4 x 290X problem

[davejcb]

Hi all,

I have the following together-

Asus Z87-WS (latest BIOS)
4x XFX 290X
LEPA G1600
Ubuntu server 12.04.4
Catalyst 13.12

I have another build with the exact same hardware, works perfectly. This one, for some reason, gives me problems.

If I display an lspci, it shows the 4 cards. If I display a clinfo, it shows 3 cards, if I display od6config, it displays 3 cards.

I've formatted, reinstalled, and it did show all four cards in clinfo after the initial driver install, then I installed other things, od6config, etc, rebooted, and can't get more than 3 cards detected.

I've uninstalled (with --force) and reinstalled the drivers, same.

All power connectors are connected to the cards, and the three 6 pin EPS motherboard connectors are plugged in.

Ideas?

[undeath]

did you test the card separately? maybe one is defective.

[davejcb]

I'm confused.

I've reinstalled the o/s, with one card installed. Added each card one by one, doing an "aticonfig --adapter=all --initial -f" each time, all four detected, both in aticonfig and in clinfo. The cards were also physically swapped to check they are all working.

Then, I install od6config- works fine, I do an "od6config --get temp", all four cards are there. I edit my .bashrc to have "od6config --adapters 0,1,2,3 --autofix --set core=1100,mem=1375", reboot, and I get the damned message from od6config telling me there are only three cards again.

I check lspci, four detected. I check "aticonfig --adapter=all --odgt", four cards. I check "clinfo | grep GPU", three cards. If I run oclhashcat, three cards.

Redid another "aticonfig --adapter=all --initial -f", same.

What the hell?!?

[davejcb]

UPDATE-

Went and installed the latest driver (14.3 beta), and now when I do an "aticonfig --adapter=all --odgt" I get-

ERROR - Get temperature failed adapter 2-supported device 67B0

I looked at xorg.conf, and it does show all four adapters.

I'm at a loss.

[epixoip]

od6config --adapters 0,1,2,3 --autofix --set core=1100,mem=1375

this command makes no sense.

[davejcb]

I actually have "od6config --autofix --adapter 0,1,2 --set core=1100,mem=1375", writing from memory.

What's wrong with that syntax? Combining the autofix with the set in one line?

Any chance od6config could have screwed something up?

Thanks!

[Milzo]

UPDATE-

Went and installed the latest driver (14.3 beta), and now when I do an "aticonfig --adapter=all --odgt" I get-

ERROR - Get temperature failed adapter 2-supported device 67B0

I looked at xorg.conf, and it does show all four adapters.

I'm at a loss.

v1.01 nor 14.x beta drivers are fully supported with 290x as yet.

[davejcb]

I'm back on 13.12 drivers.

Also, I have an other box with identical hardware and software working without issues. I think I wasn't clear- on the 14.3 drivers, the other 3 adapters' temperatures showed, only Adapter 2 did not.

[Milzo]

you said your have 4 cards detected before you ran --autofix so try setting each cards parameters manually.

od6config --adapter 0,1,2,3 --set core=1100,mem=1375,fan=100,power=+50

[epixoip]

I actually have "od6config --autofix --adapter 0,1,2 --set core=1100,mem=1375", writing from memory.

What's wrong with that syntax? Combining the autofix with the set in one line? Any chance od6config could have screwed something up?

well first you don't need to specify --adapter unless you only want to set a specific adapter. it defaults to all adapters.

but i have no idea how od6config will behave when passing both --autofix + other set params. that's not a use case i tested, as it's not one i would have ever expected, and i wouldn't be surprised if it indeed screws something up. either you use --autofix, or you use --set. it's one or the other, or neither, but not both.

and one more thing... 290X is not stable at 1100 Mhz unless you overvolt it. especially with overclocking the memory as well. and i certainly wouldn't recommend going with such an aggressive overclock right out of the gate when you're already having stability problems.

how about trying something more sane?

Code:

od6config --set core=1000,mem=1250,fan=80,power=50

or better yet, just

Code:

od6config --autofix