780 bios flash headache

[pgi947]

Ah yes, I did try a complete CMOS clear, also forcing different PCI gens, same result, added sli bridge after driver install, as it came up saying no bridge installed but sli capable prior to that I thought 'jackpot' only to be kicked in the stones again after it had installed.

My theory ATM is the stock bios on these cards is causing the plx chip to drop to 1.1/1.1x or something, which would cause the problems I'm facing. - I can't check this as anything that tries to read the cards = crash.

Its far to much dicking about just to get this board to work with stock cards so 1150 in general can do one, its been a flakey platform since launch, now with PLX adding a big bag of FU into the mix I'm done with it. Golden 4770k heading to the mm shortly.

 

[pgi947]

Ok, noticed a new mb bios is out will flash that tonight.

 

[pgi947]

No joy from a motherboard bios update either.

This is now bugging me more then anything

 

[LtMatt]

Our thoughts are with you in this difficult time. #FirstWorldBiosProblems

 

[pgi947]

Haha

Something screwy with the vbios is happening more info to come!

Edit:

Ok so with both cards flashed to the non working stock bios I managed to get into nvinspector.

Card (0) says it's a 6gb card with no 3d memory clocks and card (1) says it's a normal 3gb card and everything checks out. Could this be why I loose sli options when a bridge is fitted?

Edit 2: may be getting somewhere, fingers crossed and then i'll explain how MSI's PLX readout is the single most retarded thing on the face of the earth.

 

[pgi947]

JACK ******* POT!!!!!

Ok so how this evening went down...

Managed to get Nvinspector to fire up with the SLI bridge removed on the stock bioses (gpuz would still hang the system). Upon firing it up my cards looked like so:

Card (0) 6144mb all clocks correct to stock apart from Memory boost clock, was greyed out

Card (1) 3172mb everything looks as it should.

So, my primary card - top slot is acting up. So I thought. I can differentiate my cards by the memory IC, top card is Samsung, bottom card is Elpida. So out of interest I checked this in Nvinspector...

Card (0) Elpida
Card (1) Samsung

Wat.The.Fudge

Now I have two bioses I use for stock, each for the specific memory IC, as I presume the timings differ between the two. I even have them labeled as such, samsung.rom & elpida.rom.

I was working on the presumption that, in Nvflash my card order was as so:

i0 = PLX
i1 = Samsung
i2 = Elpida

Nope, turns out it is:

i0 = PLX
i1 = Elpida
i2 = Samsung

so, reflashed the cards as per the above and a DDU sweep and clean drivers and bingo-was-his-name-o. Everything working as normal, on stock bioses with SLI enabled...

So, seriously now MSI, what on earth is up with your lane allocation? No wonder the PCI kill switch didn't work, god knows what switch relates to what lane.

Dumb As ****!!!

 

[LtMatt]

Well done mate. Virtual pint on the house.

 

[setter]

Nice un, well does socket 1150 get a second chance then mate.

 

[Gregster]

Glad you got there Paul and what a Pita.

 

[pgi947]

Well done mate. Virtual pint on the house.

...snip...

Sod that, after this little head **** i'm going for the fire water

Nice un, well does socket 1150 get a second chance then mate.

For now...

Glad you got there Paul and what a Pita.

Totally, at least I know how to fault find this kind of thing if it happens again.

I'm a wizard with Nvflash now too!!

 

[bemaniac]

That is amazing. Well done working that one out. Makes my 4k sli and bios woes seem relatively simple

 

[almighty15]

I love this thread, you flash your card and MOD it and when you can't put it back it's somehow the motherboard vendors fault?

That's the risk you take playing with ****... do not blame anything but yourself.

 

[pgi947]

I love this thread, you flash your card and MOD it and when you can't put it back it's somehow the motherboard vendors fault?

That's the risk you take playing with ****... do not blame anything but yourself.

I was right though wasn't I? In fairness.

The lane allocations MSI have given the board due to using a PLX chip to manage the lanes is in a screwed up order.

Even NVCP says my monitor is connected to card 2.

From the top CPU slot my lanes look like so;

PCI_X16 - Lane 2 (should be 1?)
PCI_X16 - Lane 5 (should be 5?)
PCI_X16 - Lane 1 (should be 2?)
PCI_X1
PCI_X16 - Lane 3 or 4 (should be 3?)
PCI_x1
PCI_X16 - Lane 3 or 4 (should be 4?)

If MSI had the lanes allocated as everyone else does (and as Gigabyte did on the Z87X-OC I have) then I wouldn't have hit this issue in the first place and thats a fact. The numbering on their own 'Ceasefire' switch on the board doesn't even match up to the boards layout.