Gigabyte MA790FX DQ6

[Daveburt714]

http://valid.x86-secret.com/show_oc.php?id=354855

That's for you Dave

Your turn now.

XP pro x32 was the answer.

Downside, it took 1.375v.
I have to install AOD now and tweak it, as I'm not sure of stability. I had to get this screenie done first.
Priorities you know.

WooHoo!! Gratz on the 3Ghz mark Aussie
Hopefully you'll be able to get it stable in the ~2.9 range.
BTW, glad to see you have your priorities straight.

Well MSI accepted my RMA, got the UPS tracking number tonight so I should have the board by next weekend at the latest. I'm really curious to see the results with the K9A2.

[Jaivan]

Keithlm:

I have almost the same ram as you the 8000 version which also has the strange timings as well all I can say is that the "Gigabutt" DS5 just doesn't like them.

I remember using the F1 and F2 BIOSes with these and they wouldn't even post consistantly with a phenom 9500.

The later BIOSes seem to have fixed that with the 9500, but I still cannot get my 9850 to run stable with them @ DDR 800. Too bad I don't have another set or ram to test this out.

The same set of ram in my DFI 790FX M2R run perfectly stable @ DDR 1066 with both the 9500 and 9850.

I think the main difference is that the "Gigabutt" BIOS is really buggy compared to the DFI and that they don't really seem to care about sorting it out.

Oh BTW which OS are you using?

[keithlm]

Keithlm:

I have almost the same ram as you the 8000 version which also has the strange timings as well all I can say is that the "Gigabutt" DS5 just doesn't like them.

I remember using the F1 and F2 BIOSes with these and they wouldn't even post consistantly with a phenom 9500.

The later BIOSes seem to have fixed that with the 9500, but I still cannot get my 9850 to run stable with them @ DDR 800. Too bad I don't have another set or ram to test this out.

The same set of ram in my DFI 790FX M2R run perfectly stable @ DDR 1066 with both the 9500 and 9850.

I think the main difference is that the "Gigabutt" BIOS is really buggy compared to the DFI and that they don't really seem to care about sorting it out.

Oh BTW which OS are you using?

Thank you.

I am a bit "into my cups" right at the moment.

But I thank you for at least confirming my less than scientific observation.

This ram SUCKS in the DS5. It worked VERY well in the DFI board I had for "oh such a short period" of time. Before either the BIOS or the Phenom blew up. (I'd bet it was the BIOS.) I even overclocked with no problems up to about 3.0Ghz. But MEH... But it did seem to be MUCH more stable then what I have now.

I hate wasting money. I can hope that Gigabyte can figure it out. (They do seem to be large enough to have some resources. Hopefully enough.)


But I suspect when I get back from my 2 week business trip that I will have probably ordered and received a new DFI board. When I'm in a hotel... away from home... I get VERY impatient. Enough that I don't care about wasting a few hundred dollars.


I would like to wait for the next iteration... but I have no patience. So I will get a "new" DFI 790FX-M2R. Just because Gigabyte is not competent enough to keep me happy.

Of course I could be wrong. If they release a new BIOS during those 2 weeks... I will come home and try it.

Who knows.

Personally I can tell you that AT THE MOMENT I'd rather pay a few hundred dollars and NOT have to deal with this Gigabyte board. (Rather than have problems for... HOW LONG... until they figure it out???0

It seems, at least on these forums, that DFI is more responsive to our needs.

(Although I have no personal bias against Gigabyte. I would LOVE it if they figured it out sooner... but I don't have the patience. And I think that will cost me a few dollars. ON THE OTHER HAND... this DS5 board with a X2-4800+ will be a good database server.)

WHO KNOWS. It's only money. BUT THE DFI BOARD "FELT" stable. This board feels "mushy".

GIGABYTE... FIGURE IT OUT.

OH... I'm drunk. I need to go pass out.

(BTW: I might end up getting a "normal" set of ram to tryo out also. Yet more money. Oh well. I have a need to show off a bit to the INTEL "Mafia". I'm tired of hearing them.)


Actually... being drunk: I'll tell my story. I had a socket 939 with a 3500+. Then I decided to get a dual core X2. I had somebody tell me I was an idiot for not throwing away my motherboard and my 4x1Gb of 2CAS ram ($400.00 at the time) to get a new motherboard, ram and CPU for a 5% increase in CPU speed. That was the first impression of the "Intel" fanboy club. I immediately realized that they were full of garbage. Beside... for YEARS the chips from Intel mult-tasked like garbage. Like I was going to belive that they got smart? And all the recent changes they have made have proven my decision to be correct: They have still used "questionable" methods to gain speed and win benchmarks. I'd rather have a chip that uses "hammer and tongs" methods work rather than "tricks".

But I digress. (And I'm still drunk. Oh well. At least I'm not making a 3 page post on Tom's. I have been known to do that when I'm drunk.) But then those long posts usually die out fast. Perhaps because I'm an idiot. Or perhaps because I post things that they don't want to attempt to debate. I'd bet more on the idiot viewpoint. But on the other hand... I've forgotten more about computers than many of them are old enough to have learned...

[Aussie FX]

Keith, that was a nice post. I wish I could type like that when I'm pissed hell I can't even do it sober.
Anyway the pertinant point of all that is that, you've forgotten more than they're old enough to have learned. I love it.
It's not so tough being okd is it? Although turning 40 was a haze...

But I digress.........

[keithlm]

Keith, that was a nice post. I wish I could type like that when I'm pissed hell I can't even do it sober.
Anyway the pertinant point of all that is that, you've forgotten more than they're old enough to have learned. I love it.
It's not so tough being okd is it? Although turning 40 was a haze...

But I digress.........

Aye.. sorry. I get that way sometimes when I'm drunk.

BUT: back to this board.

I hate that if I use "optimized defaults" it boots but is not totally stable. It is good that if I lower the NB and the Memory bus to 1800 that it is much more stable.

WHAT really makes my head crazy is this: Theoretically I should be able to go in and set the voltages and/or speeds of either the NB or the memory and have it be the same that the AUTO settings use.

But in reality I can NOT. If I change some settings such as memory timings... the system is completely unstable.

I've also tried everything at the default except for the NB voltage... I was trying to manually set it to the same value that it automatically gets set to.

BUT I can not do it. My guess is that when you use AUTO on some of the settings that other things that you can not see get set. But if you manually set those values... then the unseen or "hidden" settings do not get set.

(Just a guess. I could be wrong.)

[Daveburt714]

Today I got a second response from Gigabyte. The first response was "We know of no problems." When I told them to just look at any hardware forum they did reply; they just asked me how to recreate the problem.

I've been thinking about this every since I read it Keith....
I think the simple answer would be.

1) Install a 9850 on the DS5 using the F5 Bios
2) Boot into Windows
3) Do a 4x run of Prime 95 Blend
4) If it doesn't fail on it's own within 10 minutes, Open a Web page...

I tried corresponding with them too but after about 6 rounds of them asking me questions that I had already explained in previous messages I gave up..

I can understand a language barrier, but to be honest it sounds like they don't give a ...

This was my 1st GBT board, I bought a 2nd the 780G S2H for my HTPC. I even had to lower the HT Link to 1400 to get it to run stable which is not a big deal because of it's application. Needless to say though, I'll never buy another Gigabutt product!!

[Fulgore0rb]

Seeing as how this has turned into a DS5 thread. heh..

What timings/settings/voltages (detailed) did you GSkill 2x2gig guys use to get 1066MHz? I can boot but I get lots of Memtest86+ errors.

This is my first Gigabyte mobo too- been having plenty of similiar problems as you guys as well. Unless they fix this BIOS soon this will definately be my last Gigabyte mobo- if not the last already. Sooo tempted to pick up a DFI board- and further deseminate the facts about Gigabyte's bad Phenom support.

[keithlm]

Seeing as how this has turned into a DS5 thread. heh..

What timings/settings/voltages (detailed) did you GSkill 2x2gig guys use to get 1066MHz? I can boot but I get lots of Memtest86+ errors.

I just use the auto options for the timings/settings/voltages.

If I try to set anything by hand... the system ends up being less stable.

I ran memtest86+ with no errors for about 8 or passes. (everything at bios "optimized default".)

I can also run Prime95 - small FFS with no errors using all the sytem's defaults. But I have to slow the NB down to 1800 to run Prime95 blend. Other than the NB/HT settings in the BIOS... I don't change anything else.

[Schneider]

To the guys with instability problems, you have disabled CPU Xpress tweak right?

[Fulgore0rb]

To the guys with instability problems, you have disabled CPU Xpress tweak right?

No... I take it that you mean to say that its a Gigabyte "feature" for "unstableness". Should I be disabling all of the Xpress tweaks?

[Fulgore0rb]

I just use the auto options for the timings/settings/voltages.

If I try to set anything by hand... the system ends up being less stable.

I ran memtest86+ with no errors for about 8 or passes. (everything at bios "optimized default".)

I can also run Prime95 - small FFS with no errors using all the sytem's defaults. But I have to slow the NB down to 1800 to run Prime95 blend. Other than the NB/HT settings in the BIOS... I don't change anything else.

What is your "Row Cycle Time" (tRC) running at? I have a 9600BE so I wonder if yours will be running different auto timings. The memory claims it needs 48 but the BIOS won't allow anything above 42 (and does 33 on auto!?!) And do you have them ganged?

[Aussie FX]

I can also run Prime95 - small FFS with no errors using all the sytem's defaults. But I have to slow the NB down to 1800 to run Prime95 blend. Other than the NB/HT settings in the BIOS... I don't change anything else.

That points directly to ram. I notice most of the complaints are with GSkill.
I'm using Team (micron) others using Crucial (micron) don't appear to have this problem, so maybe that's the deal. It still doesn't excuse GBT one bit.
It's probably been mentioned so excuse me if it has, but have you ramped up the drivestrengths?


Don't apologise for that post ^^ I loved it.

I wish we could get DFI boards here, I'm seriously thinking of quad 3870's. I already have 2 but they won't fit over the Sata connectors on the DQ6. It was even a hack job to get 2 cards in.
It will be long time until someone imports them (DFI) again after the 590 debacle.

[keithlm]

What is your "Row Cycle Time" (tRC) running at? I have a 9600BE so I wonder if yours will be running different auto timings. The memory claims it needs 48 but the BIOS won't allow anything above 42 (and does 33 on auto!?!) And do you have them ganged?

Anyway here are my settings: (Using "AUTO" in the bios.)


NOTE: What shows in the "AUTO" column while you are actually in the bios are not necessarily the values used when the memory itself is set to "AUTO".

This memory is in dual mode un-ganged.

Currently at this time everything other than the NB and HT all settings are at what the bios puts them at when you select "OPTIMIZED". (The NB/HT are at 1.8Ghz.)

I have run it with the Extreme "Tweak" stuff enabled and disable with no real change in stability. I have also tried with HPET enable and disabled. I have used both ganged versus un-ganged. 1 stick of ram instead of 2. I have also tried setting the row refresh cycle to 195ns. None of those things helped.


It could be the memory. But even if I change the speed to DDR2-400 it still has the same instability. Or I should say it MIGHT have the same instability. It takes MUCH longer to have problems. (And if I put in old DDR2-400 sticks... they have the same problem.)

It could be the CPU... and I would be much more tempted to RMA this if others did not have the same problems.

Perhaps I have two different stability problems. One that shows up at 2.0Gb NB immediately if the ram is at DDR2-1066... and another problem that shows up later regardless of the NB.


AUSSIE: When you say "ramp up drivestrenth" which ones? To what value? There appears to be 6 of them.

(I have attempted to play with them a little... but either I didn't notice too much or it immediately rebooted. But I'll try some more.)

[keithlm]

ODT?

The DFI boards seem to show ODT (Cpu On-Die Termination) at 300 ohms.

This board appears to use 75 ohms?

This might be very important in isolating IMC problems?

I don't know.... I'll keep researching. (And I have to go on a lunch date. Sigh... women. Like they don't have their priorities right. Stability or lunch....)

[G0ldBr1ck]

been reading this thread awhile so figured Id put in my results with my 9850be. Overall im pretty happy with it it out benches my 6400+be and i have got it about 90% stable at 3.0G and 100% stable at 2.9. unfortunatly i have the Gskill 2x2gig ram and have run into the same problem as others with the DQ6 in that it does not acomadate the timings needed for the ram so im stuck runing at 800Mhz. With a little research i found that the gskill ram was designed to run on intell chipsets only. Anyways here are some screenshots of my overclocks. Oh and by the way i can run 3DMark at 3.1G but i get a crash everytime i try to take a screenshot. I also made a 3DMark run with the ram unganged at 1066 and scored 17,012 but again to unstable to take a Screenshot afterwords. This is on vista 64bit so I dont realy buy into the vista being bad for overclocking, I had my 6400+ at 3.65 as full time use for 3 months.
Im not afraid of high voltages so long as proper cooling can be maintained and on this setup its all air, Zalman 9700 on the 9850 at 3.0Ghz @1.5vots never went above 57 deg. under full stresstest.

EDIT: just relised that this screenshot put me in the top 5 3DMark06 AMD score thread. http://www.xtremesystems.org/forums/...d.php?t=185979

[Aussie FX]

Good clocks Goldbrick.
I'm having the same problem at 3ghz at the moment, I can do most things but when I take a screenie - crash. 2.9 is completely stable though.

I'm getting much better clocks on XP. I couldn't even get close to 3ghz on Vista x64. Have you tried XP?

This ram thing is turning into a pattern. I wonder if the DFI guys are having the same trouble. -Campbell??

[Aussie FX]

Keithlm.

I'm not sure about all those drivestrength settings either, so I would just play and see what gives. Don't push them too hard though,because it's a bit like overvolting. (laymans term)

It pushes the data through the ram harder/faster and can help with stability.

What resolution are you using? I had to scroll a long way to see the end of your post and I've got a 24".

Sigh... women. Like they don't have their priorities right. Stability or lunch....)

Depends on how much you've had to drink.....

[G0ldBr1ck]

The GSkill ram is clearly made for intel chipsets only as you can see on there product page here http://newgskill.web-bi.net/bbs/view...=g_ddr2&no=115. With intel's mem controllers on the chipset vs AMD's on the cpu die it makes sence that the timings for one or the other would differ.

I just grabbed some 1066 dominator ram and gonna give it a try. Hopfully i can get good results.
Like i posted above my 3dmark score went from 15,704 to 17,012 just by going from 800Mhz to 1066Mhz.

Ill post my results with this and my 9850@3.0 as soon as i get time to work at it.

I have also sence my last post got my 3.0G overclock completely 100% stable @ 1.558 volts and still only would hit 57deg. max on the core's and the cpu as the whole unit temp only reached 48 deg.

Edit. To answer you question Aussie FX, I have not tryed the phenom on xp however with my 6400+be i got the same Stable OC of 3.65Ghz on xp as I have on Vista64.

[Daveburt714]

To the guys with instability problems, you have disabled CPU Xpress tweak right?

Yeah Schneider, I've tried it on&off, same with the EPP mem timings and manual adjustments...

I'm on my 3rd Phenom and 3rd Mobo and they all actually act pretty much the same reguardless of settings. I see these guys with 3.x Ghz benchies and can't help but think they are all just Suicide Shots, benchable but not stable...

I've been OC'ing AMD's since the K6/400's and I've never seen a chip as unpredictable as the Phenoms, it will run stable for hours, then the next time you boot it crashes for no apparent reason...

That being said, I finally found a stable setting with my gear (see Sig) 245x11x8x6 - Vcpu 1.275, Vimc 1.15, Vmem 2.0. Yep, 2.7Ghz, with an IMC slightly under 2Ghz and a seriously degraded HTT (1470).... But it has proven stable for the last few days and actually it doesn't perform all that bad. But not nearly what I expected...

G0ldBr1ck:I had the same thoughts, blameing it on the G.Skills, so I bought some OCZ Reapers (1066/Micron D9's). I wish you luck, but in reality I don't think you'll see any benefit.

Take it for what it's worth, but running your CPU or IMC over 1.35v's will just cause your chip to degrade very quickly (talking weeks here!)....

Wish you luck, and hope you prove me wrong..

[KTE]

Dave, yeah it exists with Phenom not the board, it'll happen on any board.
Just figured mine out with extensive testing by individual core clocking -- its when one core is weak which causes the freezing
No more freezing for me

Still stock volts.

[Schneider]

Well hopefully the SB750 will help with the htt clocking as I think thats the biggest problem when it comes to boards booting.

I can confirm my 3.2+ ss are just bench stable, but I'm not fussed with lower clocks, I can't tell the difference between 2.75Ghz Phenom, Phenom 3+ and my e8400 @ 4Ghz except in benchies.
Phenom 2.75-3.25 is bout 1500 3D Marks in 06 and finally starts nippin at the e8400's heels

[Aussie FX]

I can confirm my 3.2+ ss are just bench stable, but I'm not fussed with lower clocks, I can't tell the difference between 2.75Ghz Phenom, Phenom 3+ and my e8400 @ 4Ghz except in benchies.

Exactly, the more I compare my Phenom with a q6600 in particular and C2D's in general the more I think intel have made them to win benches and that's all.

In real world usage my Phenom is certainly smoother. Even my 9500 @ 2.4ghz. They just don't hesitate.

I'm on my 3rd Phenom and 3rd Mobo and they all actually act pretty much the same reguardless of settings. I see these guys with 3.x Ghz benchies and can't help but think they are all just Suicide Shots, benchable but not stable...

Would I do that?
2.9ghz is completely stable though.

[G0ldBr1ck]

Daveburt714: I have read plenty of reports of the chip degrading and im willing to test this out. I have been overvolting mine for almost 2 weeks now and have not yet had a problem with degrading. Also your theory on the suicide screenshots may be correct in allot of cases but i have been running everything i have including allot of gaming (Crysis) and benchmarks for a couple days now at 3.0Ghz@1.56v and have only had a couple crashes and they were from the display adapter drivers crashing (ATI) wich i have happen when im not CPU overclocked most likely do to the high overclocking I have on my 3 cards. Maybe i just have a very lucky CPU, maybe i have a lucky DQ6 board or chipset who knows.

What sucks is i can only get stable with multi. cant get a decent OC at all with HT bus.

Im just happy realy that im finely getting near the 3dmark scores that Intel chips have been easily hitting. hopefully with the die shrink they will get the clocks up and maybe just maybe the phenom will finely bench like an Intel.

Edit: I have installed my 6400+be in my wifes DS5 board and cant get as high of a stable OC as I had on the DQ6 for 3 months. 3.425 seems to be max on the DS5, had 3.65 on DQ6. Im a little confused with this?

Another issue i could use some advise on is how to keep the bios from reverting back to F2 anytime there is a checksum error and the backup bios wrights to the primary? This is a massive pain in the arss! have to put in a old dual core to post then update back to F4.

[keithlm]

Another issue i could use some advise on is how to keep the bios from reverting back to F2 anytime there is a checksum error and the backup bios wrights to the primary? This is a massive pain in the arss! have to put in a old dual core to post then update back to F4.

Read the notes for the BIOS. For BIOS "F3" the last change was:
"Added bootblock checksum support (The back up bios will be updated)"

SO if your board had F2 when you got it... and you immediately put F4 onto it... anytime there is a flashback it will go back to F2.

But if you flash F3... and THEN you flash F4... if there is a need to revert... hopefully it will go back to F3. (I'm assuming when they say "back up bios" they mean the "Dual BIOS". But I have never seen the BIOS revert... so I can't say either way.)



EDIT: Just got another answer from Gigabyte. Apparently since Prime95 wasn't written for Vista... they don't seem to consider instability of that application a problem; they want to know what else has problems. Okay... how about the Vista OS itself. It reboots itself with nothing else running while using the Gigabyte DS5 motherboard at default settings. I'd say that was a "slightly bad" thing.

I am SO getting fed up with their responses.

[G0ldBr1ck]

Read the notes for the BIOS. For BIOS "F3" the last change was:
"Added bootblock checksum support (The back up bios will be updated)"

Ah, thanks i shall try that. On my DQ6 just about everytime i try to set manual timings on the ram it creates a checksum error and reverts to backup bios.