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

[roller11]

Update: Make that 5 920s tested. Just got another from BuyNow.com. Looked at the sticker, another week 41 chip, version#002, a bad sign. Sure enough, a dud. This one runs cool, but can do only 3927Mhz in my standard prime run of Vcore=1.344 HT off, Tuniq Tower 120 at 2100 RPM. My 'good' chip has done 4116 Mhz with identical settings and components. My two good chips are both week 37, version #001. Are the later chips generally weaker, or is this just a coincidence?

[jcool]

Guys,

I am having real issues here.. Can't get the damn system to boot with EIST + turbo enabled. Not even at 4Ghz (I gave 1,5V+ Vcore so that's not the issue). Same settings work fine when I disable EIST (200x19 and 200x20 even tho 200x20 needs enourmous amount of Vcore, way more than 220x19 for example - I think my 920 just hates the 20x mult which is why I need the freakin 21x to work)

Any suggestions?

[vega22]

It's a horrible story. We hope that Eva lights this problem.
BTW my 940 in 20x200 in ex58 needs tons more of Vcore than 19x211.

[The Nemesis]

Guys,

I am having real issues here.. Can't get the damn system to boot with EIST + turbo enabled. Not even at 4Ghz (I gave 1,5V+ Vcore so that's not the issue). Same settings work fine when I disable EIST (200x19 and 200x20 even tho 200x20 needs enourmous amount of Vcore, way more than 220x19 for example - I think my 920 just hates the 20x mult which is why I need the freakin 21x to work)

Any suggestions?

what r u trying to run the ram at using this multiplier and what do you have cpu vtt/qpi voltage set at? If your running your memory on the 10 multiplier this board seems to have a need to boost the vtt/qpi voltage. It happens to me @ 20 with turbo disabled. Enabled no issue.

[eva2000]

I'm sorta in the same boat, I have a giga ud5 and my first task is to get the DFI dialed in so I can do a true "apples to apples " head to head comparison between the DFI and the giga with same components. I'll use a voltmeter to insure same Vcore and do two prime runs, 1.34v and 1.45v Vcore. I'm using a Tuniq Tower 120mm w/120mm fan at 2100 RPM
for benching (900 RPM for 24/7). I have to turn off HT for benching cause heat overwhelms with HT enabled.
BTW, for what it's worth...I have tested 4 i7 920's to date. Two were from Amazon.com LLC (sold and shipped by Amazon.com as opposed to one of their 3rd party vendors).
The first of these two was very weak, could do only 3.7 Ghz or so. The second was defective in that no matter what I did, it would max out temps on all cores at 100C within a couple minutes. I sent both of them back, the point being that they both were very similar in their batch numbers and version nums (#....-002) and pack dates. I ordered a third chip and when I got it, I saw the same info as the other two bad chips and decided to send it back without opening it. Now maybe there's nothing to it, just bad luck, but I would stay away from Amazon.com LLC as a source. the two 'good' chips were version num #....-001 and were right in the ball park of expected performance. Not great like eva2000, but at least average from what I've seen on this forum. To be clear, I'm only complaining about Amazon.com LLC, not BuyNow.com, which is also a vendor on the Amazon site. One of the two 'good' chips was in fact from BuyNow.

Update: Make that 5 920s tested. Just got another from BuyNow.com. Looked at the sticker, another week 41 chip, version#002, a bad sign. Sure enough, a dud. This one runs cool, but can do only 3927Mhz in my standard prime run of Vcore=1.344 HT off, Tuniq Tower 120 at 2100 RPM. My 'good' chip has done 4116 Mhz with identical settings and components. My two good chips are both week 37, version #001. Are the later chips generally weaker, or is this just a coincidence?

ouch there are average, better than average and gem i7 cpus out there, seem I got a good pair with 3836A756 batch



I've seen good i7 920s in week 36, 37, 38 batches so there could be some truth to this ?

Guys,

I am having real issues here.. Can't get the damn system to boot with EIST + turbo enabled. Not even at 4Ghz (I gave 1,5V+ Vcore so that's not the issue). Same settings work fine when I disable EIST (200x19 and 200x20 even tho 200x20 needs enourmous amount of Vcore, way more than 220x19 for example - I think my 920 just hates the 20x mult which is why I need the freakin 21x to work)

Any suggestions?

what batch/fpo ? and yeah 20x multi is harder to clock on most boards as it's running cpu at full power, turbo with eist I don't think the cores are always loaded and there has to be some form of speed step involved as i benched wprime which stress all 8 threads at 200x215 vs 210x21 = 4.3ghz (non-turbo) vs 4.4ghz (turbo) and times were identical

21x has higher bclk max than 20x too http://i4memory.com/102697-post7.html

It's a horrible story. We hope that Eva lights this problem.
BTW my 940 in 20x200 in ex58 needs tons more of Vcore than 19x211.

read paragraph above

what r u trying to run the ram at using this multiplier and what do you have cpu vtt/qpi voltage set at? If your running your memory on the 10 multiplier this board seems to have a need to boost the vtt/qpi voltage. It happens to me @ 20 with turbo disabled. Enabled no issue.

yup true http://i4memory.com/102545-post5.html high mem clock = higher cpu vtt needed

More turbo mode notes i posted http://i4memory.com/102693-post9.html

I tested 21x205 and 21x210 for Super Pi 1M, 8M, 32M and 3dmark06.

Notes:

• Seems at high bclks like >210, when you move the cpu multiplier up and then raise bclk i.e. from 205 to 210, it can get to a stage were the you'd need to increase CPU VTT even more despite the same higher bclk (210) being stable at a lower cpu multiplier at a set CPU VTT value.
• I experienced this as I passed Super Pi 32M @21x210 = 4410Mhz at 1.4875v but couldn't get 3dmark06 to pass cpu tests even with 1.55v vcore by itself with CPU VTT at 1.48v which was stable for 220bclk on lower cpu multipliers.
• Tried raising CPU VTT by itself to 1.55v and vcore at 1.4875v still couldn't pass cpu tests in 3dmark06. Then remembered sometimes Core i7 cpus like to have vcore and CPU VTT within 0.02-0.03v of each other. So tried vcore 1.525v and CPU VTT at 1.55v and 3dmark06 cpu tests passed with no problems!

don't forget to use bios template to share you settings used http://i4memory.com/f80/dfi-lp-ut-x5...emplate-12919/

[roller11]

EVA...Looking at your settings I don't know what 'VID' is although it's referenced in giga boards also. Explain? After 5 weeks or so with my giga x58-ud5, seems it's pretty straightforward to get max OC. Just increase Vcore and Vtt and get higher CPU clk up to the limit of the chip. Is it pretty much the same with the DFI board, I'm asking cause a lot of your settings don't exist on the giga board, so I have no idea if they are critical or not. I want to leverage your superb efforts as much as possible so I can get to max CPU clk as quickly as possible when I get my DFI on Tuesday. Finally, did you obtain your CPUs in Australia?

[jcool]

what r u trying to run the ram at using this multiplier and what do you have cpu vtt/qpi voltage set at? If your running your memory on the 10 multiplier this board seems to have a need to boost the vtt/qpi voltage. It happens to me @ 20 with turbo disabled. Enabled no issue.

No, was using the x8 mem divider and x16 BCLK.
Tried VTT from 1,25 to 1,55V to no avail.
WTF guys...

@eva: It's a 3838A so brand-new. How do you set the fixed 21x mult? In bios I can only select the 20x max and the turbo 21x is there in grey.. I am guessing my CPU might just hate the 20x so much that it won't even get to run the 20x with mult set to 21x, EIST and Turbo enabled
Not even at 1,55V and -20C which is weird tho?

[palee02]

for fixed 21x multi, you have to set Vista Power managment to High performance ...

[jcool]

Thanks... but I think I know what the problem is already
My CPU is just really really crappy... it can hardly do 4,2Ghz stable with 1,58V
I need a better one, and fast.

[vega22]

@eva: It's a 3838A so brand-new.

Same here: 3838A. Already on sale

[jcool]

Yep mine definitely sucks. 213x19 is all I am gonna get out of this crappy thing stable... It doesn't work right with 20x or 21x mults, not even at 1,6V Vcore

[msgclb]

I'd like to ask Eva Unit 02 a couple of questions but first let me say that my DFI LP UT X58 will ship in a couple of days. Since this is going to be my first DFI motherboard I've been doing a lot of reading to prepare for the occasion. The online manual that I downloaded omits the BIOS information. If it wasn't for your website I would be totally in the dark.

First, what's the purpose of 'Memory LowGap' and what determines the value used?

DRAM Timing
Memory Control Setting: AUTO
Memory LowGap: 1536M

Next, in the 'Voltage Setting' section I don't see how you set Vcore. There's 'DRAM Bus Voltage', 'CPU VTT Voltage' and 'CPU PLL Voltage'. Will I be surprised to find a BIOS section in the hardcopy manual?

I know that I've got a lot of other question but I'll stop with those.

I retired my 920 in favor of my 965 but I just might try it first.

920 batch #: 3837A902
965 batch #: 3838A610

[eva2000]

EVA...Looking at your settings I don't know what 'VID' is although it's referenced in giga boards also. Explain? After 5 weeks or so with my giga x58-ud5, seems it's pretty straightforward to get max OC. Just increase Vcore and Vtt and get higher CPU clk up to the limit of the chip. Is it pretty much the same with the DFI board, I'm asking cause a lot of your settings don't exist on the giga board, so I have no idea if they are critical or not. I want to leverage your superb efforts as much as possible so I can get to max CPU clk as quickly as possible when I get my DFI on Tuesday. Finally, did you obtain your CPUs in Australia?

CPU VID Control = Vcore



yup got my i7 920 from initial first shipment that landed in Australia

No, was using the x8 mem divider and x16 BCLK.
Tried VTT from 1,25 to 1,55V to no avail.
WTF guys...

@eva: It's a 3838A so brand-new. How do you set the fixed 21x mult? In bios I can only select the 20x max and the turbo 21x is there in grey.. I am guessing my CPU might just hate the 20x so much that it won't even get to run the 20x with mult set to 21x, EIST and Turbo enabled
Not even at 1,55V and -20C which is weird tho?

for 21x set Turbo and EIST to enabled

for fixed 21x multi, you have to set Vista Power managment to High performance ...

Yup in Vista depending on power management you need to set to high for fixed 21x.

Thanks... but I think I know what the problem is already
My CPU is just really really crappy... it can hardly do 4,2Ghz stable with 1,58V
I need a better one, and fast.

ah ha poor clocker won't help ya any will it

Same here: 3838A. Already on sale

what's your folks full batch number for 3838A???? so we can avoid these heh

I'd like to ask Eva Unit 02 a couple of questions but first let me say that my DFI LP UT X58 will ship in a couple of days. Since this is going to be my first DFI motherboard I've been doing a lot of reading to prepare for the occasion. The online manual that I downloaded omits the BIOS information. If it wasn't for your website I would be totally in the dark.

First, what's the purpose of 'Memory LowGap' and what determines the value used?



Next, in the 'Voltage Setting' section I don't see how you set Vcore. There's 'DRAM Bus Voltage', 'CPU VTT Voltage' and 'CPU PLL Voltage'. Will I be surprised to find a BIOS section in the hardcopy manual?

I know that I've got a lot of other question but I'll stop with those.

I retired my 920 in favor of my 965 but I just might try it first.

920 batch #: 3837A902
965 batch #: 3838A610

haha ya found Eva Unit 02 photos

straight from http://i4memory.com/102537-post3.html

Memory LowGap option is for mapping local memory (add-on-cards) to system memory address size. It most like AGP port Aperture size or PCIE1gen 250M fixed mapping size. Extend it can solve when if "add-on cards" request a bigger size as like we would like to enable Qual SLI on NV new 295, it must be extend over 2048M.

Can confirm that with 2x HD4870x2 in Crossfire I had set my LowGap to around 2176 at least. With Highpoint 3520LF PCI-E x8 256MB cache controller added into HD4870x2 Crossfire, I needed at least LowGap of 2560 otherwise windows boots to a blank window and doesn't go any further. As to how this effects windows available memory, with single HD4870x2 installed and 3x1GB memory, if I set LowGap to 2560, WinXP Pro SP3 32bit sees 1.5GB available memory. If I set LowGap to 1024, WinXP Pro SP3 32bit sees 3.0GB available memory IIRC.

As to vcore above photo CPU VID control main vcore options with CPU VID Special Add which allows additional volts on top of main vcore in percentage increments up to 14.8%

[msgclb]

straight from http://i4memory.com/102537-post3.html

It seems that I've got a lot more reading to do.

[vega22]

what's your folks full batch number for 3838A???? so we can avoid these heh

3838A335 - i940
In EX58 Extreme i have spent a couple of weeks with Linpack stability @ 19x211 1.38750 Vcore/1.28 IOH/1.34 Qpi/1.66 Vdimm. (20x200 tons of more volts needed). But for a couple of days the system has become insane. He has lost all its stability and is asking more and more volts. Board on sale (tomorrow arrives my DFI), cpu on sale, mems on sale...

[roller11]

I see a bad trend...my 3x week 41 were all the weakest. jcool is unhappy about his week 38, which is better than my 41's, but worse than my 2x week 37s. My week 37 is way worse than EVAs week 36. So 36>37>38>41. Seems that the later the chip, the worse the AC characteristics. At this rate, by week 44, we won't see any chip booting into windows above 3.7G. I can't think of any reason Intel would deliberately mess up the chip to make it run slower. I suggest that everybody start including their batch nums in their posts at least once so we can correlate week number with OC ability.

[jcool]

Mine is a 3838A685. I should have taken the 3835A but thought newer was better, duh.

Something interesting: My chip runs WAY better with HT off, even though I am obviously not temperature limited. 4,4 to 4,5Ghz easily 3D stable, the setting below even went through 10 mins of prime at 1,550V load Vcore:



This score any good for a single 4870X2? Haven't benched in ages really

[eva2000]

Mine is a 3838A685. I should have taken the 3835A but thought newer was better, duh.

Something interesting: My chip runs WAY better with HT off, even though I am obviously not temperature limited. 4,4 to 4,5Ghz easily 3D stable, the setting below even went through 10 mins of prime at 1,550V load Vcore:

http://database.he-computer.de/Bilder/temp/3d06i7.jpg

This score any good for a single 4870X2? Haven't benched in ages really

without HT you loose a bit in 3dmark06 at least - stock 4870x2 750/900 = 27,360k

[Solarfall]

i've been playing with this board for a few days now and its starting to look promising.. did this with watercooling hhahahahah MAN these things get toasty.

[eva2000]

i've been playing with this board for a few days now and its starting to look promising.. did this with watercooling hhahahahah MAN these things get toasty.

http://i163.photobucket.com/albums/t...tos/06937s.jpg

not bad.. enable HT for faster times though

Next up HD4870x2 CrossfireX with i7 920 still on TRUE120 air cooling

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 - 12/31 bios
• HIS HD4870x2 + Gigabyte HD4870x2 both flashed to Asus TOP bios - CAT 8.12 Official Drivers
• 3x1GB Samsung HCF0 based - 2x1GB Corsair DDR3-2133C9DF Dominator + 1GB G.Skill PC3-12800HZ
• Ram cooling: 3x 60x25mm Sunon 23.5cfm Maglev fans
• 750GB Samsung HD753LJ
• 4x32GB OCZ Core V1 SSD Raid 0 + Highpoint RR3520LF PCI-E Raid controller 256MB Cache
• Pioneer DVD-RW
• 1KW Corsair HX1000 psu
• Vista Ultimate SP2 Beta 64bit vLite'd

While my HIS HD4870x2 can do 850-860Mhz on GPU, my Gigabyte HD4870x2 hits a wall at <840Mhz, so settled for 837Mhz on GPU in CrossfireX. Weird GPUZ v0.31 isn't registering my ATI CCC set GPU/MEM clocks ??

System peak wattage draw = ~915 watts! (probably need a beefier psu ? )

• 3dmark06 @837/985 = 29,412
• 3dmark05 @837/995 = 37,175
• 3dmark2003 @837/1000 = 127,708 <- some reason I haven't been able to publish 3dmark2003 results with HD4870x2 lately ??
• 3dmark vantage = unable to bench keep getting create pixel shader failed message in cpu tests even at stock cpu/gpu settings ? Vista SP2 Beta 64bit related ?

[vega22]

I see a bad trend...my 3x week 41 were all the weakest. jcool is unhappy about his week 38, which is better than my 41's, but worse than my 2x week 37s. My week 37 is way worse than EVAs week 36. So 36>37>38>41. Seems that the later the chip, the worse the AC characteristics. At this rate, by week 44, we won't see any chip booting into windows above 3.7G. I can't think of any reason Intel would deliberately mess up the chip to make it run slower. I suggest that everybody start including their batch nums in their posts at least once so we can correlate week number with OC ability.

Very interesting. In my own experience, this is true. I removed the chip from the sale, it makes no sense to sell and to buy a newer and worse. Now we must wait (until summer?) to see if Intel launch new steps.

[jcool]

Really nice eva... my 4870X2 can't go too far either, seems 850/950 is max stable without artefacts - really crappy OC if you consider it never gets above 33C Core
Is there a way to increase voltages via software, like back in the X1900 days?

Anyway, here are a few quick & dirty pics of my testing setup if you are interested:

[Solarfall]

thats really nice clocking there eva yes i know that HT would give me better scores i was just playing around with the system ..
i will try again with my cascade i just have to finish up my new hold down for the evaporator and im good to go..

[cardnut99668]

I just ordered the DFI x58 board from Newegg, should be here Tues Jan 13. I'll be keeping my giga x58 ud5, so I'll be able to do a direct head to head shootout with same components, same Vcore so I'll report which board OCs better. I'll be glad to run any experiments on a request basis, just let me know.
i7 920
3x2gb Patriot 1.65v ram
2x gtx260-216 in SLI
Tuniq Tower 120

I'm looking forward to your comparison because I'm somewhat torn as to which one to get.

[Solarfall]

im seriously starting to love this board.. i hooked up my cascade today and was able to achieve 230 bus speed.


http://valid.canardpc.com/show_oc.php?id=483561


here is the bios settings that i used if any one else wants to try and beat me

DRAM Timing
Memory Control Setting: AUTO
Memory LowGap: 2048

DRAM Command Rate: AUTO/1N/2N : 1N
CAS Latency Time (tCL): 8
RAS# to CAS# Delay (tRCD): 8
RAS# Precharge (tRP): 8
Precharge Delay (tRAS): 24
REF to ACT Delay (tRFC): AUTO
Write to PRE Delay (tWR): AUTO
Rank Write to Read (tWTR): AUTO
ACT to ACT Delay (tRRD): AUTO
Row Cycle Time (tRC): AUTO
Read CAS# Precharge (tRTP): AUTO
Four ACT WIN Time (tFAW): AUTO

Voltage Setting
O.C. Shut Down Free: Disable O.C.S.D.F
CPU VID Control: 1.5=================with my Processor
CPU VID Special Add Limit: Enabled
CPU VID Special Add: Auto
Vcore Droop Control: disabled
DRAM Bus Voltage: 1.78 =============== with my Memory ( corsair dominator 1600 mhz ddr3 tri kit
DRAM PWM Switch Frequency: Reduction 15%
DRAM PWM Phase Control: 2 Phase Operation
CPU VTT Special Add: AUTO
CPU VTT Voltage: 1.55 Volt
VTT PWM Switch Frequency: Reduction 15%
VTT PWM Phase Control: 2 Phase Operation
CPU PLL Voltage: 1.80v
IOH/ICH 1.1V Voltage: 1.21v
IOH Analog Voltage: 1.20v
ICH 1.5 Voltage: 1.7v
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: 115
x DIMM 3/4 ADDR/CMD Bus VREF: 115
x DIMM 5/6 ADDR/CMD Bus VREF: 115
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: Disable
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: 20x

* BCLK/UCLK/QPI Controller Settings *
QPI Control Settings: Enabled
QPI Link Fast Mode: Enabled
QPI Frequency: auto
CPU Base Clock (BCLK): 223 Mhz
Boot Up CPU Base Clock: AUTO
PCIE Clock: 103 Mhz
DRAM Frequency: AUTO
UnCore Frequency: AUTO

CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled