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

[Brother Esau]

[unkn0wnhacka]

I would like someone to post all of their m2rs bios settings with the bios version.

I cannot get mine any higher and i know i can. Im trying to raise voltages and stuff to get it higher but all they do is add more heat lol.

Even with 1.48 vcore i am only at 35c


gimme gimme

thanks

[aGeoM]

Maybe people are forcing their Phenoms and pushing higher just in the hope that they'll clock stable higher than they really will?
I mean if you run 2.8G, test it, while that's actually unstable for the CPU, then no matter what you try, it will be unstable, even though your overclocked heart can't accept to lower it anymore. Just because you benched it doesn't mean it's stable or doable MHz
Yeah... like I said earlier, you're too rich for me

BTW, I have 2.8G stable in 32b and 64b, not tried more yet.

I'm not, just sold my 5000BE+250 HDD+80 euros(beer savings)=Crosshair II

now DFI/9500 is 2.8 stable (Benching+Gaming+Surfing+Prime) WinXP x86/x64, Vista/Server2008 x64.

aGeoM given up on the 6400BE ?

Never had one, but no more Duals for me

[AlabamaCajun]

Yea. I posted it about a month back, Ubuntu 64b is 100% stable for me for a while now. Infact I'm reinstalling it at those settings.

Problem is it needs high voltage, 1.480v for 2800MHz to be stable. Doesn't give me any error at that and at 2700MHz 1.352v. Haven't tried anything else nor did I change any other parameter from stock.


Nearly all settings of mine used and what not have been disclosed across the AMD section of this forum but I haven't replied to a concern of yours that I can remember you asking or wanting to clarify. In case you haven't read before a month, my 9850 has had no boot since about a month on all boards I had - until now. What is it you want to know exactly?

Thanks Bro but you know you don't have to

Sorry if that came off a little harsh, I was more just trying to for a little info on your clocks. Above, you answered my curiosity. This is a well informed board with a lot of power hitters, that I respect. I look at sigs as a reference, it puts a personality with posts making a persons post mean more. I change parts every month so I'm updating regularly. (BTW Nothing free here, internet sales pay for these ). I certainly hope it helps people to know where I stand when I answer a question or post a result.

I did some reading from AMDs tech pages, these little power munching beasts are seriously complicated with a lot of parameters to keep then moving. Some of the settings for the RAM could easily take it down if not tweaked right. I will read up on it when I have the time. I now see where some of these timings are coming from but I'm not up on what to do yet. Might be some tweaks in the NB and DRAM controller that will help.

[KTE]

No problem
It's quirky to work with Phenom trying to get it stable.

My update: http://www.xtremesystems.org/forums/...postcount=1840
Hit post there before here, so might as well.

Not 64bit though - that's on a different drive and fudged up with 800x display size.
I'm running 800 RAM, its much more stable and is the only way for me to avoid C1 errors at boot. Not 800 mode but actually 800 rated RAM and not 1000/1066 rated. Everything else but CPU multi is at stock.

AMD hasn't provided drivers for lm-sensors but they did say they're coming, so we'll be able to get some info through them. I can't get CPU frequency scaling or changing to work for some reason, it did on older CPUs.

See this, in WinXP 32b, this is easy for my 9850: http://www.xtremesystems.org/forums/...&postcount=935
AOD full load for about 2 hours starting from stock, this is what it finds stable->
Whilst working: http://picsorban.com/upload/au2.png
Final max stable: http://picsorban.com/upload/3055.png
These are my current settings, multis, voltages and VIDs that I've changed are all there (Sandra Java benchmark):
http://picsorban.com/upload/mt-ar-2.8-2.4a.png
http://picsorban.com/upload/mt-jv-2.8-2.4b.png

But... now for a Mesonychoteuthis hamiltoni sized spoon of reality;
None of those are truly stable apart from the 2.8G setting which I'm still testing!
Yes yes, the normal trash happening: easily bench/load stable, freezes idling or during common tasks. The critical freezing period is anywhere between 0.1 to 28 hours.

And yes, the normal drum of Prime95, Sandra, EVEREST, OCCT, AOD, Systool, CPU Stablility Test, MaxCPU stability testing all fail, none catch any errors in this stability range; something we found with the first B2 9500, mine could bench 2832MHz easily and repeatedly but below that was showing the same freezing issues all the way down to 2615MHz.
I even ran FAH and WCG for 24 and 5 hours on the 2.975GHz setting about 5 weeks back on stock cooler without TIM just to see how good they are for catching Phenom B3 errors with an overheating core, but none of those really test stability anyway and as expected, even they fail to find any errors. They're just for higher than average heatloads, my B2 and B3 Phenom intentionally at an unstable setting ran WCG fine for 5 days and that setting failed P95 within 20 seconds.

As a brief recollection of the ones I tested, my unhappy findings;

9500 = 2860MHz max - 2832MHz max bench - 2772MHz max load stable - 2615MHz max fully stable
9600 = 2622MHz max - 2599MHz max bench - 2553MHz max load stable - 2553MHz max fully stable
9600 = 2599MHz max - 2553MHz max bench - 2530MHz max load stable - 2530MHz max fully stable
9600BE = 2560MHz max - 2525MHz max bench - 2500MHz max load stable - 2472MHz max fully stable
9600BE = 2940MHz max - 2849MHz max bench - 2805MHz max load stable - 2656MHz max fully stable
9850BE = 3155MHz max - 3100MHz max bench - 3059MHz max load stable - 2700MHz max fully stable [still pending]

All the ones having a different max load stable to a max fully stable showed idle freezing problems.

So I'm left trying to figure out what is causing them all this "idling" instability, why and how to avoid it -- so far with multiple Phenoms, I've not found a way, on all motherboards I've tested and on all the CPUs, the problem existed. I am trying to cut it down and work things out, tested everything but the following; either to do with MRL of DRAM or the NB volts needing to increase as you increase CPU MHz [as Achim found]. I might have a Toliman soon to see if that also has it (or is it that core3..), depends on time.

Only one I recall never showed the freezing phenomena was ASUS 790FX and I've not heard anyone on that board report it either.. but my time with that board was very limited and short, since they had poor early BIOSes.

[Rammsteiner]

Interesting KTE.

I thought at least about MRL... I think for example since the MRL can be variable, we shouldnt use Memtest since that forces the crrent applied MRL. Although if MRL doesnt change during uptime it's not an issue, but if it does change will PC is on it's a bad idea.

I think also there's something wicked going on. And it has definately something to do with Voltages and NB is very suspicious too. I downclocked NB to 2.2Ghz and set both CPU and NB to 1.35V and got a freeze in about 10 minutes 24/7 usage... That's just like impossible (was running 220HTT, 12.5 CPU multi and 10 NB multi).

Also if you didnt read this yet:http://www.xtremesystems.org/forums/....php?p=2957907

Might be interesting that it might be the SB600 causiong this stuff. I mean, in general freezes are caused by just a notch of Voltage short. If it's too less Voltage it either wont boot or BSODs (AFAIK?). But this freeze... what about just something causing instability. And since we see it almost on every board and Phenom this might actually be true.

KTE, did you try the Phenom on a 780a chipset yet? I wouldnt like to get a 780a though since Im running CF. But if it was just to check if it's an issue with SB600... Not a lot of people have the 780a yet and AFAIK everyone who suffers from freezing actually runs a 790FX board. Though it's strange the Asus one doesnt seem to suffer from it. They dont use reference boards though, but so doesnt DFI.

[Rammsteiner]

Also, KTE, those REAL stable values, did you notice that the NB had to be at 2Ghz or was it afterall the CPU speed what caused the freezes if over a certain point?

[KTE]

I've kept the NB at 2GHz all the way, just to eliminate all factors. Usually I run 10x, 11x and 12x on NB at each setting just to check.
2.4GHz NB didn't freeze any quicker than 2.0GHz did at a idling instable settings.

Voltage - if it Primes/EVEREST tests and benchmarks fine, then something is iffy and something other than the CPU itself can be the cause, very possible. You have to have those basics in place though.

I've not played with a 780a yet, nope. Wanted to buy ASRock 780G for home but I just missed out, they sold out within 24 hours.

MRL - reason is this. At times I have booted low MRL and had higher bw then at other times. When I was testing across BIOSes and boards, the first thing I tried to keep the same is MRL, due to its high bw effect. Problem was, at the exact same settings and hardware but on a new bootup, you may not be able to get the same MRL value back without crashing. Hence it "could" be a cause for some people.

For me, I doubt it. I've tried upto 62 MRL at 800, which is high enough when stock is 51. The freezing continues...

[Rammsteiner]

Oh well,

As you can read in my sig Ill wait for the SB750. Im done with this crap really. Im around 2.65Ghz now and will see how that performs. If it's stable Ill add now and then 1Mhz to HTT. Since this is my 24/7 system as well I prefer stability a lot lol.

If SB750 launches and it proves to solve everything Ill be the first one in the line to get one for sure and do some new testing. But atm we're just 'fighting' something about we dont know why and what exactly causes it to be instable.

[AlabamaCajun]

Woe, might have something there I have a 9500 that was a pain on the GBT790FX dropped it into the ECS740G and it is rock solid with Vista 64, albeit stock. I never saw that stability on the GBT. I did tun the 9500 on a 770 board w/SB600 OCed to 2.39 for months but the it started acting up. About that time I got a 5000BE and replaced it.

What is interesting with the 790 and SB600 is that when I see the crash, it's at a point of loading an app or booting. I'm still at stock with folding running 20 hours a day while I surf, watch Vids etc. I shut off folding when I run the 3D stuff and games. Ran 2 hours of gaming and some simulators then put back up to fold. It's been solid with no issues now for 2 days. I'm going to go for 2.7G and just leave it there. The only problem I keep seeing is that it boots to 2.5G even though I have it set on X13.5. X13 does the same thing. Is this related to the C1 error?

[Tony]

I can't unhide these, DFI have added so much of their own code it would need a direct hack to get this to show...thought I would share though

Is this something new on the way with Phenom ?

[jonspd]

interesting that would be really nice wouldn't it

[illogical]

Is this something new on the way with Phenom ?

Are you OCZ people planning something incredible with DDR2 ?
That DFI know about, 2048 !!! and I've just ordered OCZ 4GB PC2-8500 Reaper HPC (2x2GB), from Aria, knew I should have waited.

[AlabamaCajun]

I can't unhide these, DFI have added so much of their own code it would need a direct hack to get this to show...thought I would share though

Is this something new on the way with Phenom ?

Them are the DDR3 settings 800-1666 (non-Selectable ones) 2048 will be fore the OCZ PWN Reepers
From the Phenom specs on AMD I found out our controllers are ready to support DDR3 but the MOBOs don't. I doubt a B2 or B3 stepping will run much over 1300 though.

[KTE]

Officially it supports DDR3-1600 max already.

[AlabamaCajun]

Officially it supports DDR3-1600 max already.

You first

[KTE]

Hehe, physically the circuitry is not wired up though
I discussed with some MEM MFG workers when 9500 was released if I could mod the sockets to try DDR3, the answer was just no.

[Tony]

I have a feeling we may see 1333 from DDR2, i never expected for us to see higher

Just weird that this bios had the options and earlier builds did not...

I will show you all the voltage options they have hidden next...if only modbin worked LOL

[AlabamaCajun]

Based on DDR2-800.
DDR2 works from 100-200Mhz at the circuit level and dual multiplexed in the DIMM module meaning one signal access 2 chips. one on the even cycle, the other on the odd cycle. The chips still operate at 200mhz but the multiplexer handles it at 400.

Based on DDR3-1600.
DDR3 works from 100-200Mhz but is quadplex or 2x Duplex. This adds an extra cycle to the already double data rate which quad pumps the IMC-DRAM link. The chips still operate at 200 but there are banks four on the DIMMs sharing a single lane which is effective 800.

Sneaky way to get the speed but it works because it fattens the pipe where the bottleneck is, getting data loaded into the rows and columns of cells.

Bump the base line by just 66mhz and both go from 800 to 1064 and 1667 to 2124 respectively.
2*266 = 533 or DC 1066
4*266 = 1064 or DC 2128

[Oese]

new spi1M personal rekord over here

[sea_man]

I'm not, just sold my 5000BE+250 HDD+80 euros(beer savings)=Crosshair II

now DFI/9500 is 2.8 stable (Benching+Gaming+Surfing+Prime) WinXP x86/x64, Vista/Server2008 x64.



Never had one, but no more Duals for me

I`m glad if you share the complete config for the 9500@ 2.8 ( volt , and other settings)

and few question....
1-in the phenom is necesary the NB and Vcore volt are equal ?
2- what is the best method to cooling the chipset an PWmic ? ( i have 44º Pwmic an 49 chipset , in idle..... i fell is to Hot..)
3- is correct the temp showed in the smart-gardian ?

Thanks.

[Final8ty]

I'm not, just sold my 5000BE+250 HDD+80 euros(beer savings)=Crosshair II

now DFI/9500 is 2.8 stable (Benching+Gaming+Surfing+Prime) WinXP x86/x64, Vista/Server2008 x64.



Never had one, but no more Duals for me

My Bad! got you mixed up with Oese.

[aGeoM]

I`m glad if you share the complete config for the 9500@ 2.8 ( volt , and other settings)

and few question....
1-in the phenom is necesary the NB and Vcore volt are equal ?
2- what is the best method to cooling the chipset an PWmic ? ( i have 44º Pwmic an 49 chipset , in idle..... i fell is to Hot..)
3- is correct the temp showed in the smart-gardian ?

Thanks.

1- Not necessary, but recomended by DFI engeniers as Brother Esau posted.
2- I use CoolerMaster Gemini II, it cools MEM,CPU,NB Chip and PWM , yout temps are OK.
3- I don't use SG, but you can check against Everest or HWMonitor.

[aGeoM]

My Bad! got you mixed up with Oese.

No problem here. Be well...

[Oese]

uhmmm.. well... was planning to use a lab chiller for some more oc and/or replace my grafics cards with a pci card to get some more headroom on the 12v lines...

but "unfortunately" weather was too nice that weekend and after looking at that chiller i had in mind i'm not sure whether that thing is gonna cool that 200w away... will try both things another time..