**Official DFI LanParty UT 790FX-M2R Review/Overclock/Guide Thread**

[MrBallpoint]

I prepared going buy this board, but can i put regular waterblocks within the nb and sb?

Is there someone who will give me the dimensions of the mountingholes?

thx.

[DPeter]

Hi every one,

I have this board now for up to 3 weeks and having big problems with!
It just won't oc, what have I done:
first bios 2007/12/11 installed everything and booted up fine, install xp oke, tried to oc up to 250fsb no problem after that 260fsb no boot. Lowerd the mem to 400mhz and continued up to 278 fsb (3200mhz) htt multi 3 stock vcore, prime error after 20 minutes. 3100 still errors even @ 1.42 vcore.
Memory runs fine @ ddr1000 @ cpu 2500 multi 10 fsb 250.
But combining these things is just hell: 260fsb + htt multi 3 is boot with mem @ 400mhz divider. higher mem speed is no boot. higher vcore doens't help higher voltage on southbrigde doens't help either.
But now the most absurd thing, even @ fid 5x vcore 1.35 and 308mhz htt (multi 5) and memory @400mhz divider it won't boot. Lower htt (200mhz setting) won't boot either and upping vcore and southbridge voltage wont help also. With bios 2007/11/22 I can't even boot with fsb 225 and higher.
And bios 2008/1/30 it's the same as the original.
Also I can't boot from cd when my dvd drive is connected to the si3132 while it is in mass storage mode, the drive is spinning up strange and read light keeps burning. Is this normal not being able to boot from that controller with dvd drive (drive is a sony nec optiarc 7200s).
First I thought it might be my ram but 3oh6 in this forum has the same and is booting fine as I read (did you also use this ram with the 5000+ be?) it here.

So Please people help me is my board cooked or something and how do I prove it's faulty for rma when it runs stable @ stock except for the dvd drive issue.

[Oese]

I think your probs are similar to mine, just a bit worse..

I blame it on the imc.

could you boot 278 x 11,5 @ ddr667 divider?, so that mem is at around 800?

maybe your cpu is not the best one if prime errors at 3,1ghz @ 1,4v.. is it g1 stepping? no one would guarantee getting over 3ghz stable with such cpu, its not seldom that these chips dont go that well..

someone from forumdeluxx is testing on a gigabyte and is getting results that seem to indicate weak imc's as well..

btw: could you please post your everest mem speeds? I get >10000 @ ddr1066, but with half multipliers its only around 5000-6000 at comparable speeds and timings..

[justapost]

I haven't tried checking CnQ settings load watts. [added below]

Scripts don't work for me still. Debugging doesn't get far. RWE just freezes and you have to use Task Manager to kill it. Will need more time for it.

Your 1st power figures (2600) should read AC / DC.. you missed that but edit comes handy

Odd that the scripts dont work for you, it's damn neat that way. Gonna ask sam2008 for the source code of it's tlb-fix app.

...
I'm going to pick up the Gigabyte GE-S550A-D1 PSU soon and use it's P-Tuner software which is pretty damn accurate (known problem is +25W DC showing and 12V3+ below 180W DC is wrong) to show me the amps/volts each rail is pulling and work out the DC power draw by the CPU 8-pin.

That PSU is nice, might get one too.

I just tried everything again, found a bug in CnQ. Good news.

What bug? Did you post something about it in the MSI thread?

@others: sry, did not try an x2 cpu and using no watercooling here.

[KTE]

Yeah Achim I want him to make the source code available or just edit it himself to add MSR values.

PSU: Picking it up tomorrow. Would be good if you had one to compare if we both pull different currents/watts for the same volts/settings.

CnQ: If I change values whilst having CnQ enabled (through Windows+BIOS), CnQ doesn't fully kick in until I disable it within Windows and then enable it again. Loses me 3-5W AC idling.

As you know, I've found I'm still stable at lowish volts for 204x13x9x9x2.66, 94W AC idling now writing this and having at least 11 other apps open. Still at the same volts you can see in my 3DMark06 run I posted in the MSI thread. It's been running for over 35 hours now at this setting, many reboots etc, but hardly any downtime. I'll move it up slowly, testing all CPU/IMC/HT/RAM for instability is a bit of a long winded hassle.

[justapost]

Yeah Achim I want him to make the source code available or just edit it himself to add MSR values.

Well, found the WinRing0 library (http://openlibsys.org/manual/), should not be too difficult to write such an app with that lib.

PSU: Picking it up tomorrow. Would be good if you had one to compare if we both pull different currents/watts for the same volts/settings.

Just ordered one, p-tune looks promising. Hope i can run the monitoring stuff on a different pc (assuming it connects via usb or similar).

CnQ: If I change values whilst having CnQ enabled (through Windows+BIOS), CnQ doesn't fully kick in until I disable it within Windows and then enable it again. Loses me 3-5W AC idling.

I think this is normal behaviour, if the p-state changes the register get's copied to c0010070. Here it was enough to generate some load start/stop prime95 to force the switch.

As you know, I've found I'm still stable at lowish volts for 204x13x9x9x2.66, 94W AC idling now writing this and having at least 11 other apps open. Still at the same volts you can see in my 3DMark06 run I posted in the MSI thread. It's been running for over 35 hours now at this setting, many reboots etc, but hardly any downtime. I'll move it up slowly, testing all CPU/IMC/HT/RAM for instability is a bit of a long winded hassle.

Must say i'm ok with 2.6GHz at 1,3375V (1,325V just froze). Max i got stable for more that 4hrs is 2.63GHz(~1,39375V). 2.64GHz freezes after ~half an hour with volts up to 1,45V, which is too much for me to run 24/7.
Can not get my idle consumption below 105W AC.

[justapost]

Well, found the WinRing0 library (http://openlibsys.org/manual/), should not be too difficult to write such an app with that lib.

Visual C++ Express makes his way to my laptop atm. Already installed it on my phenom system and was able to build the OpenLibSys DLL. Samples require MFC so i can't build em with the express edition.
Never wrote a windows app in c++, gonna skim over a few tutorials first. Should not be too difficult.

[KTE]

Well, found the WinRing0 library (http://openlibsys.org/manual/), should not be too difficult to write such an app with that lib.

Hehe, excellent find. It makes it look so easy Need to remove a ")" from the end of that link to make it functional.

Just ordered one, p-tune looks promising. Hope i can run the monitoring stuff on a different pc (assuming it connects via usb or similar).

Not sure what you mean, but if you mean the PSU is used and monitoring one system while you're logged on another and seeing the metrics with the software, then I'm not sure.

I think this is normal behaviour, if the p-state changes the register get's copied to c0010070. Here it was enough to generate some load start/stop prime95 to force the switch.

Yep. But in my case, total P-State change didn't occur even if it was copied to c0010070 and c0010065 and c0010062 set to P1 unless I enabled and disabled CnQ again.

Must say i'm ok with 2.6GHz at 1,3375V (1,325V just froze). Max i got stable for more that 4hrs is 2.63GHz(~1,39375V). 2.64GHz freezes after ~half an hour with volts up to 1,45V, which is too much for me to run 24/7.
Can not get my idle consumption below 105W AC.

Wow, things are a little different but my volts are lower so that's why. I had to increase volts to get more stability but no need to increase IMC/NB/HT/RAM volts yet. I'm at this since around 12 hours now for stability:



One notch less in volts (1.264) would fail on core3 after 30 mins while other carried on past 4 hours smallFFT.

CnQ Power: Idle 93W AC
No CnQ Power: Idle 105W AC

It's getting better with optimizations, pretty weird.

Visual C++ Express makes his way to my laptop atm. Already installed it on my phenom system and was able to build the OpenLibSys DLL. Samples require MFC so i can't build em with the express edition.
Never wrote a windows app in c++, gonna skim over a few tutorials first. Should not be too difficult.

Yea I don't like clogging my drives up with big software etc, not even got word/etc installed on these. Only the basics. IMO writing C++ is boring but that's not the worst bit, deassembling them is the biggest pain.

BTW, Sapphire board I may skip testing. I'm too short of time and got the Abit 770 and GBT 780G as well to test yet.

[LIKMARK]

LOL my Crosshair idles at 165w with Phenom 9500@stock and 7900GTO@stock, one hardrive, no cnq enabled. (275W max under load, gaming) I'm getting the abit 770 board too in a few days, looking forward to testing it

[Sparky]

Quick Q, surely having the phenom black edition means I can take the multiplier of the CPU up a notch from 11.5 to 12, surely it can handle that right? It posts no problem, but it freezes on its way into windows!

Oh and a side note, my system hung yesterday randomly. Came back from work and found that the monitor wouldn't wake up

[DPeter]

I think your probs are similar to mine, just a bit worse..

I blame it on the imc.

could you boot 278 x 11,5 @ ddr667 divider?, so that mem is at around 800?

maybe your cpu is not the best one if prime errors at 3,1ghz @ 1,4v.. is it g1 stepping? no one would guarantee getting over 3ghz stable with such cpu, its not seldom that these chips dont go that well..

someone from forumdeluxx is testing on a gigabyte and is getting results that seem to indicate weak imc's as well..

btw: could you please post your everest mem speeds? I get >10000 @ ddr1066, but with half multipliers its only around 5000-6000 at comparable speeds and timings..

278x11.5 @ ddr667 won't do any good need to go to ddr533 and lower htt multi to 3x to get it stable. Btw my cpu is a g2 brisbane so I would have expected it to atleast do good @ 3.0ghz but thats even a problem. Speed I can't test right now pc is @ my parents.
About the optical drive problem on the si3132 does any one have similar probs?

[KTE]

Sparky, should easily be able to do 11.5 to 12 multi change. Bump the V to 1.265V and check, if it fails, you most likely have other problems somewhere.

Can you give us a ss of what AMD Power Monitor along with Memset 3.5BETA and CPUZ CPU/SPD tabs show?

LIKMARK, 165W AC idle is mostly due to your card+chip I'm gathering. I have more peripherals connected than you and have to increase Volts to 1.29V idle now because after 15hrs, trying SupCom, it failed after 1hr37m play at the above post#608 setting, and still, I'm idling 106W AC without CnQ and 95W AC with tweaked CnQ.

I'm wanting to trying 5 boards ATM (RD790: Sapphire/Jetway, ASUS Crosshair II, Abit AX78, Gigabyte 780G) and give feedback and I can pick all of them up by Friday evening, incl the ASUS with 10 phase PWM ()... problem is actually time and space. I'm too overloaded nowadays, I haven't even managed to post 8 different and very large testings I've done with K7.5/P4C/P4/X2 BE 45W/X2 G2 BE/C2D G0/C2Q G0/X4 95/96/96BE since 6 weeks now, not even arranged the +600 odd pics, they're scattered all over UK on diff. systems, then theres the in-depth memory test I ran similar to that recent article posted here but at least 4 weeks before it and not yet posted, incl not answering around 104 old PMs, 19 vital member posts and 24 old emails yet and skimping on some lengthy testing for software devs and site reviewers.

It's why I 'aint really doing much, and also dropped writing an intro AM2+ guide linking to 5 in-depth walk through guides for Phenom systems and overclocking on different boards, setup, requirements, best procedures, known issues, experiences, oddities, expectations, base performances, performance tweaking 'n' tricks (exact values and their effects), limitations, stability testing and how to deal with them, especially the biggest headache causers and things for newcomers, incl. which CPU (Phenom/C2Q) is better for which sort of real-life segment and which program (can bench 2.8GHz Phenom and 4GHz C2Q all the way down to air to show each application). There's a big difference between oc now and before. Before with C2Q, quads were either the market for those who spend on nothing but computers, rich, very limited specific need segment or are too into them beyond reality or need so don't mind waste credit card purchases to pay-off over the next year. Now quads are open to near enough everyone of the general public, again this because of AMDs mainstream competition and pricing, to whoever can afford any standard single core PC, and so a lot more newcomers are getting them, want to oc and get started on them, especially with the 9500/9100e and the various 770/780G/690G chipsets incl. the MSI/DFI 790FX and their prices. You can't exactly write "MSR001_0070 [32:29] in CCPUID MSR Editor and you're done!" but you have to write towards the intended audience most of whom know nothing of that sort.

Daily IT onliner, gamer, mainstream, overclocker/bencher, budget, general desktop, HTPC, science calcs, rendering/multimedia are different segments, with different needs, perspectives, uses and thus tips and advices.

[GunterFalstaff]

hey guys, i have the dk version of this board and i am having some problems with getting into bios or (doing anything prior to os boot) with a usb keyboard... so far the only thing that works is an old ps2 keyboard for when i ever want to change anything in the bios... has anyone else had this problem and is there any fix for it? thanks! >.<

[Oese]

@ DPeter: Sounds exactly like mine... higher mem dividers make the system not to boot or fail in prime at clocks both cpu and mem are stable at testing them each for themselves..

i really think its the imc. i dont know whether some tweaking with the dram driver settings and/or timings will do something good but i have little hope..

the imc's in the k8 am2 cpus are simply ddr1 imc's that were little modified to run ddr2, but they are at their ends clockwise, speaking literally..

its a shame, ddr1100 would be very nice but its not possible..

all you can try is playing with the dram settings or testing another cpu..

and wait for phenom b3..

[Final8ty]

hey guys, i have the dk version of this board and i am having some problems with getting into bios or (doing anything prior to os boot) with a usb keyboard... so far the only thing that works is an old ps2 keyboard for when i ever want to change anything in the bios... has anyone else had this problem and is there any fix for it? thanks! >.<

Thats the norm with most DFI motherboard & bios version on them.

[Sparky]

Sparky, should easily be able to do 11.5 to 12 multi change. Bump the V to 1.265V and check, if it fails, you most likely have other problems somewhere.

Can you give us a ss of what AMD Power Monitor along with Memset 3.5BETA and CPUZ CPU/SPD tabs show?

I'll try that. Here is the screenshot you wanted. I'm trying to figure out why refresh cycle on DIMM 0 is 105ns while the rest are 75ns Also it appears CPU-Z is reading the SPD for my memory wrong or the memory SPD is wrong, should be 4-4-4-12 2.0V or 2.1V at 400MHz, not 5-5-5-15 1.8V.

[Oese]

a question: how about temperature read out using a phenom? core temps/cpu temp within reasonable range??

my windsor gives much too high cpu temps, the brisbanes core temp readouts are much too low..

@ DPeter: You use a 2x2GB Kit (like me), right? I think this is causing some more stress on the imc too..

[justapost]

Hehe, excellent find. It makes it look so easy Need to remove a ")" from the end of that link to make it functional.

It is easy to use if you are familar with c no cryptic c++ inside.
THX link fixed.

Yep. But in my case, total P-State change didn't occur even if it was copied to c0010070 and c0010065 and c0010062 set to P1 unless I enabled and disabled CnQ again.

Ah ok, that makes a difference. Noticed here that without phenom drivers installed only the multi changes and the vid stays.

Wow, things are a little different but my volts are lower so that's why. I had to increase volts to get more stability but no need to increase IMC/NB/HT/RAM volts yet. I'm at this since around 12 hours now for stability:
One notch less in volts (1.264) would fail on core3 after 30 mins while other carried on past 4 hours smallFFT.

1,3375 CPU VID froze in the morning exactly as i powered on the monitor and tried to move the mouse. 1,35V froze after a few minutes, had it stable at that voltage for 10hrs already. Restored my save 2.5GHz settings (higher nb and chipset voltages and mem in 800 mode).
p-state tweaking does not help in stability here, maybe if i also touch idd stuff. But it saves ~30W AC here, lot more than on your system. Also the system is still pretty snappy at 800MHz for normal tasks

Yea I don't like clogging my drives up with big software etc, not even got word/etc installed on these. Only the basics. IMO writing C++ is boring but that's not the worst bit, deassembling them is the biggest pain.

Hope i can port that stuff to msys or cygwin once i'm finished.

BTW, Sapphire board I may skip testing. I'm too short of time and got the Abit 770 and GBT 780G as well to test yet.

No big deal, with that PSu we can use the exact same testing setup's, so no need to do everything twice.
Those 780G boards are only listed at a few sites but not jet at stock here. Hope I can get the J&W RS780UVD or something similar with side port memory.

[GunterFalstaff]

Thats the norm with most DFI motherboard & bios version on them.

so you're saying theres no fix? why would a brand new motherboard not support usb keyboards? usb has been around a loong time...

[MotF Bane]

I just finished mounting the board into a case, water cooling on processor and graphics, NB/SB/mosfet heatpipe remounted with MX-2... and first it said Boot disc error, which didn't make any sense, but then after rebooting twice, it went to post code 88. Any guesses?

[3oh6]

First I thought it might be my ram but 3oh6 in this forum has the same and is booting fine as I read (did you also use this ram with the 5000+ be?) it here.

sorry for the late reply...just saw this post and happened to catch my name in there.

i haven't done a whole lot of overclocking the 5000+ with the OCZ2P10004GK in the 790fx yet as i am running a mountain of benches for a review but will get on it shortly. i am running the OCZ kit with the 5000+ right now at 255*11/467MHz and have this screenshot on the 10X multi with the /5 divider at 520MHz 5-5-5...

http://img.photobucket.com/albums/v2...0_hci1hr-1.jpg

will get into more systematic clocking of this setup when i get my work done, but it won't be for a little while yet. HTH

[Final8ty]

http://www.bit-tech.net/news/2008/03...er_version_2/1

[MotF Bane]

http://www.bit-tech.net/news/2008/03...er_version_2/1

That's an Intel board, unfortunately.

Anyways, my board is going for RMA.

[Final8ty]

That's an Intel board, unfortunately.

Anyways, my board is going for RMA.

I posted to point out that a new cooler for the NB will be available.

[DPeter]

Oese I haven't yet been able to test 1066 @ half dividers cause I've taken it apart just one more time.
But when it's up again I will check for you.
About the stress on the imc with 2gig modules, do you think the imc and htt controller are connected?
I mean will the 2gig modules cause the low max htt (305mhz x5) even @ low mem speeds just because the imc is having a hard time managing 2 gig sticks?
Can you give me the link of the forumdeluxx gigabyte thread?
It's nice to have some one with the same problem cause I really feel like with this brisbane! (2.88ghz is not stable in vista 64 )

3oh6 thanks for your info, when your ready to go for some more oc-ing, could you check your max htt speed with these sticks? just to check the above theory! Btw could a motherboard grounding issue also be able to cause this problem?(it's fully stable @ stock)

If it's true than I will be the first to run a Kuma dual core K10.