Gigabyte Aorus 1080ti - card not initialising #2

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Sorry, already soldered in before I read this.

However, I still have 419,000 errors on F0 chip in MATS (all others are fine).

I reflowed the chip after the 1st test but not any better. Beginning to think the pcb has taken a hit and possible microfractures somewhere in the data lines or a lifted pad that's not immediately obvious.

Will have to pull the chip again and double check the data pads. Not easy to trace them though as they go direct to the core.

There is only so many times this can be unsoldered/soldered before the traces are wrecked

So it increased from 28917? Still wrote only errors? Same bits or different ones ? So you have the equipment to reball the GPU ?

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Whoop! Success at last.

Removed the F0 chip, was really slow and careful cleaning the board and the chip to get the pads as flat as possible on both surfaces.

I had been using 0.45mm balls but decided to use 0.35mm this time as there is less chance of them bridging and they seem to be tall enough to clear the small ridge in the centre of the ram chip.

Hand reballed the chip again (I actuallly find it quite enjoyable moving 190 balls gently onto their pads . Gently cleaned the chip with isopropol and smeared a little more flux than I usually would onto the board.

Preheated the board to 120c and then used the Quick 861DW at 440c / 70% speed to finish reflowing the solder. Other difference this time was I gave the chip a little nudge and then nudged it back into place and I think this helped the solder correctly join to the pads on the board this time.

Let the board cool for 25mins and crossed my fingers and plugged it into the test bench.

Fired right up with an image! ran MATS 10mb test, no errors, PASS

Need to stress it in windows a little before I get too hopeful, but altogether the best news I have had so far

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Whoop! Success at last.

Removed the F0 chip, was really slow and careful cleaning the board and the chip to get the pads as flat as possible on both surfaces.

I had been using 0.45mm balls but decided to use 0.35mm this time as there is less chance of them bridging and they seem to be tall enough to clear the small ridge in the centre of the ram chip.

Hand reballed the chip again (I actuallly find it quite enjoyable moving 190 balls gently onto their pads . Gently cleaned the chip with isopropol and smeared a little more flux than I usually would onto the board.

Preheated the board to 120c and then used the Quick 861DW at 440c / 70% speed to finish reflowing the solder. Other difference this time was I gave the chip a little nudge and then nudged it back into place and I think this helped the solder correctly join to the pads on the board this time.

Let the board cool for 25mins and crossed my fingers and plugged it into the test bench.

Fired right up with an image! ran MATS 10mb test, no errors, PASS

Need to stress it in windows a little before I get too hopeful, but altogether the best news I have had so far

Sounds great and even more helpful the level of details. Thanks for continuing to update.

[Ulhmann]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Nice study case!!!. Please keep us informed!. Hope finally you had solve the problem.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Installed windows drivers fine, all details are fine in GPU-Z.

Ran MSI Kombustor stress test for 30 mins with the artefact scanner on with no issues at all, card runs cool and gpu core and ram all run at rated speeds with no issue.

Will run some Heaven benchmark later, but have no reason to believe there are any issues with the card now.

I am glad my running commentary is useful for people reading, I know it would have been useful for me to read stuff as much detail as possible about troubleshooting these kinds of things (even down to "how much flux used" and "hot air temperature and speed" etc) as many threads leave out the details and are less useful to others.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Heaven benchmark running fine, no issues.

Added a poll to this thread to see if there is interest in repairing more cards and providing a running commentary of how I am getting on.

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Do please give the commentary on the next card and others!!

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Damn it.

Went to test the card again getting ready to sell it. Won't boot.

Read errors on all banks. Voltages/resistances fine - starting to think the GPU needs a reflow....

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Going back over this video from TechCemetery that I posted in my previous thread as same symptoms again - https://www.youtube.com/watch?v=9M2K...l=TechCemetery

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

So booting card as secondary results in blue screen when initialising drivers. Booting as primary and connecting over remote desktop results in no crash, error 43 in device manager and GPU-Z showing card with bios version but can't see memory size (very much like the last card in my other thread).

Can read bios fine (main and inforom) so don't believe its bios related - its just really odd that this card is now exhibiting the same issues as the other card (apart from the fact that it's not got any shorts, low resistance, all voltages seem fine).

Ram strap resistors could be at fault, checked the other card and they are fine (will need to pull the backplate to test this card, but not expecting any surprises)

Card was running great when tested which is the odd thing - I literally just removed he cooler to realign the thermal pads on the VRMs and put it back together.

There must be something about these Aorus cards or I have done something dumb accidently....

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Thought I might have found something, but think its possibly ok.

Tested the 12v rail resistance on the 8 pin connectors and was reading low (circa 130 - 140ohms) - thought aha, bad ram vrm mosfet.

Then realised the plugs were plugged into the psu (switched off of course). There must be something on the 12v rail with that kind of low resistance (pc fan maybe?).

Thought it might have been a quick win... the search continues...

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Thermal camera image.

Hot spots are GPU core (middle) and the chokes for the memory VRM (bottom).

Also, 1 of the caps there is hotter than the other.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Annotated the thermal image to label what I am referring to

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Been examining start-up voltage sequences, mainly the VREFC for the ram.

For the ram to report write errors on all chips, it must be a common issue that affects all of them (so not a specific chip).

This could either be:

1) Supply voltage issue to every chip
2) Missing initialisation signal / voltage
3) Issue in GPU core

I have checked the VREFC initialisation and notes are on the attached image. The signal (digital) from GPIO_10_FBVREF_SEL appears fine (this was not pulling high on other card). VMEM (FBVDDQ) seems ok, 18V_AON is fine....

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

So this is weird.

Thought I would run MATS again (using 400.250 version), now normally to run MATS when the card to test is secondary you need to pass the -skip_rm_state_init parameter to MODS first.

I didn't do this out of curiosity, ran MATS: "mats -n 1 -e 10" and received:

"Found an NV device at domain 0 bus 5 device 0, but it's memory cycle enable bit in PCI configuration space is turned off"

Ok, so I ran MODS : "mods gputest.js -oqa -skip_rm_state_init"

Then ran MATS again (same 10mb test)

PASS, no errors????

Rebooted with 1080ti as primary, no boot.

Repeated the same steps above - PASS, no errors....

Will now run a much longer MATS test to see if anything else comes up (full 11GB).

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

If this passes I might flash the bios.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

So this is weird.

Thought I would run MATS again (using 400.250 version), now normally to run MATS when the card to test is secondary you need to pass the -skip_rm_state_init parameter to MODS first.

I didn't do this out of curiosity, ran MATS: "mats -n 1 -e 10" and received:

"Found an NV device at domain 0 bus 5 device 0, but it's memory cycle enable bit in PCI configuration space is turned off"

Ok, so I ran MODS : "mods gputest.js -oqa -skip_rm_state_init"

Then ran MATS again (same 10mb test)

PASS, no errors????

Rebooted with 1080ti as primary, no boot.

Repeated the same steps above - PASS, no errors....

Will now run a much longer MATS test to see if anything else comes up (full 11GB).

Ok, gave up on 11gb test as think it will take hours.

Ran 100MB test - PASS, running a 1024MB test now.

Just thinking that if its passing these MATS tests then the ram and memory controller are probably ok and its more like a bios issue (either corrupted or chip is bad)

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

1024mb test took 50mins - PASS - onto the bios...

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Lol, plugged cable into card to reboot into windows and it fired right up with a picture

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Interesting....

1st boot, had picture but was in black and white and stretched (device manager said device is having problems)

Flashed bios to latest from Gigabyte site - F3P, rebooted and now image is correct. Device manager fine.

GPU-Z screens attached