uses
Must either be the chip (great IMC as opposed to my sucky one) or the fact he's using 3x1GB

Oh yeah, I think I have a confirmed crappy IMC. Tried your settings for ram subtimings EVA, also 2T and went up to 1,6V VTT, 8x ram mult still crashes at 23x192 after 20-30 minutes of prime or 5 minutes linX
Uncore seems to work fine, currently running 6x Dram / 20x uncore, 2h prime blend stable

I just don't get how it could be THAT crappy. I mean, it's supposed to support Tri-Channel DDR3-1600 right? Well obviously mine doesnt

2000 CAS9 worked fine on the 920, so I know it's not the board or ram holding me back. It really is the IMC

[Dami3n]

Definitely i haven´t luck. I have seen people with the same gskill pi black than me doing 1800+ estable at 8-8-8-21 1,65v, but my kit can´t load windows at those settings, even with 1,5vtt

[eva2000]

Definitely i haven´t luck. I have seen people with the same gskill pi black than me doing 1800+ estable at 8-8-8-21 1,65v, but my kit can´t load windows at those settings, even with 1,5vtt

try playing musical chairs (dimm slots) with your 3 modules as they might like particular dimm slots - helped me with my 2 sets of 2x1GB G.Skill PC3-12800HZ D9JNM based modules on DFI UT X58-T3H8 and also play with DRAM/VTT PWM Switch frequencies as well - some of my modules like Reduce 15% over Nominal setting but can vary depending on cpu/mem clock for some modules. My 2x1GB OCZ PC3-14400 Platinum D9GTR, 2x1GB Cellshock PC3-14400 D9GTR and G.Skill PC3-12800HZ D9JNM based modules handle better with like Reduce 15% over Nominal setting but my Samsung HCF0 based G.Skill PC3-12800HZ and Corsair DDR3-2133C9DF like Nominal.

[Dami3n]

Thanks eva. I play with sticks and slots, and voila! rock solid 1800


PD-Don´t look vcore(high), im testing only ram.

[jcool]

Lol? Are you serious? Man, time for some DIMM swapping I guess

[Dami3n]

Yes, totally serious. Strangest thing ever

Are there any option to recover from a bad overclock settings with default config instead remove battery?

[The Nemesis]

Yes, totally serious. Strangest thing ever

Are there any option to recover from a bad overclock settings with default config instead remove battery?

U tried this already?

[roller11]

I just got my DFI x58 board a couple days ago. I got it mostly because EVA tested vista sleep mode and reported that it resumed out of SM in 6 seconds. To my great disappointment, my board resumes in a very slow 14 seconds. So either there's something wrong with my particular board, or, 14 seconds is normal and therefore there's no point in getting a replacement from Newegg. I have to make a decision in the next day or two so I'm appealing to anyone who has a x58 board to do a simple test...put your x58 board to sleep, and time how long it takes to resume, power button to desktop reappearing. Thanks.

[eva2000]

Thanks eva. I play with sticks and slots, and voila! rock solid 1800


PD-Don´t look vcore(high), im testing only ram.

it always pays to test each memory module and in all combinations on dimm slots for whatever board you use you will find tiny differences which may help

Are there any option to recover from a bad overclock settings with default config instead remove battery?

tried power down rig, the holding Home key when you hit power on button ? that will try to boot system from default bclk setting of 133 while retaining all other settings i.e. voltages, timings etc once boot in just change to a stable bclk or try another setting if you are testing stuff

hitting Inset key instead of Home key while powering up clears everything to default IIRC

[eva2000]

I just got my DFI x58 board a couple days ago. I got it mostly because EVA tested vista sleep mode and reported that it resumed out of SM in 6 seconds. To my great disappointment, my board resumes in a very slow 14 seconds. So either there's something wrong with my particular board, or, 14 seconds is normal and therefore there's no point in getting a replacement from Newegg. I have to make a decision in the next day or two so I'm appealing to anyone who has a x58 board to do a simple test...put your x58 board to sleep, and time how long it takes to resume, power button to desktop reappearing. Thanks.

latest hot fix updated vista ? I was using Vista Ultimate SP2 Beta 64bit with latest hotfix integrated drivers see http://i4memory.com/f78/vista-ultima...otfixes-13441/

for Vista Ultimate SP1 latest post-SP1 hotfixes i used http://i4memory.com/f78/list-vista-p...ixes-etc-9492/

8 extras too long to wait ???

How devices you have installed on pc ? usb connections etc ? check your windows event log as it will log your resume process