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

[i7luver]

hey eva so the power and hdd led situation is not normal right? I might just rma the whole board then, I am using a OCZ elitexstream 800w psu and HAF932 case, reasonably good quality components i think, my ram is OCZ Gold 12800 3x2gb kit, one of them's dead for sure, the other two seems fine until i flashed it, should i rma with dfi or my e-tailer? its 1 day old...

[PRJ]

Maybe this is'nt possible with the new flashable USB method

What we did then, as i remember was...

• Booted up working computer
• Change Bios chip on running motherboard
• Found the correct DFI bios
• Used DFI flash utillity, running on XP to flash
• Flashed
• Changed Bios chip back again

[eva2000]

Power led (above PCI-E #2 slot - PAGE 27 of manual) should be on as long as psu is connected and psu switch is on even if system is powered down - as it's a standby power led. I thought you meant power led is on after psu power is cut off. Or you mean your pc cases led ? not sure about that as i haven't used pc case in years LOL

non detecting of triple channel (report as dual channel) when 3 modules is installed is usually sign that vdimm and cpu vtt aren't correctly set for your mem frequency/timings

[The Watcher]

[QUOTE=eva2000;3644286]965 has unlocked multi so you can use any multiplier without turbo mode anyway so set vr current limit max won't effect 965 in terms of turbo mode although i guess tdp and tdc will still be in effect ? As to SG no idea why never seen such a problem before sorry

NP EVA thanx for the quick reply. well then i should try to intall a new fresh windows, and see if it work.

[i7luver]

Power led (above PCI-E #2 slot - PAGE 27 of manual) should be on as long as psu is connected and psu switch is on even if system is powered down - as it's a standby power led. I thought you meant power led is on after psu power is cut off. Or you mean your pc cases led ? not sure about that as i haven't used pc case in years LOL

non detecting of triple channel (report as dual channel) when 3 modules is installed is usually sign that vdimm and cpu vtt aren't correctly set for your mem frequency/timings

Yea i know about the standby led, I meant the case led, but seems like you won't know :P my brother's board does the same thing, is this a widespread issue? I'm not sure if its a triple channel problem as that single stick by itself won't even post, i was diagnosing the problem when i decided to flash it and now its stuck at C1

[OverDozeD]

eva2000
thanks
recently I saw EVGA X58 @ 241 BLCK,
Is it WR or you can push it more?
coz T3eH8 seems can brake almost all records yet

[eva2000]

ways to improve bclk

1. better cpu cooling, sub zero > water > air
2. more uncore, vdimm and ioh/ich voltages
3. better cpu itself
4. memory used and memory settings
5. motherboard design/bios

so runmc's eVGA X58 241 was on sub zero cooling with ioh at 1.45+ v and 1.6 uncore IIRC and only validation

Batch - 3837B512
vcore - 1.7v
vdimm - 1.80v
qpi-pll - 1.50v
cpu-pll - 1.80v
ioh-vcore - 1.45v
cpu vtt - 1.6v
ioh/ich/io - 1.55v
ich vc - 1.05
Single stage evap temps held around -40c. Room temp was 78f today

My 228bclk is super pi 32m stable and validation 231bclk all on air cooling with ioh at default 1.11v bios (dmm real same) and 1.5v uncore volts for i7 920 #2 and i7 920 #1 need 1.6v uncore (illustrates differences among cpus).

I can probably push bclk higher with right memory and voltages and sub zero cooling instead of air cooling

[NBF]

ways to improve bclk

1. better cpu cooling, sub zero > water > air
2. more uncore, vdimm and ioh/ich voltages
3. better cpu itself
4. memory used and memory settings
5. motherboard design/bios

so runmc's eVGA X58 241 was on sub zero cooling with ioh at 1.45+ v and 1.6 uncore IIRC and only validation





My 228bclk is super pi 32m stable and validation 231bclk all on air cooling with ioh at default 1.11v bios (dmm real same) and 1.5v uncore volts for i7 920 #2 and i7 920 #1 need 1.6v uncore (illustrates differences among cpus).

I can probably push bclk higher with right memory and voltages and sub zero cooling instead of air cooling

eva2000, since you tested Micron, Samsung and new Elpida RAMs, what are your findings so far about the memory x bclock relationship?

[eva2000]

eva2000, since you tested Micron, Samsung and new Elpida RAMs, what are your findings so far about the memory x bclock relationship?

Everything laid out at http://i4memory.com/showthread.php?t=12855 and particularly http://i4memory.com/102537-post3.html

Basically bclk first and foremost is cpu related. Then finding the right vcore, uncore, vdimm balance for memory stability will go along way - but each type of memory has their own balance for voltages even amongst same ic types.

[StimulatedBoy]

Hi Eva2000,

Got board installed last night and are now up and running. No luck until flashed with BIOS 0206 for some reason but now fine on default set up at least. Really wish DFI had a BIOS upgrade procedure like ASUS EZ flash in BIOS but thx to StoolMan for creating the bootable USB stick.

BIOS setup is kind of different compared to my previous ASUS boards and I am wondering if CPU Non-Turbo Clock Ratio is the same as Multipliers? Does not go beyond 24 for me. Have a i965

Again thx for all the information and knowledge you have provided us with in the various threads.

[Sam_oslo]

[QUOTE=The Watcher;3644320]

965 has unlocked multi so you can use any multiplier without turbo mode anyway so set vr current limit max won't effect 965 in terms of turbo mode although i guess tdp and tdc will still be in effect ? As to SG no idea why never seen such a problem before sorry

NP EVA thanx for the quick reply. well then i should try to intall a new fresh windows, and see if it work.

You can try reinstalling th SG first. I had the same issue with SG showing only 0 after un-installing ABS-crap, but reinstalling the GS solved the problem.

[faster3200]

I have a stupid question for the collective here . What does it mean if increasing vDimm and VTT past a certain point cause mem instability (vDimm past 1.6v) and raising BCLK causes mem instability if if you loosen timings and lower the mem divider? I can pass LinX just fine but memtest fails pretty quickly.

[The Nemesis]

ways to improve bclk

1. better cpu cooling, sub zero > water > air
2. more uncore, vdimm and ioh/ich voltages
3. better cpu itself
4. memory used and memory settings
5. motherboard design/bios

so runmc's eVGA X58 241 was on sub zero cooling with ioh at 1.45+ v and 1.6 uncore IIRC and only validation





My 228bclk is super pi 32m stable and validation 231bclk all on air cooling with ioh at default 1.11v bios (dmm real same) and 1.5v uncore volts for i7 920 #2 and i7 920 #1 need 1.6v uncore (illustrates differences among cpus).

I can probably push bclk higher with right memory and voltages and sub zero cooling instead of air cooling

My Max BCLK seems to be 233-234, My max boot from bios bclk is 220 and my max stable bclk is 229 1 mb pi only. My voltages all read .05v or greater over in windows, but 1.65vcore and 1.6vtt gets me over 4.8GHz PI stable, still working an 3DMark Stable clocks.

[jcool]

Holy crap nemesis, thats one insane 920.
I finally found a decent 920 btw, doing prime at 21x200 with 1,42V on water, so depending on the max BCLK it should be a good candidate for my SS
Good old 3835A it is.

[ReverendMaynard]

My Max BCLK seems to be 233-234, My max boot from bios bclk is 220 and my max stable bclk is 229 1 mb pi only. My voltages all read .05v or greater over in windows, but 1.65vcore and 1.6vtt gets me over 4.8GHz PI stable, still working an 3DMark Stable clocks.

holy snappin arse-hole man that is fantastic.

What are you using for cooling btw?

[StimulatedBoy]

[QUOTE=Sam_oslo;3644857]

You can try reinstalling the SG first. I had the same issue with SG showing only 0 after un-installing ABS-crap, but reinstalling the GS solved the problem.

Have to admit that I don’t understand what you are trying to tell me. I have not installed ABS after Eva2000 advice. What is SG and how should I set it? The damn i7 systems kind of brought me into unchartered territory, so have no clue any longer... Yes I know I am a noob!

[The Nemesis]

holy snappin arse-hole man that is fantastic.

What are you using for cooling btw?

cooling is ss phase -54c idle -41.5c load. Last 32MB pi i ras was @ 4752MHz but it was a lot less volts. It seems my board has run out of blck and is not quite as good as eva's even with extreme cold.

[Sam_oslo]

[QUOTE=Rob Halford;3645164]

Have to admit that I don’t understand what you are trying to tell me. I have not installed ABS after advice. What is SG and how should I set it? The damn i7 systems kind of brought me into unchartered territory, so have no clue any longer... Yes I know I am a noob!

That was a good advise, but i didn't see it. I installed ABS-crap and it didn't do anything good, but has written some craps that i can't remove. I believe it is a good advise, stay away from ABS.

I see it is not easy to catch all ABC-stuff on a new platfrom, SG is Smart Guardian, the monitoring tool that comes with DFI. SG has been usefull.

[jcool]

Guys,

something is definitely off with my DFI x58. It can't be a general problem because you guys are getting sky-high ram OC figures, but I can't OC the mem for .

Now I tried varies CPUs and Ram kits of course, but to be really sure I did the ultimate experiment today: Pulled a 920 and 6GB of DDR3-1600 out of a GA-EX58 Extreme that was running 200x19 with 8x mem divider (1600 9-9-9-24) with 12GB of ram for the last few weeks. Rig is 53h prime blend stable and passed 12h+ of LinX using 11GB of ram, so it's as stable as it gets.
VTT was 1,35V bios, Vdimm 1,62V. Cooled by water it ran around 65C on all cores.

Now I went ahead and put that exact CPU and 3 of the 6 mems in my DFI. Set the exact same settings, and sure enough.. I get a reboot while trying to boot into vista. Setting the mem divider down to 6x "fixed" the issue once again. Running memtest, I could not detect any problems, but that has always been the case for me. Memtest can pass fine all day while Windows won't even load or freeze/BSOD after a few seconds.

Now I tried setting the subtimings manually to very relaxed values (used eva's recommendations from earlier), still no go. The only thing that seems to help is raising the Vdimm, which is BTW severly overvolted on my board, as is VTT. At around 1,8V bios I could get a boot into windows, but the system still was pretty flakey. Prime blend froze after a few minutes.

Right now I am back to my Diablos, running 210x21 with a 3837A707 920, pretty good CPU. LinX stable at 1,52V load Vcore. VTT (1,48) and Vdimm (1,8V) are pretty high, but at least it's stable now. Again, using the 6x ram divider.

Any suggestions what else I could try? I am drawing a total blank here guys. Why do I have to use so much Vdimm for very low OCs if I can pass memtest fine even at 2Ghz ramspeed?

[The Nemesis]

@Jcool,

Have you measured ypur voltages. All my voltages are reported at least .05v over in windows vs whats set in bios. When I measure them they are actucally pretty close to whats set in bios. This caused me all kinds of headache because I thought I was overvolting when actually I wasn't giving enough volts for the clocks I was trying to acheive. Also, your vtt/qpi will need to be 1.5v-1.55v to start to run high memory clocks.
***edit***
Also there's a funky relationship between vtt and vcore on this board for it to post depending on cpu. For eva it was .27v- .31v on one cpu on the other is was less. For me it's .20 or less. You'll need to get that figured as well.

[jcool]

@Jcool,

Have you measured ypur voltages. All my voltages are reported at least .05v over in windows vs whats set in bios. When I measure them they are actucally pretty close to whats set in bios. This caused me all kinds of headache because I thought I was overvolting when actually I wasn't giving enough volts for the clocks I was trying to acheive. Also, your vtt/qpi will need to be 1.5v-1.55v to start to run high memory clocks.
***edit***
Also there's a funky relationship between vtt and vcore on this board for it to post depending on cpu. For eva it was .27v- .31v on one cpu on the other is was less. For me it's .20 or less. You'll need to get that figured as well.

Thanks for your efford nemesis. I too was wondering WTF went on about this overvoltage thing reported in Smart Guardian, so you are saying SG reports crap and the voltages set in bios are pretty close to actual? I sure hope so.. gotta measure it, where can I find the spots to measure on?

The strange correlation between VTT and Vdimm I noticed as well, for instance if I go too far up or down with Vdimm while leaving VTT the board will only post showing 4GB instead of 6. The range on my current 920 seems to be 0,25-0,3V difference between the 2.

And yeah I tried 1,55V all right, went as high as 1,65V with some CPUs, to no avail. And remember I am not talking about 2Ghz+ here just stupid 1600Mhz. Impossible with normal/low volts... if I feed it 2V Vdimm then sure it'll do it.. but I sure as hell ain't runnin 2V for 1600Mhz.

It's also not the uncore speed, I just upped the uncore to 16x at 210 BCLK and its still linX stable at 4,4Ghz, no change whatsoever. Changing the ram mult to 8x causes the system to freeze on boot

I'll maybe play with it some more tomorrow, and I wouldnt be complaining if I had trouble hitting 2Ghz+ ramspeed, but we are talking officially supported 1600Mhz here. Somehow I feel like an idiot running 4,4Ghz with HT rockstable but running the memory at 1200 something

And I ain't exactly new to OCing or to DFI either, to say the least... tried 5 different CPus so far and 3 different mem kits. Maybe I need to try a kit with XMP profile or something, cause the ones I tested were all CSX and they don't have XMP or anything.

[The Nemesis]

I understand your feelings, this board has been a headache for me too at times. I've decided I'm gonna pair it with my i7 920 3838A and get the evga classified for my benching as soon as its available. I may even pick up the current evga x58. My 3838A chip clocked much easier in a gigabyte board than this dfi so I'm thinking the board is a little fussy.

[The Watcher]

[QUOTE=Sam_oslo;3644857]

You can try reinstalling th SG first. I had the same issue with SG showing only 0 after un-installing ABS-crap, but reinstalling the GS solved the problem.

Tq Sam, but i've tried that also before posting it and still no luck with that. actually it doesn't concer me so much i just curios, that's all. i don't really need it cos it's way better to make sure with the multimeter volt, i think it's way more accurate.

tq for all the help.

[StimulatedBoy]

Will running VTT of 1,5 or thereabout 24/7 damage anything or is that acceptable? Thx for any input.

[jcool]

No idea, but I am currently running 1,48V bios as well. Tried lowering it, not happening at 210 BCLK or I lose stability.
If you can you should try to lower it to around 1,45V.. if not just pray nothing happens ^^