Fixing an Alienware 18 with seemingly faulty GPU's

I was discussing this in another thread I made, but this title fits it better to help get the help I need

So I bought a alienware 18, 460CAD, and the owner claimed the gpu's wouldn't accelerate.

After picking it up, he told me he had taken it to a shop to get them to swap the secondary GPU into the primary slot, to no avail. (Read as a 3MB ATI card)

So I swapped the original first one in by itself, and it booted and seemed to work great. R9 M290X.

But I was at a computer tech's place who I was using his table, and he tested furmark. and it seemed to work great. good fps, and the card seemed to work. but the fan didnt come on. 90 Degrees before the pc made a fan noise. Not even a loud one though. 95 Degrees hit, I reached to stop it before it did damage, but he said just wait a second- Then the pc crashed. Duh, Overheating.

But now I cannot boot into Dedicated Graphics, only Intel Integrated. FN F5, press yes, Reboot, and still in intel. I have yet to try the first card again. Guessing we killed it, but unsure.

Tried putting one of my 460M's into it, from my M18X R1, but 8 beeps. Guessing incompatible cards.

Put the R9 back in, boots fine, but wont swap to dedicated. (I havent tried the original 3MB one again that seemed to be fried)

Then I encountered a new problem after re installing windows, after i log in, its just a black screen. Explorer.exe wont open. Then one time it did, about 4 reboots later. It was about 3AM at this point, getting pretty tired. So I updated my bios, A13. Just now am i realizing this is probably a mistake to have done, because Im guessing I cant flash to unlocked A12 now can I? Or can I?

TL/DR - Trying to determine if mobo or cards is the problem - Ill order a new Motherboard if needed, or ill order 980M's if its just bad cards. (and their appropriate heatsinks) But I wouldnt want to just fry more cards if its the motherboard. :|

Thank you in advance anyone!







@ electrosoft @ Raidriar , you guys were on my other thread so I thought id tag you

 

You ran furmark and let the card overheat, chances are you killed it.

Also R1 is not compatible with maxwell, only 800series or older.

 

Yeah, honestly im not that suprised. Oh I should mention - The guy I bought it off of said he bought it, played Duke Nukem Forever for about half an hour before the pc crashed, and wouldnt play games. so it sounds like his first card overheated and bit the dust too. I just partially wanted to make sure Its not like a motherboard issue thats messing up the cards.

 

He said Alienware 18 R1....can definitely use maxwell cards

@ColaMonster you can send me the machine (or the cards) and I can test and return to you. More than likely motherboard is fine and the R9 M290X are fried

 

@Danishblunt yeah ive got an M18X R1 and i newly purchased an AW18, which im trying to troubleshoot and upgrade

@Raidriar i might just take you up on that. i might test the M290X in my m18xr1 as well first, since i do believe they are compatible

 

Yes it will work in the M18x R1, but fan control doesn't work, you will need hwinfo to control the fans

 

my 460M's are not compatable in the AW18 im guessing from it just 8 beeping me when I tried?

 

Possible, but weird that it would not work. Do you have any other GPUs laying around? Can the system POST with the R9 M290x before loading into windows?

 

So with the original one that was reading as a 3MB ATI card, it still posted with that card.
With the newly fried R9 M290X, it still posts, but refuses to even boot into dedicated graphics, or seemingly even attempt to. any time i FN F5 to swap over, it tells me it needs to reboot to make the swap, seems to reboot fine, but immediately is still in integrated graphics. in abit i can try the first fried one again, try to get into dedicated graphics.

Unfortuneately I only have 2 460M's and 2 (dead) R9 M290X's, and i live in a small town where I dont think anyone has any extras to sell me

Edit : Managed to get rid of the black screen error, idk what it was, reinstalling windows fixed it.

Would rolling back to A12 from A13 Bios , then Going to Unlocked A12 allow me to force the system into trying to use the graphics card?

 

BIOS has no effect, stay on A12. If you can POST with the R9 M290X then fail trying to get into windows, the motherboard is good. Black screens or BSODS after POST mean failure to hardware accelerate graphics, typical of graphics card failure. 460M may be incompatible with the Alienware 18 BIOS structure.

 

My bad, thought he was talking about his M18x.

@ColaMonster
No it won't the card is most likely dead.
You went without fan control with furrmark on graphicscards and killed it with it.

 

I had upgraded to A13 last night, so i should roll it back then? (If thats possible)

Technically I havent seen any BSODS, and the only black screen ive seen was after I cooked the R9 M290X, I fixed the issue of windows not openening properly with a reinstall, I still had a cursor and was able to use task manager with my black screens failing to get into windows. But thats great news that the mobo is probably good!

 

Send me the cards alone and I can tell you for sure if they are toast or not

 

You can also make pictures of both cards, maybe we can see which VRM died or something. Maybe you can get them repaired relatively cheap.

 

I popped in the card that i cooked, no response, i should have tried fn f5'ing it, but i got ahead of myself and just took it back out, and put in the one that earlier read as 3mb ati card. it appears to be alive again? exept one problem. the moment i get amd drivers installed, black screen. rebooting shows windows symbol, then black screen the moment it tries the driver. without the driver it boots though?

also i did take photos of the cooked card, nothing out of the ordinary as far as I can see.

EDIT :

So I ran userbenchmark to test if the gpu still worked, and had hwinfo to monitor temps - cpu temp climbed fast, hit 90C before i was able to crank fans on it, because i think i had it set manually to a low RPM accidentally, but it went sys auto and it worked fine

But then the GPU test hit, and the gpu temp climbed and climbed, and the fan controller had no effect on it - it hit 90 degrees and kept climbing, till i just used the power button to force the pc off - no damage done. so my question became - why is there no fan.

unable to get it going with speedfan or hwinfo, and running the system diagnostics from the boot menu, it was like 8 tries before it sped up at all.

so i replaced the fan with the same one out of my M18X (which 100% works) and tried the diagnostics again. took a photo of the result. basically same results as before though. fans spin up during boot, and when the diagnostics first initializing them? confusion. I see why I burned the one, and possibly why this one got messed up to begin with.

 

This is classic GPU failure with AMD, dead 7970Ms act exactly in this fashion. The core is dead/damaged, no way to fix this short of a repair that costs as much or more than the card itself. It works without driver because windows is software accelerating the graphics (CPU) instead of hardware accelerating graphics with the GPU.

 

What @Raidriar said is the most likely scenario.

Usually when gpus are starting to fail they will:

Black screen
Fail to initialize
Show up with an error code under Device manager
Fail after driver install when gpu acceleration kicks in
Random hang
Fail at lower thermals
Artifacting
Freeze/reboots

Usually when you get to the 8 beeps of total death, you don’t realize you’ve actually experienced one of thee above as a sign of impending failure.

The 7970m’s were dogs.

I’m not sure why manual fan controls failed.

You could send them to Raid for testing since he keeps a literal arsenal of alienwares (sans his AW18 that died).

But if the system works fine with no gpu(s) installed then focus on them. You could always pick up a dirt cheap 765m to fully test your primary gpu slot.

As for your cpu spiking into the 90s, that is every AW18 I’ve encountered basically until you repaste with something like nanogrease (to compensate for any pairing issues) or LM if you have a tight enough fit and use XTU to find your sweet spot.

I took possession of (yet another lol) a brand new AW18 4710mq and dual 970m”s last week (deployed 2016 tag and shipped with A13 already applied) and right out of the box the stock cpu settings spiked immediately into the 90s before crashing running wprime1024. How these left Dell all those years still with atrocious pasting Jobs is beyond me.

 

Not entirely dells fault. The haswell chips ran stupid hot (poorly designed FIVR) and any high clockspeed haswell almost required liquid metal and an undervolt

 

I did manage to get amd drivers to work last night. Took a lot of messing around, but I got it working I believe. But the fan won’t spin up, so I can’t keep it cool. :s

I’m worried if I get a new video card it’ll just cook too

 

This is a stock factory configuration, you shouldn't need any fan control applications. You may have an unplugged or damaged fan.

 

When I fried the other one with furmark, no fan spun up till 90ish degrees. I swapped fans out and put my m18x fan in that I know works, same thing.

Im wondering if the fan controller on the mobo died/is dieing or something.

In HWinfo I can only change the speed of the CPU fan, not either gpu fan. is there any workaround i could do do you think? I know piggybacking on the cpu fan speed is probably possibly, or if there is another way to get that fan header to work properly? :s

 

userbenchmark.com/UserRun/8684687 I hope links are okay. I ran this quick benchmark and it seems like the card is still alive. yet the fan wont spin up?


@Mr. Fox I wonder if you know why i cannot get my GPU fan to work properly? (i tried 3 seperate fans now, 1 from my m18x that i know works, and the left and right fan from this AW18 that seems to spin up in diagnostics.)

 

EC chip fried most likely and cannot properly communicate. This was a stock card in the Alienware 18 it should work 100% otherwise assume defective. If both cards act the same way then it could be motherboard, unlikely both cards fried their EC chip exactly the same time

 

Other card wouldn’t boot since I fried it from its fan not kicking on. Could try it again after. This one boots but won’t spin the fan. Is there no way to force it to spin?

 

If hwinfo didn’t work, then no

 

Does it sound like a blown EC chip still though?

 

Did you try reflashing the vbios on the card(s)?

Also in HWinfo are you able to access the fan controls for fan1 and fan2 or are they greyed out?

 

I haven’t, was having a little trouble finding r9 m290x vbios for the Alienware

And it says fan 1&2, but only effects the cpu fan :s

 

I have dell r9 m290x bios if you need it. I back it up when I had those cards.

 

Id love it. it should work the same as the alien ware one correct?

 

Yes it's the dell r9 m290x from alienware 18. I'll look for it when I'm home and free.

 

I've sent you a PM with the stock m290x bios attached.

 

Thanks a tonne! Ill let you know how it turns out

 

Apologies everyone for never letting you all how it worked out, I got very busy lately, lots going on.

Unfortunately updating the VBios had no effect, didnt seem to change anything at all, and I noticed in GPUz that it wasnt even registering GPU temps properly, so Im going to assume it was the cards issue.

I ordered a 1060 , that should be here in a few days, which I will make a new thread for that project Hoping its not the motherboard, but Ill know by the end of the week