Gigabyte MA790FX DQ6

[nate39]

Go back to the previous page and look what happened to me.
I think you have the same problem.

Thanks, Aussie, I used my old flash drive and it worked fine. I guess my new Corsair falls under legacy usb but I haven't been back into bios to see how mine is set yet. Right now I am using F5.

Does anyone know if the settings under autoexpress have any value with a 5000BE? thanks

[Aussie FX]

Does anyone know if the settings under autoexpress have any value with a 5000BE? thanks

No, it doesn't make any noticable difference with Phenom either.
You can try it, you won't hurt anything.

[Daveburt714]

Just an Update on the DS5 F5 Bios..

First off I was wrong about the "Memory Controller Volt" option raising the voltage on the IMC/NB. It was an honest mistake , looks like what they did was raise the base NB v's, so when I changed it from AUTO to "-.00xxx" it lowered the V's from 1.138, which was still higher than what it gave me with the F4 bios default, confused yet?

To make a long story short, changing the IMC volt option from AUTO (Now 1.138v's) to the "-x.xxxx" numbers it shows actually does lower the voltage by that amount from 1.138.. Moral of the story, leave it at AUTO.

This bios has actually given me a very good boost in NB performance, I could barely get it stable at 1930Mhz with F4. Here's a SS of my new OC using the F5 version.



The higher NB helped overall performance quite a bit! I don't have a really strong 96BE, and this is the best stable OC I've ever gotten out of it, so I'm fairly happy. Now they just need to give us some PLUS voltage options for the NB/IMC...

nate39: I agree with Aussie. I played with the autoexpress feature too to no effect using a Phenom processor. But it didn't cause any problems either.

I'm running 2643(12x) with ht@220, NB@1760, HTT@1540 and I've just hit 60C with Prime's small FFT. I have Thermalright 128i and I wonder if it's ok for Phenom 9600BE.

Carath, I have 2 Phenoms and both of them run better with lower voltages (~1.25v's). I could squeeze a little more out of them by raising the CPU v's, but it ussually wasn't much and generated a lot of heat that lead to serious instability. I realize no 2 chips are the same, but at least with mine anything over 50c was shakey (see SS above).

If your not running high CPU v's, you may want to check your HSF and make sure it's seated well.

[Aussie FX]

Thanks Dave.

Why would they consider the option of lowering the NB volts
I gave up on my support guy, I told him the language barrier was too great, which it was. It wouldn't surprise me if that happened here.

I saw you were contemplating putting your Phenom under water, I got mine out the other day and it has dropped temps by 15deg.

[Daveburt714]

Why would they consider the option of lowering the NB volts
I gave up on my support guy, I told him the language barrier was too great, which it was. It wouldn't surprise me if that happened here.

Well Aussie, at least they raised the base NB v's to 1.138. It's a step in the right direction. I totally agree with you though, you should be able to take it up to 1.25 at the very least!

I saw you were contemplating putting your Phenom under water, I got mine out the other day and it has dropped temps by 15deg.

I was, but I got lazy... Hey I was on vacation!
Anyway, it's over now, back to work tommorow.
Just curious as to whether the lower temps helped your OC?

[Aussie FX]

Dave, Amd power monitor reports my cpu NB at 1.25v.

As for my overclock my pc is going downhill fast since I installed SP1. I used to be able to get a 2.5ghz stable overclock and now I am at 2.4ghz I upped the voltage to 1.3 to try and keep it there.

[Majmac]

Could it be that the chip is degrading, or are you sure it is the SP1?

[Aussie FX]

Could it be that the chip is degrading, or are you sure it is the SP1?

Yeah the chip could be degrading although I would be surprised because it has always run at 1.25v except for the occasional increase to try for the last bit of an oc. But it has never gone over 1.375 and it was only ever that high for about 5 minutes.
I'm frustrated with SP1 that's for sure.

[Carath]

Carath, I have 2 Phenoms and both of them run better with lower voltages (~1.25v's). I could squeeze a little more out of them by raising the CPU v's, but it ussually wasn't much and generated a lot of heat that lead to serious instability. I realize no 2 chips are the same, but at least with mine anything over 50c was shakey (see SS above).

If your not running high CPU v's, you may want to check your HSF and make sure it's seated well.

Thanks Dave, now I see the problem. I just can't make my cpu fan go faster then 1100 rpm! My vendor sold me xthermal BTF 120 Pro and it's obviously not a good choice... What do you recommend for my cpu that has a good balance between quiet and performance (and it's 120 mm of course)?

[Daveburt714]

Dave, Amd power monitor reports my cpu NB at 1.25v.

As for my overclock my pc is going downhill fast since I installed SP1. I used to be able to get a 2.5ghz stable overclock and now I am at 2.4ghz I upped the voltage to 1.3 to try and keep it there.

Sorry to hear about your OC problems bud... I finally ditched Vista 64 for the 32 bit version because I was having alot of OC problems myself. Vista32 just seems to run a whole lot smoother and gives a better OC.

I wasn't real happy with the SP1 install of Vista32 either, the fact that it had the TLB fix built in was a pain, but Sam's patch fixed that problem and it's pretty painless once you set it to run at loggon.

SP1 for Vista32 didn't have any effects on my OC though...

I flashed back to the F4 bios tonight just to check the base v's of the NB and you are right it was 1.25v's... And it still won't do 2100Mhz on the NB! No idea what they did in the F5 version, but I get better NB speeds with lower v's, so what ever they did.... It's better!

Carath: Heres a good site for HSF information:
http://www.frostytech.com/articlevie...id=2244&page=4
Normally 120mm fans don't run at really high speeds. 1100 RPM sounds about right for a 120 to me. Lot's of info at Frostytech though!

[KTE]

Aussie FX: what voltage do you need for 2.4G stable now and what did you need when you first managed it stable?

Just keep my two Phenom deaths and one of Achims in mind...

[Aussie FX]

Aussie FX: what voltage do you need for 2.4G stable now and what did you need when you first managed it stable?

Just keep my two Phenom deaths and one of Achims in mind...

KTE, I've been keeping your problems in mind with my overclock, that's why I rarely go over 1.25v unless I have been going for max overclock which I haven't done for ages.
Right now I'm running 2360mhz @ 1.264v (1.25v in bios) When I get a bsod I have been ratcheting down the frequency and keeping voltage stable. I have run the voltage up to 1.35 to try and keep stability but it doesn't want to play.

If the chip dies, then it dies there are plenty more where that came from but it's frustrating watching the thing degrading. I don't know the reason, that's what is frustrating. It started happening since I installed SP1 (probably just coincidence) but I have never given it a hard time voltage wise.
Now it's watercooled it's running smp F@H 100% on all cores at 35deg. On air it hit 52deg max.

To answer your first question, I could get 2550 stable with 1.25v, this chip never needed any more as the NB always crapped out on me at ~2050.
Since then I have just been lowering the frequency and keeping cpu volts at 1.25.

I didn't realise you lost yours, I thought they came back to you over time?
Now you can join Cooper in finding out the best way to get the lid off.

[KTE]

*At first, I could run high MHz (+2700), low volts, stable, and you'll know what I mean by stable is every test I can throw at it for a week incl. 24hrs P95 with high and low fanspeed. First dead one reached 2750 1.24v stable and the last one reached 2754 1.28v stable.

*At the start with both, one benched 2914 max small apps and 287x and one screened 3000 and benched fully around 2830 max in all short and long benches. That needed 1.4v, nothing higher would change this.

*After 3-5 weeks in use at my stable settings and only lower, both degraded. I needed higher volts: first needed 1.36v 2700 now and second needed 1.38v 2700 for stable. Max MHz was much lower, max bench was also lower.

*So I worked downwards in finding what was low volts stable, 1.28v was personal my limit. I reached running 2.64G 1.24v, 1.1-1.25v IMC, 2.1-2.2v DDR for the first one and 2.69G 1.28v, 1.1-1.25v IMC and 2.1-2.2v DDR for the second one as perfectly stable again.

*I dropped settings to 2.64G 1.25v for both and ran that.

*After 2-3 more weeks, both died. They died suddenly, one day, never to boot, under no load, just idling mostly. The problems started appearing early on after 6 week mark, as when you cleared CMOS, they would not start up again at default volts, but luckily. The last time when I cleared CMOS or switched them off totally, they failed to start again on 3 boards in 14 days. IMC died, not the cores, same error.

Some people often chant "CPU temps are not important!!", well they ignore the obvious that temps *are* important for this exact reason; for us to troubleshoot and understand CPU happenings. Here I have no idea what the real temps of the cores/nb was at any time because we had no way to find out, so I can't tell if that was a possible cause or how to avoid it and to what extent.

[Daveburt714]

KTE: Hope you know I totally respect your observations! And I agree with you on the heat factor, temps have ALWAYS been a serious consideration when overclocking. I'm not sure Phenom/MB's do a good job of reporting temps properly. Really when you think about it, you would expect that 4 cores on 1 die @ 65nm would produce alot of heat.

But man, I hope your not projecting a problem that all B2 Phenom oveclockers will face. I've had this 96BE for at least 8 weeks and so far I haven't experienced any degredation (Knock on Wood)!

Aussie: I wouldn't be so quick to ignore the Vista64 SP1 factor. The reason I say that, is that I have 2 9600's, one BE and one Std. They both acted exactly the same under Vista64, to the point that they could have been the same chips! When I switched to a 32 bit OS the OC and especially stability got much better...

If you have an extra drive and a 32 bit OS, it may be worth your while to pull your 64bit drive (so you don't lose your current settup), and try doing a fresh install on the new drive using a 32bit OS.

I could be wrong, but if you have the parts it would be worth a try..

[gOtVoltage]

Well... I may have a 9600BE going to the Glue factory.......It did 2600mhz at one time....Then 2590mhz then 2550mhz then 2500mhz then 2430mhz maybe the max and will still get a ramdom reboot here and there....

""It just wont stay stable and it now likes MORE VOLTS Hmmmmm"",...Funny that my temps from what they showed on STOCK HS and Water never went above 45c at full load...It works fine in Ganged and UnGanged too.


Im testing this chip untill Sunday and if it seems like its getting worse ill have to drive to AMD and drop it off ...To think i started under 64bit XP Pro then slowly got worse,Then read how crummy the doo on X64bit ,,I experianced it..So i grabbed Vista32bit and XP32bit. Currently using 32bit XP Sp3 and max is barely 2400mhz stable no matter what i set NB/HT or Volts(kepping them resonable or Stock)...Ive had two Random Reboots @ 2300mhz but i had the FSB too 270+... waiting for the chip to Honestly roll over and play D_-E-_@-D ..Before i send it on its way to RMA-.

Chips Age young,,Im only on 8th week of CPU Life too! 0747 MPMW stepping was pretty common...

Note:My X2 6000+ @ 3.5mhz priceless in the same mobo...I like Phenom but B2's are just plain painfull.

P.S. Does formica/particle board count as wood...

[Aussie FX]

I'm not sure what's going on here, I'm going to be away for a few days so I will just switch it off and leave for it a while.
I have flogged some chips in my day, I have a FX-53 right in front of me that I have given 2.1v and it's still as strong as the day I bought it. This Phenom has been treated like a baby, so I'm confident it's still ok.

Daveburt you are right and that's what I will do when I get back.

Also I will be getting a 9850 when they come so it will be an interesting test with this install.

KTE, it sounds like yours were delivered to the gods in the name of science

[KTE]

Put it this way. I received a 9600 BE replacememt 22nd March and put it on sale the day before that without even opening, testing or overclocking it. That's how much I feared it dying out of the blue and that's how much I didn't bother having the hassle again. It is too unpredictable and I don't know why, so I avoided it altogether. That's 2x 9600 BE in a row for me.

The difference between what I and Achim did with those dead chips from what you guys are doing is: heavy benching, heavy stress testing, higher voltage, lower voltage, higher [relative] temperatures.

So, assume nothing, plan wisely.

[pablo NZ]

^Absolutely. It's the same with any component really. I chewed through 3 straight sets of Microns by leaving them to stress overnight. There's no way in hell that I would leave anything stressing for 24 hours. Very few people need that kind of stability.

[BrowncoatGR]

**edit** Nevermind i figured it out.

[Daveburt714]

OK... I finally figured out what is actually going on with the IMC/NB voltage in the F5 Bios for the DS5 Mobo. I know it's still not released for the DQ6 yet, but if they implement it in the same way at least you'll have a heads up!



Setting this option to "Auto" results in the IMC being set at 1.138v's. From there on setting it to the negative numbers it shows are correct, but they are subtracting the numbers from a 1.25v IMC! How wierd!!

Here's a partial table, you can calculate the rest yourself, the pattern stays the same:

Setting / Actual (per AMD Power Monitor)

AUTO = 1.138v
-0.0125 = 1.238 (1.25 - .0125, rounding in effect)
-0.0250 = 1.225
-0.0375 = 1.213
-0.0500 = 1.200
-0.0625 = 1.188 Etc....

Geez, no wonder I was confused!
If you want the highest possible IMC v's with this Bios, don't use Auto, just set it to 1 click down (-0.0125) and you will have the highest NB/IMC v's available.

[Lightman]

Thanks Dave for heads-up!!
I can't wait for this BIOS!

[Aussie FX]

Just dropped my 9850 in.
I thought I had reset the htt in bios to 200. Anyway she booted up straight away at 2.7ghz. So I'm currently at 12.5 divider at 215htt with 1.3v.
All cores at 100% gives me 50deg load temps.
It's a week 11 build.

More to come....

[Aussie FX]

Does anyone know what's up with AOD?
I was using it perfectly with my 9850 then I restarted the PC and it no longer detects my cpu.
Any ideas?
I have reinstalled it and I'm using v 2.0.0.17

[Aussie FX]

I had a feeling this might happen as it's only logical, but the bloody bios workaround for the TLB that MICROSOFT PUT IN SP1 IS STILL THERE FOR B3 chips.

So guys if your using xx50 Phenoms you will need AOD to get around it or Sam2008's fix.

Bloody Microsoft.

[Schneider]

Started playing around with my 9850 + 790FX DS5. My board or chip seems to disagree with changing the NB volts, i get the feeling my max NB vid is 1.138v as even AOD can't change it.

Are you getting decent NB clocks aussie? I'm topping out around 2.3Ghz which is hardly better than my old 9500.