Gigabyte GA-MA790X-UD4 Thread

[devguy]

Thought I'd start a thread for all those who have/are interested in this board. Here: is the link at Newegg.

Pros: 790X chipset, which is the 790GX everyone is familiar with, but without the unnecessary (well... for most of us enthusiats!) onboard video and dual PCIe 2.0 slots (dual x8 when in crossfire). It has heatpipe northbridge/mosfet cooling, and has the "newer" SB750. Also, it has 4 right angle SATA ports, plus two upward facing ones. Also, this thing has USB ports GALORE; 8 on the backplate (no 790GX video connectors here), plus two more USB connectors on the motherboard! Lots of overclocking options!

Cons: Even though it has the ALC889A onboard chipset, this does NOT support DTS: Connect, which was a huge (and unpleasant) surprise to me. Southbridge not heatpipe connected with the northbridge/mosfets (not really a big deal given that the heatsink doesn't really get that hot, but I'm only using AHCI and not RAID5). Unfunctional ACC, but see below.

Attached is a testament of about 3 hours working with this board to try to find the sweet spot. I did only manage to get to 3.6Ghz stable with under 1.5V on a 720BE, which is kinda disappointing, but take a look at the Northbridge/HTT frequencies! I never got over 2.1Ghz stable on either one of those with the Phenom I X3, and this has a huge impact on memory intensive programs. I gave the northbridge VID an extra .2V and the northbridge/HTT clock (thing) and extra .1V, neither of which really heats up anything. And a big WOOT for my first processor to get under 20s in 1M superPi!

Temperatures are at a max of 47C LOAD under a Zalman 9700LED running at fairly slow speeds (I can easily hear my GPU fan over the Zalman). I did manage to get the PC to boot at 267x15= 4005mhz, but it was only Memtest86+ stable at 1.55V and I couldn't even boot windows. I believe I can get this chip to 3.75Ghz if I up the voltage to 1.53V, but that is a lot of voltage for only 125 more mhz, so I probably won't bother.

ACC seems completely broken with this board as none of the BIOS options seems to do anything (or unlock my fourth core). Also, check in my AMD Overdrive picture where the ACC control is greyed out, as if I didn't have an SB750. However, I just proposed this dilemma to Gigabyte...

Hello, I recently purchased your GA-MA790X-UD4 motherboard for my Phenom II x3. This may be a strange question, but is there something wrong with the Advanced Clock Calibration on this board? The reason I ask, is that no difference is made at all when enabling it in the BIOS, as in no difference in overclocking stability (which is to be expected with the Phenom II), but more importantly, my fourth core does not unlock on one of the batches that has a good fourth core (0904, stepping RB-C2). I've tried using the setting, "Auto", "All Cores" (adjusting the values), and "Per Core" (also adjusting the values). And it isn't just my chip, many others across the web have tested their Phenom X3 in other boards and the fourth core unlocks stably, but the same chip does not allow it with this board. Furthermore, using the two latest builds of AMD Overdrive, the ACC section is completely greyed out, as if I did not have an SB750...?

I understand that the ACC at least works as expected with your GA-MA790GP-DS4H board. I wonder if you could try and compare the BIOS code with my new motherboard and release a BIOS that allows for the same adjustments. Judging from AMD Overdrive having a greyed out section for ACC, I'm sure you'll have some Phenom I owners very upset, as ACC is supposed to be the savior for Phenom I overclocking.

And keep in mind that ASUS has silently released a beta BIOS for their AM2+ 790FX boards with SB750, allowing users to unlock the fourth core of an X3 720, so please do not say their is no motive for you to do this.

Thanks!

...and here is the response:

Dear Customer,

A new BIOS will be available soon. Thank your for your patience.

Thank you for choosing Gigabyte products

I can hardly wait!

Post if you have any experience with this board, or have any general inquiries/things you'd like me to try.

 

[suraswami]

Thanks for posting this.

Someone on egg commented that CPUZ and related tools when opened crashes or BSOD. Do u have any such issues?

 

[devguy]

Thanks for posting this.

Someone on egg commented that CPUZ and related tools when opened crashes or BSOD. Do u have any such issues?

Sounds like they have an RMA in their near future...

As you can see from my picture, I have three instances of CPU-z open, as well as overdrive, and Everest. And no BSOD.

Edit: Added a Vantage pic of my system on a one week old install of Server2008 x64 RTM with the 4870 @830/1000 using Catalyst 9.2. Not too shabby!

Edit2: Does anyone else think Jane Nash is kinda hot?

 

[devguy]

Gigabyte just released a new BIOS for the board here.

It says, "Update ACC code for Phenom X3 CPU" as I requested. I'll try it out and report back to you guys!

Edit: OMG, look at me, they call me AT!

Gigabyte rocks! I'll be trying some overclocking with my new quad later today!

 

[sniviler]

OH YEAH just updated with the F2A bios .... MY 720 BE X3 is now an X4 Phenom ll 20 , running fine @3.6Ghz 1.4v looped 3DMARK06 for an hour , played FARCRY2 with no BSOD's, GOOD WORK GIGABYTE !

 

[devguy]

OH YEAH just updated with the F2A bios .... MY 720 BE X3 is now an X4 Phenom ll 20 , running fine @3.6Ghz 1.4v looped 3DMARK06 for an hour , played FARCRY2 with no BSOD's, GOOD WORK GIGABYTE !

Did you set your settings to "Auto", "All Cores", or do it per core? And, with your fourth core unlocked, could you do some stability checking with Prime95 and OCCT? For some reason, with my fourth core enabled, the computer completely freezes as soon as the first Lucas Leimar iteration has finished on every core. That happened to me consistently no matter what ACC settings or clock frequency I tried. With OCCT, it crashed my computer as soon as it began stressing the chip.

However, I did a Vantage run at the stock 2.8Ghz and it plowed through no problem.

 

[sniviler]

OCCT freezes me up with any ACC setting, wish I knew which core is disabled, anyway to find out ? .. as far as games go no lockups so far , probably not stressing the disabled core enough while gaming.

Gonna mess around with ACC alittle more, i'll check back.

have you tried ganged , unganged setting ?

 

[aymeric106]

Ran the test with the new bios:

1) I lost temperature reading for all CPU core (-255 °C)

2) Four core with ACC on auto

3) No problem to boot XP

4) Test Prime95 with 4 threads -> FREEZE

5) Test Prime95 with 4 threads one core at a time (with CPU affinity) -> OK for all core

6) Test Prime95 with 4 threads enabling one, then two,then three, then four core (with CPU affinity) -> FREEZE as soon as the fourth core is enabled. What ever the 4th core is.

7) Lowering the CPU multiplier, Prime95 works fine with 4 core

Comment:
- I've 'only' a 430W Earthwatt PSU, and i'm wondering if it can cope with the 4 core at full charge ?
- I'll run more test tweaking the NorthBridge to see what happens

 

[devguy]

It's interesting, because I've read online that when you enable ACC when using Phenom IIs, often the result is unstable performance at 100% cpu usage across the four cores. However, I've tried disabling ACC after enabling it, and that fares no better (to get the Phenom x3 back, you have to clear CMOS, btw).

With Prime95, a different core would finish the first Lucas Leimar iteration last everytime, yet it never failed that as soon as that final core finished, my computer would freeze up, and never a second before that!

I've reported the issue to gigabyte, so maybe we'll see a BIOS F2B in the next few weeks. And aymeric, yes, enabling the fourth core kills the temperature sensors with OCCT/AMD Overdrive, but not with Gigabyte's own EasyTune6, which I've also reported to Gigabyte.

 

[sniviler]

I noticed after disabling ACC the bios still reads x3 20, turning the pc off and on fixes it.

no need to clear the bios here.

EVEREST reads the correct temps with four cores.

 

[SystemViper]

I love the UD4P, I also had the Extreme, but the UD4P is the same thinkg but with no frillz...

UPDATE: i just realized this isn't a X58 UD4P i7 thread! but Still GIG rocks






'Freq : 4662.81 MHz (222.04 * 21)









.

 

[devguy]

Lol. Gigabyte actually released a GA-MA790X-UD4P motherboard, very silently. It just cropped up on NewEgg about 2 days ago, and is exactly the same as this board, with 3 exceptions:

1) Has the GSATA additional controller for adding two more SATA ports
2) The original 6 from the SB750 are all right angle, vs 4 right angle and two flat (as on this board)
3) It has a full 8+2 power phase like the AM3 790X/790FX board (something I'm actually sore about missing)

I wish Gigabyte had announced this board, instead of silently introducing it, as I definitely would've preferred it (and oddly, it is cheaper too)!

 

[LittleLizard]

kinda sounds that gigabyte plays you a joke :shadowshu

 

[devguy]

kinda sounds that gigabyte plays you a joke :shadowshu

I don't think this is a joke...

And Gigabyte took down the updated BIOS from their .tw site, but it just got uploaded to the .us site. I received a reply from Gigabyte support and said they are going to work on what I described to them, and have an F2B bios relatively soon.

 

[lateralus1123]

Ran the test with the new bios:

1) I lost temperature reading for all CPU core (-255 °C)

2) Four core with ACC on auto

3) No problem to boot XP

4) Test Prime95 with 4 threads -> FREEZE

5) Test Prime95 with 4 threads one core at a time (with CPU affinity) -> OK for all core

6) Test Prime95 with 4 threads enabling one, then two,then three, then four core (with CPU affinity) -> FREEZE as soon as the fourth core is enabled. What ever the 4th core is.

7) Lowering the CPU multiplier, Prime95 works fine with 4 core

Comment:
- I've 'only' a 430W Earthwatt PSU, and i'm wondering if it can cope with the 4 core at full charge ?
- I'll run more test tweaking the NorthBridge to see what happens

Hi, I just joined after seeing your post because I have the exact same issues. It isn't a power supply issue. It appears to be an inability of the fourth core to work properly in conjunction with the others. I ran Prime with all different combinations of affinities:

Core 0 passes
core 1 passes
core 2 passes
core 3 passes

cores 0,1 pass
cores 0,2 pass
cores 0,3 pass
cores 1,2 pass
cores 1,3 pass
cores 2,3 pass

cores 0,1,2 pass
cores 0,1,3 instant hard lock
cores 0,2,3 instant hard lock
cores 1,2,3 instant hard lock

cores 0,1,2,3 instant hard lock

This happens no matter what frequencies and voltages are set to, I even underclocked quite a bit with no luck.

As you can see, the problem arises when core 3 (the 4th core) has to work with 2 or more other cores. It appears to be fine when working with any one other core or by itself.

The strange thing is, I have all 4 enabled and OC'd to 3.6 GHz and I can run 3dMark, SuperPi, WPrime, Crysis, Left4Dead, etc. no problem. Only Prime95 and OCCT cause a hard lock.

It makes me wonder if there is a defect with this BIOS (F2B) or the way it enables the core, or perhaps the defect is on the chip itself.

Edit: I just noticed you all have the AM2+ board. I have the AM3 version of this board which is the GA-MA790XT-UD4P. Still have the identical symptoms though.

 

[devguy]

lateralus, thanks for that. Be sure to list your findings to Gigabyte, here. They certainly will listen, and the more people they have reporting the same problem, the more likely they will fix it quicker.

 

[lateralus1123]

Thanks, I wrote them a very in-depth description. I kind of doubt they will do anything since unlocking this fourth core is in no way something they are obligated to support. If they can do something that would be great.

 

[devguy]

Gigabyte just uploaded a new bios. Version F3F. It says that this improves audio compatibility. However, it has an older BIOS date than the F2A, and it shows up below F2A in the listings? The AM3 boards also got a BIOS update that lists "Improved Lan Compatibility."

Anyone wanna give it a try and see if it has any fixes for the ACC issues we've been having (and/or see if audio sounds better)?

 

[aymeric106]

about F3F bios:

4th core unlocking is gone ... :-(
Correct temperature reading is back

Didn't notice any change about the audio (I had no issue before anyway)

I'll stick with this new bios since my 4th core seems to be unusable. As i said before, my computer freezes as soon as 4 cores are 100% in charge BUT ALSO from time to time, with no reason.

 

[devguy]

Thanks for the report, aymeric. And Lateralus, what was their response? Remember that to read their response, you have to go to that site where you asked the question (they do not email their responses to you).

I asked Gigabyte what the improved audio compatibility meant, but they didn't really give me an answer. They did assure me that it does not mean they added DTS:Connect support.

 

[lateralus1123]

Thanks for the report, aymeric. And Lateralus, what was their response? Remember that to read their response, you have to go to that site where you asked the question (they do not email their responses to you).

I asked Gigabyte what the improved audio compatibility meant, but they didn't really give me an answer. They did assure me that it does not mean they added DTS:Connect support.

They just got back to me today. It took them 5 days and their response was more questions. More specifically, questions to which the answers could be easily found in my original message had they taken the time to read it. It's very frustrating when you write in such detail to be certain to leave no remaining questions for the reader, yet they ask them anyway. I replied to their questions politely but if they don't have any ideas in another week I'm letting it go. I'll update after their next reply, if there is one.

 

[devguy]

Well, I never received a reasonable response from Gigabyte about what the F3F BIOS did, so I just went ahead and tried them. My report:

-I love the new addition in the MIT that tells you the exact frequency of your overclock for a certain part (the part in the BIOS that tells you the effective RAM speed is now done for all frequencies).

-I don't like their setting of the auto on the PCIe frequency to 108, so I have to manually input 100 for peace of mind

-ACC has been messed with. What I mean is that the fourth core no longer unlocks, but that ACC still works, and well I might add, but with a huge caveat. I'll explain my finding with ACC below, but the huge caveat is that once you mess with ACC in the BIOS using Per Core adjustments (I haven't tried Auto/All Cores), you cannot cold boot the machine. In otherwords, the machine works fine until you actually turn it off and then try to turn it back on (warm reboots work fine). And, ACC still isn't supported via AMD Overdrive.

My report on ACC:
I've read countless times in the Internet that ACC is "built in" to the K10.5 architecture and is no longer helpful for the Phenom IIs. What I really think happened is that someone previewed a Phenom II chip, tried basic ACC adjustments (probably Auto), and either lost stability, or had no difference. Then due to his/her preview, there has been a "knowledge" that Phenom II doesn't benefit from ACC. Now, it could be that either this person was incompetent and didn't know how to use ACC, or maybe it was his board/BIOS were not ready yet for the combo of the two. However, I can tell you that that assumption is quite wrong.

I started testing ACC with Phenom II due to the results posted here. It is surely an interesting read through and has all the dynamics/typical players including some guy who thinks he knows everything and tells the OP that his results have to be anomaly because what he read on the Internet contradicts the OPs findings. Then, after several pieces of significant evidence are shown, he finally realizes that maybe he's got something!

1). Anyways, my first step for using ACC is to determine the sickliest core. This is assuming you've already found your "max" overclock for your processor (using only multiplier overclocking and not reference clock overclocking), such that increasing the multiplier by .5 causes instability. This is done by opening up AMD overdrive in Windows and setting all cores to stock speed where you know they are safe, but with plenty of extra voltage to work with (I've been using 1.5V, which is actually 1.52V for my system). Then, using per core adjustments, set core0 to the highest multiplier value + the .5 you couldn't previously attain. If your computer instantly freezes, then you probably need more voltage/better cooling, and I don't think these next steps will help much. If not, then open up OCCT and run it for at least 10 minutes. Make note of if it crashes or not. If it did crash, reboot into windows. Now, try core1 and do the same thing. Continue until you run out of cores. If your results are such that only one cannot handle the overclock, then you are very likely to benefit from ACC. If two, less likely, but still a good chance. If three or more, you are unlikely to attain benefit from ACC (but it is still worth trying).

2). Now, boot into your BIOS and set ACC to "per core". Now, adjust the first core that you found the instability on to "+2" and all the rest to "0" (including the locked fourth core on Phenom II X3). Boot back into windows and try to bump up the multiplier on that core like before (still with the other cores at stock speed). If it goes at least ten minutes without a crash, congratulations, your results are very promising! If not, boot back into the BIOS and adjust the ACC number to "+4" and try again. Repeat until you run out of positive numbers. If you still have no difference, it looks unlikely that ACC will help you, but it is worth trying the negative adjustments, starting from "-2" and decreasing. If you run out completely, give up, go home.

3). If you make it here, cool! Now, if you've run out of cores that originally gave you instability, retry the stability test on each core again (because sometimes setting an ACC value helps one core gain stability and alienates a different one). If all are good, you are done and can try more overclocking! If either of those scenarios don't work for you, then you need to move to the next unstable core and repeat step 2 for that core.

My findings are that ACC really is only useful for multiplier overclocking, and has no reasonable effect on reference clock overclocking (necessary for 710, 810, 920), so sorry you guys. Also, this method is good for when you've tried standard overclocking and get to a point where no matter how much voltage you pump into the f(*#*(ng chip, it won't get stable at a faster speed. But otherwise, your problem lies elsewhere (perhaps it does need more voltage). And keep in mind that altering ACC for one core can hurt the stability of another core, and it is possible that a never ending duel can arise between two or more cores effectively eliminating any benefit from ACC. As to what ACC does? I'm not entirely sure, but it seems to be a setting that effects the clock multipliers to make each core more compatible with the others (perhaps adding a phase shift to the frequency wave to try and make the waves of all four cores identical). As to how it works? I have no clue. One extra thing is that choosing higher values for the ACC choices (well, the absolute value of them) doesn't seem to be likely to damage your chip say in the way that a higher voltage value would, but I believe that the higher the value necessary to achieve stability, the more incompatible that core was with the others.

Hope that helps some of you all. As always, I take no responsibility for your actions and these findings are just that, findings. They worked for me (minus that fact that the PC wouldn't boot once I turned it off) where I previously could get a max of 3.5Ghz at 1.48V and now I can get 3.7Ghz stable (maybe more) with 1.52V (maybe less volts is necessary, too).

 

[Josh81]

So you can only use 3 out of 4 cores with this motherboard? (Amd newb)

 

[devguy]

So you can only use 3 out of 4 cores with this motherboard? (Amd newb)

If you have a Phenom II X4 (810, 920, 940), then you can use all four cores fine. If you have a Phenom II X3 (710, 720), you can use all four cores (if the fourth isn't completely defunct) if you use BIOS F2A. But F2A is riddled with problems of not being able to 100% load all four cores (I hope Gigabyte are working on that).

 

[Ozpa]

I've read everything in this thread, great stuff.

I'll be assembling my 720 + MA790X-UD4 soon too. I'm not going to do any overlocking just yet while the stock fan is in use, maybe 3months+ into using that CPU I will get something more serious for overlocking. As with the BIOS and info have up until now I'm not going to unlock the 4th core either.

I just have a quick question. What BIOS should I flash my mobo with (no oc, 3 cores)? (F2A or F3F)
What about when I get some serious hsf for overlocking?