5870m VBIOS

[radji]

Gots a question for all y'all. I've been going back and forth with someone trying to get a pair of 5870ms to work in his M17x R1. Attached are the GPU-z screenshots he took for each card. We were successful (I think) in flashing the latest VBIOS to his cards using the dell flasher utility. The user stated the flash utility said Update Complete but also gave red letters at the bottom: 0FL01

Now that code is for an invalid card or GPU detected IIRC. Would that correspond with the fact that his GPU-z info for one card doesn't match the other? I suspect one of his cards isn't the MXM 3.0b version cause the subvendor is invalid and the PCIe version reports as 1.1, which I thought was only present in the MXM 2.1 type III card? Any clues on this?

Card 1

Card 2

[Michael]

Só the problem is that crossfire isn't working? Did you try flashing the card from dos?

[radji]

Yup. I had him flash the card from DOS using the "official" Dell VBIOS for the 5870m. Then I had him take card 1 out and put card 2 into the master slot. The system wouldn't boot. Gave the GPU error beep code. I think he may have just been given a bad card since the BUS interface is not reporting PCIe 2.0 like it should.

[svl7]

The card can't be MXM 2.1, first of all there's no 2.1 5870m, and second it has a different connector, you'd have to use serious force to put it in a 3.0 socket

The difference in the PCI bus speed can be explained easily, the card slows itself down at idle for saving power, that's absolutely normal, and will also occur with a 680m.

0FL01 is an error of atiflash, at least one card didn't get updated, however both cards already seem to be on the latest Dell vbios according to GPU-Z. I don't know why the subvendor ID is unidentified, but as GPU-Z seems to read the subvendor ID not from the vbios but from the adapter it is possible that there's an issue with the second card, or adapter.

Also it surprises me that the first card seems to work fine with that vbios in the R1, I thought that it required the older vbios in that system. Anyway, check the CF cable.

[radji]

I know it can't be MXM 2.1, it was just the first thing that came to mind. I was thinking maybe he got a Clevo version of the 5870m or something. At worst he was sent an MXM 3.0a, but I told him to check the appearance of the slave card and make sure it was totally identical to the master card. I checked on the possible issue of the R1 and the new VBIOS. None that have the 5870m crossfire setup have any issues with running the latest VBIOS with their cards. I didn't think about the crossfire cable though. I told him to check that for damage justin case. I still think his secondary card is bad.

Here are the before shots with GPU-z (before he did the BIOS update):

Master:

Slave

So either the system (or drivers) are not properly interfacing with the card, or there is something really weird with his card. I also told him to take the secondary card out, clean the contact face with ISPA and hit the MXM slot with a can of air to get rid of any lingering contaminants that could cause the contacts on the card to not make contact with the MXM port. I've also got the master and slave ROMs for the 5870m from StamatisX so we can try a BIOS that we know is for the slave card. Other that that I'm out of ideas.

[svl7]

There's no slave / master vbios, both are the same. Also there's no MXM 3.0 type A version of this card. Did you check the CF cable? What bios is that system running?

[radji]

There's no slave / master vbios, both are the same. Also there's no MXM 3.0 type A version of this card. Did you check the CF cable? What bios is that system running?

That's what I though. StamatisX says the master and slave VBIOS are slightly different though. He hasn't gotten back to me on the crossfire cable. His system is running A10 unlocked BIOS.