[Resolved] [Kind Of] SLi Mode Corrupt / Distorted Screen

[BladeRunner]

I've been trying to assist you in the nvnews topic, but you need to accept it's not easy when you don't have the issue yourself.

I have that 600 watt Enermax in my SLI, and so far it's worked 100% perfectly with none of the graphical issues.

PSU's can cause issues, but I don't think this is your problem. It's got to be more a timing, interference, incompatibility issue as it's only happening in full screen. The fact you are getting it to work in windowed mode points to the fact SLI does work ok, just not full screen.

Not sure what else to suggest. Have you stressed each card individually with the other card removed from the PC. It might not be related but it should be done to ensure they are both completely 100% artifact free on their own.

Have you checked both cards bios version & rev version are exactly the same, again they should be but confirm it.

have you tried a different monitor cable, unlikely but try it if it's easy to do.

Next try reinstalling the o/s on another, (ide if that's all you have), hdd to see if it's a install corruption problem.
unless is a hardware fault it must be solvable, and if it were I would doubt it would work in windowed mode ok. The only slight possibility is fullscreen is taxing the NB chip and PCI-E bus more, but that still wouldn't explain why you get perfect windowed mode.

Another thing is all of you with this exact problem list you total hardware & driver bios info to see if there is a common denominator at all.

I haven't checked everything you've tried so far, therefore ignore anything I've mentioned that you've already tried.

 

[Last Starfighter]

Originally posted by: BladeRunner
I've been trying to assist you in the nvnews topic, but you need to accept it's not easy when you don't have the issue yourself.

I have that 600 watt Enermax in my SLI, and so far it's worked 100% perfectly with none of the graphical issues.

PSU's can cause issues, but I don't think this is your problem. It's got to be more a timing, interference, incompatibility issue as it's only happening in full screen. The fact you are getting it to work in windowed mode points to the fact SLI does work ok, just not full screen.

Not sure what else to suggest. Have you stressed each card individually with the other card removed from the PC. It might not be related but it should be done to ensure they are both completely 100% artifact free on their own.

Have you checked both cards bios version & rev version are exactly the same, again they should be but confirm it.

have you tried a different monitor cable, unlikely but try it if it's easy to do.

Next try reinstalling the o/s on another, (ide if that's all you have), hdd to see if it's a install corruption problem.
unless is a hardware fault it must be solvable, and if it were I would doubt it would work in windowed mode ok. The only slight possibility is fullscreen is taxing the NB chip and PCI-E bus more, but that still wouldn't explain why you get perfect windowed mode.

Another thing is all of you with this exact problem list you total hardware & driver bios info to see if there is a common denominator at all.

I haven't checked everything you've tried so far, therefore ignore anything I've mentioned that you've already tried.

Hey, thx for the help it is appreciated!

Not sure what else to suggest. Have you stressed each card individually with the other card removed from the PC. It might not be related but it should be done to ensure they are both completely 100% artifact free on their own.

Yeah, I've run both cards on their own under some fairly heavy load for a sustained period, I've been unable to get them to falter.

Have you checked both cards bios version & rev version are exactly the same, again they should be but confirm it.

Card 1:

$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA specific display adapter information
$ffffffffff ---------------------------------------------------
$0100000000 Graphics core : NV40 revision A1 (12x1,5vp)
$0100000001 Hardwired ID : 0044 (ROM strapped to 0044)
$0100000002 Memory bus : 256-bit
$0100000003 Memory type : DDR (RAM configuration 07)
$0100000004 Memory amount : 262144KB
$0100000005 Core clock : 351.000MHz
$0100000006 Memory clock : 448.875MHz (897.750MHz effective)
$0100000007 Reference clock : 27.000MHz
$010000000b HW masked units : pixel 0100b, vertex 001000b
$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA VGA BIOS information
$ffffffffff ---------------------------------------------------
$1100000000 Title : nv45 p211 sku10 VGA BIOS
$1100000002 Version : 5.40.02.17.15
$1100000100 BIT version : 1.00
$1100000200 Core clock : 200MHz
$1100000201 Memory clock : 600MHz
$1100010000 Perf. level 0 : 350MHz/450MHz/1.30V
$1100020000 VID bitmask : 00000011b
$1100020100 Voltage level 0 : 1.10V, VID 00000000b
$1100020101 Voltage level 1 : 1.30V, VID 00000010b
$1100020102 Voltage level 2 : 1.40V, VID 00000001b
$1100000003 SW masked units : none

Card 2

$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA specific display adapter information
$ffffffffff ---------------------------------------------------
$0100000000 Graphics core : NV40 revision A1 (12x1,5vp)
$0100000001 Hardwired ID : 0044 (ROM strapped to 0044)
$0100000002 Memory bus : 256-bit
$0100000003 Memory type : DDR (RAM configuration 07)
$0100000004 Memory amount : 262144KB
$0100000005 Core clock : 351.000MHz
$0100000006 Memory clock : 448.875MHz (897.750MHz effective)
$0100000007 Reference clock : 27.000MHz
$010000000b HW masked units : pixel 0100b, vertex 001000b
$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA VGA BIOS information
$ffffffffff ---------------------------------------------------
$1100000000 Title : nv45 p211 sku10 VGA BIOS
$1100000002 Version : 5.40.02.17.15
$1100000100 BIT version : 1.00
$1100000200 Core clock : 200MHz
$1100000201 Memory clock : 600MHz
$1100010000 Perf. level 0 : 350MHz/450MHz/1.30V
$1100020000 VID bitmask : 00000011b
$1100020100 Voltage level 0 : 1.10V, VID 00000000b
$1100020101 Voltage level 1 : 1.30V, VID 00000010b
$1100020102 Voltage level 2 : 1.40V, VID 00000001b
$1100000003 SW masked units : none

have you tried a different monitor cable, unlikely but try it if it's easy to do.

I've tried a different monitor (with a different cable) - a flat panel that only goes up to 1280*1024 - It's shows the same artefacts / distortion at the equivalent resolution (albeit relatively minor) - fact is at TFT resolutions I cant push the system hard enough to get the major distortion to appear.

Next try reinstalling the o/s on another, (ide if that's all you have), hdd to see if it's a install corruption problem.

I reluctantly tried this a while ago (I hate OS reinstalls, mental note to make a ghost image of a vanilla setup). - Absolutely no difference.

The only slight possibility is fullscreen is taxing the NB chip and PCI-E bus more, but that still wouldn't explain why you get perfect windowed mode

It just *feels* like there is something on the boards that cant handle the combined hi-res data of both cards at high frame rates. Is there a chips that converts the onboard processed data into a signal monitors can understand? I also wonder if it's got anything to do with me having to use a DVI -> VGA convertor.

Another thing is all of you with this exact problem list you total hardware & driver bios info to see if there is a common denominator at all.

Athlon FX 55
Asus A8N-SLi Deluxe (Beta 1003.04)
1 Gb (2 x 512 Mb) Corsair TwinX-XL, DDR, PC3200XLPT, Cas 2
2 x MSI 6800 GTo, Twin DVI
http://www.scan.co.uk/Products...sp?WebProductID=154668
Forceware (currrently) 67.03

I think thats everything....

 

[BladeRunner]

Originally posted by: Last Starfighter

It just *feels* like there is something on the boards that cant handle the combined hi-res data of both cards at high frame rates. Is there a chips that converts the onboard processed data into a signal monitors can understand? I also wonder if it's got anything to do with me having to use a DVI -> VGA convertor.

Yeah as if there is a chip issue with detail or timing under stress but that still doesn't explain the windowed mode what res is it running at windowed, (the highest you can get) ?

Definitely not the DVI-D -> D-Sub dongle as I'm using one of those because my Leadtek GT's annoyingly have dual DVI's, I'd personally prefer one D-Sub on a gaming card, as I still see CRT as the best choice for gaming.

Once data arrives at the primary GPU via the PCI Express bus, all GPU to GPU communication is handled via nVidia's video bridge. The video bridge is a bus that connects directly to the GPU and is used for transferring data from the frame buffer of one GPU directly to the next. nVidia isn?t offering too much information on the interface, other than saying that it is capable of transferring data at up to 10GB/s.

That quote is from a basic overview of what's going on behind the scenes with SLI in an easy to follow Anandtech article HERE

 

[Zebo]

It's a software issue.. try different nvidia drivers to include latest betas

dx9c?
Different mobo bios?

 

[Last Starfighter]

Originally posted by: Zebo
It's a software issue.. try different nvidia drivers to include latest betas

yeah I've tried that - probably tested most of the available SLi compatible forceware drivers, inc. the betas.

incidentally, just flashed to 1003.005, no effect on SLi or anything else discernable, seem stable here.

 

[Last Starfighter]

yeah dx9c too

 

[Zebo]

Originally posted by: Last Starfighter
PSU Specs

I dont think it's the PSU but I'm going to stick a new one in anyway just in case.

Possibility

though some people are recommending against dual rail for SLi setups

They smart, dual rail sucks for SLI cause 2nd rail (molex rail) is too weak when powering those cards, all the fans, drives etc. Each card uses ~90watts off 12, add each HD uses 25, fans ~3 add it all up and... you could easily exceed 25A!!!! most aux 12's are ~18

But I don't think that's your problem... comp would shut itself down under gamming if it were

If you want to try another PSU this model is the way to go, member exdeath is exteme SLi overclocking with it.. Fat 33A 12 volts rail and high value at only 77USD

http://www.newegg.com/app/View...=17-103-456&depa=0

 

[R1ca]

Has anyone checked their hardware with the list at Nvidia for NVIDIA SLI Certified Products.
http://www.nzone.com/object/nz...ardware_certified.html
Just because it has the capability does not mean it works properly.
Thats whats so good about Hardware revisions, they can be changed.
Last Starfighter
MSI has a 6600 GT and a 6800 Ultra but no 6800 GTs on the list.
Maybe just a fluke but maybe not.
Did everyone that has a problem check to see if their non working hardware is on list list
or more importantly NOT on the list.
I would have checked the list before buying.
SLI is still new and is bound to run into problems.
Just my 2 cents.

 

[R1ca]

Did you try the setup with a different SLI Bridge chip.
Ask Asus for a new one.

 

[Last Starfighter]

Interesting - I was looking for that list when I was purchasing, couldn't find anything like it.

What prompted me to get the cards was the retailers website:

Link

which says

Yes this card DOES support SLI.

that and it's the only 6800 card available in the UK afaik.

Easy enough to RMA them I suppose! Hope we get some more frickin 6800s on PCI-E in the UK a bit frickin faster.

 

[R1ca]

Because you said that corruption is allways on the bottom
maybe the second row of 16x is not functioning properly.
Did you allways have the bridge chip facing the same way.
Try putting it on the other way and see if the corruption is on top.
I dont know just a suggestion.

 

[Last Starfighter]

Originally posted by: R1ca
Did you try the setup with a different SLI Bridge chip.
Ask Asus for a new one.

It had occured to me, but they look so uncomplicated & hard to imagine there's much that can go wrong with them.

Asus aren't particularly interested tho :-/ :

***

Dear Alex,
We haven't encountered any problems but obviously we can't speak for MSI who make the cards. I would suggest you contact them and see if they have had any problems.

***

Dear Alex,
I can only reiterate that there are no known issues with any cards. perhaps the games themselves need patches or updates.

***

Perhaps the cards need newer drivers, or maybe there's a PSU problem, other customers have set the SLI up without these problems and there isn't a list of approved SLI cards.

***

Dear Alex,
This is more likely to be related to the graphics cards themselves. Try the latest Nvidia drivers or check with MSI.



..... hmmm

and there isn't a list of approved SLI cards

 

[Last Starfighter]

Originally posted by: R1ca
Because you said that corruption is allways on the bottom
maybe the second row of 16x is not functioning properly.
Did you allways have the bridge chip facing the same way.
Try putting it on the other way and see if the corruption is on top.
I dont know just a suggestion.

I did wonder about that - alas it makes no difference

 

[R1ca]

Originally posted by: Last Starfighter
Interesting - I was looking for that list when I was purchasing, couldn't find anything like it.

What prompted me to get the cards was the retailers website:

Link

which says

Yes this card DOES support SLI.

that and it's the only 6800 card available in the UK afaik.

Easy enough to RMA them I suppose! Hope we get some more frickin 6800s on PCI-E in the UK a bit frickin faster.

Because the list at Nvidia for supported 6800 GT SLI cards is smaller than the other 2
there might be other problems there.
They dont support the higher clocks of the ultas for a number of reasons maybe
some graphical problems are incured on a small amount of Ultra rejected chips
when the Largest bandwidth is used in SLI.
When you go to a window its not running full on and is using some memory for the
desktop window which isnt graphics intensive.
Clean inside the PCIe banks with compressed air REALLY good maybe there is some
contamination from the factory.
Look at the PCIe trace lines any bulges on the board or indents are both banks tight
no looseness at all.
Just some suggestions

 

[Last Starfighter]

The board looks ok on close inspection, but I'm far from an expert in such things.

What you said did remind me of this article though:

Driver Heaven GTo Review

Particularly the bit about them being Ultras with duff pipelines.

 

[BladeRunner]

You may have something here, I'm sure before xmas OCUK, (where I got my Leadtek 6800 GT's from), had their budget GTO's shown as SLI compatible. It no longer mentions SLI in the specs, so I wonder if they've had returns / complaints? I'm 90% sure of this as I nearly got them because GT's were so rare, but held off as I didn't like the look of the 12 pipe 900mhz ram spec.

I know that you can get some graphical weirdness with an X800 pro vivo that's bee softmodded to X800 XTPE if one or more of the re-enabled pipes is faulty.

http://www.overclockers.co.uk/...py_of_6800_Series.html

 

[remmerst]

I'm running a pair of eVGA ultras, though, and hsving the same problem. I don't think there is anything soft modded out in that card is there?

 

[BladeRunner]

I wasn't talking softmod as such, more refereing to the fact one pipe could be ever so slightly below par so only showing up under the full screen sli load. I admit I / we are clutching at straws now, but we're running low of things to suggest for you to try. Id be the first to blame the hardware, chip, or PCI-E bus if it weren't for the fact it works ok for you in windowed mode.

 

[bkzshabbaz]

General question....does this oddness happen in any other game? Do u have Doom3?

 

[Last Starfighter]

happens in Far Cry, HL2, Eve - I can install more stuff I suppose!

 

[R1ca]

I would like to know the percentage of the Ultra cards that are failing.
It would seem that more GTs are failing.
Maybe its a problem in the SLI Traces or interface somewhere
that are exaterated when the Pipes are not formed properlly.
Even a small anomally in a few Ultra boards, not bad enough to fail
Pipe inspection but fail when put in SLI.
Im just suggesting some things.

 

[genemd]

Same problem with HL2. I have one 1/2 inch wide vertical green bar on the left that fluctuates its vertical size continuosly like its monitoring something and a thin horizontal geen bar that is permanant right across the middle of whole screen. Otherwise the graphics are awesome. The bars disappear in windows mode. Don't have any other high demand video games loaded yet. This is a new build and I am having problems with spontaneous rebooting (see my other thread) so not anxious to put much else on yet.

 

[xtknight]

Originally posted by: genemd
Same problem with HL2. I have one 1/2 inch wide vertical green bar on the left that fluctuates its vertical size continuosly like its monitoring something and a thin horizontal geen bar that is permanant right across the middle of whole screen. Otherwise the graphics are awesome. The bars disappear in windows mode. Don't have any other high demand video games loaded yet. This is a new build and I am having problems with spontaneous rebooting (see my other thread) so not anxious to put much else on yet.

You need to disable that status option in the nVidia control panel :laugh:

 

[genemd]

xt
looked around but not sure what you are referring to?

 

[Ballistyx]

Ummm.. just a quick observation (feel free to ignore it if you like).

Doesn't the Asus A8N-SLI have a "special" feature called P-E-G mode, where it (covertly) overclocks PCI-x graphics cards? You might want to try turning it off and see if it resolves it. It could be either a timing issue between the two cards when PEG mode is enabled, or because PEG mode is overclocking the cards (and the cards' can't take it).