GA-990FXA-UD5 issues (need quick answers)

[Night Gyr]

Well. I've read the AsRock documents and have a new idea.

It looks like the problem lies in the conflict between the Am3 CPU's and the Am3+ Boards. They say in the keynote that the power design has changed. But you are trying to use the old CPu's with the new power design. Who wins in this situation? Does the system behave according to what the CPU wants or what the VRM wants?

The other factor is that the UD3 behaves differently. It's a lower tier board and it's behavior is generally qualified by Gigabyte as slightly worse, so I think it's ok if the 5 and 7 differ from the UD3 in a certain way, that's how it should be.

[absic]

It looks like the problem lies in the conflict between the Am3 CPU's and the Am3+ Boards.

I feel you have probably hit the nail on the head with this Night Gyr. I believe the new 9 series boards and newer 8 series boards with the AM3+ Socket, have been designed primarily with a view to the new AMD Fusion FX AM3+ CPU's and not for the existing AM3 processors. Backwards compatibility is always going to present some issues and I wonder how people would feel if these boards were only designed for AM3+ Processors with no backwards compatibility, in the same way that Intel releases their new CPU designs.

I feel the real test for these motherboards will only come when the new CPU's are available and that it is only then, that we will see their true strengths and weaknesses. I am not dismissing the current issues or ignoring them but I am a realist and it is obvious to me that, changing the socket design and the power design is not a task undertaken lightly and it is bound to have an impact on the existing technology. The Vdrop issue is more noticeable on the current 6 core CPU's than on those with four or less cores and the reason for this must be in the basic design of these processors. If you lock off two of the cores in these processors then the Vdrop is not so dramatic.

At present, as individuals, we do not know any firm details of the new AM3+ processors or their full specifications. The motherboard manufacturers must have more details and AMD would have supplied them with technical guidelines as to how the motherboards are to work. I would assume, that is the reason for the current problems and it is not down to bad design or faulty components.

[Beekeeper]

The other factor is that the UD3 behaves differently. It's a lower tier board and it's behavior is generally qualified by Gigabyte as slightly worse, so I think it's ok if the 5 and 7 differ from the UD3 in a certain way, that's how it should be.

Agree. For sure looking at it's price UD3 must be kind of lower end mobo

@ absic
Yes it would be interesting to see the tests with am3+ cpus, but how long we have to wait for amd to release them?
I've heard they will have also many cores at least 6, is this true?

[absic]

@Beekeeper

Still no firm release date for the new AM3+ processors. They were originally scheduled for early June but there were some speed issues that AMD felt needed addressing first. Latest info I have is August/September but this still hasn't been confirmed.

I believe that initially there will be 4 and 8 core versions released but again, I have no confirmation that this is true.

[iSeries]

What I don't understand then is if its a new power design for AM3+ processors, why arent the owners of other brands of 990 boards complaining?

With that said though - my 1055t is OC'd at 3.75ghz @ 1.35v. At full load this drops to 1.31v but stays stable. On my previous board anything less than 1.35v was not stable so maybe there's more to all this than what meets the eye...

[gareth170]

well the asus crosshair V formula board is great,

so far no vdoop at stock under load.. i set the vcore at 1.30v in the bios, it ideas at 1.308v and under load it goes to 1.320v....  love it ... it has the LLC options which u can control the vdoop or let it do it automaticly..

if the GA-990FXA-UD5 had the LLC options then maybe it would of be ok

[toysoldier]

I feel you have probably hit the nail on the head with this Night Gyr. I believe the new 9 series boards and newer 8 series boards with the AM3+ Socket, have been designed primarily with a view to the new AMD Fusion FX AM3+ CPU's and not for the existing AM3 processors. Backwards compatibility is always going to present some issues and I wonder how people would feel if these boards were only designed for AM3+ Processors with no backwards compatibility, in the same way that Intel releases their new CPU designs.

I feel the real test for these motherboards will only come when the new CPU's are available and that it is only then, that we will see their true strengths and weaknesses. I am not dismissing the current issues or ignoring them but I am a realist and it is obvious to me that, changing the socket design and the power design is not a task undertaken lightly and it is bound to have an impact on the existing technology. The Vdrop issue is more noticeable on the current 6 core CPU's than on those with four or less cores and the reason for this must be in the basic design of these processors. If you lock off two of the cores in these processors then the Vdrop is not so dramatic.

At present, as individuals, we do not know any firm details of the new AM3+ processors or their full specifications. The motherboard manufacturers must have more details and AMD would have supplied them with technical guidelines as to how the motherboards are to work. I would assume, that is the reason for the current problems and it is not down to bad design or faulty components.

I don't understand why the BIOS doesn't adjust to the CPU being detected ? So when it detects an AM3 CPU it runs with no vdroop/LLC enabled.

I'm currently running a 1100T BE on a GA-990FXA UD5 board while I wait for BD to be released and it's very frustrating to deal with the high idle vcore on this board. Normally it would be a breeze to get it to run >4GHz but not on this board with the vdroop. This is  the first time ever I have opted for a Gigabyte board instead of my normal Asus choice ... I hope I'm not regretting it. I will keep my fingers crossed hoping there will be a fix to this issue.

[ack65]

i have heard nothing from here or gigabyte tech for a week, beyond the F5b bios flash which did nothing to help. the answer i received from tech last week was this...

"
"Dear Customer,

We followed the AMD AM3+ load line calibration guide line, therefore
Load will cause the voltage to drop is normal"

which sounds like they expect this and are done with the issue. makes me wonder why other boards, including the UD-7, are not having the v-droop problems.

i have my UD-5 RMA set up and am waiting for my replacement MSI board tomorrow.

[Nightrain]

The UD7 has the vdroop as well but it doesent bother me all. I can still get 4.0ghz stable on my 1090T.

[toysoldier]

The UD7 has the vdroop as well but it doesent bother me all. I can still get 4.0ghz stable on my 1090T.

I can live with the vdroop. What I can't live with is the overvolting. This board is feeding my 1100T (1.52v while idling at stock speed).
The VID for my CPU is 1.3750 not 1.52. The only time the board detects the correct VID is when I set the multiplier myself. Otherwise it want's to feed the CPU 1.4750v which is consistent with the fact that this boards overvolts by 0.1 in idle mode. Bios F5b is by far the best for this board up until now. But I'm still crossing my fingers for a fix for AM3 CPU's and the voltage issues.

[gareth170]

it sounds like gigabyte are just being lazy.

they should add the LLC option in the bios, to let the user choose how much LLC is used.

the lower the LLC the more vdoop there is,  higher the LLC the less vdoop there is,

[iSeries]

Another happy customer

http://www.overclock.net/amd-cpus/1066717-crazy-vdrops-please-help.html

[gareth170]

i still say they should add the LLC option in the bios, to let the user choose how much LLC is used.

it would stop these issues.

[absic]

Yes, I agree it would be nice to see some sort of LLC control on these boards and as they have it on the Intel platform I can't see why not.

[Dark Mantis]

No I would have to agree with you there absic as the Intel boards have had it for years. I sometimes wonder why so many issues with the motherboards come down to BIOS files in the end. Good hardware bad BIOS.