Bitcoin Forum

Hello!
Died 5 cards rx 480 ref(xfx, gyga, 2 powercolor, his) After downloading Windows error thread stuck in device driver. All 5 of the graphics card died in the first 24 hours. The remaining 30 pieces of work from a week to a month. All the cards worked on the same settings. GPU 1130mhz 900mv, memory 2200mhz 900mv. Mining only eth. Flashed the Bios, only the timings. 2 of 5 dead cards were not flashed. Visible damage no. In safe mode is determined (disabled drivers are also).

Could it be because of settings?(Chip 1130mhz 900mv, memory 2200mhz 900mv)


 The monitor sees this picture.
https://pp.vk.me/c604329/v604329301/29def/TAWP1MMW2HI.jpg

Gpu-z other working card on screen.
http://dl1.joxi.net/drive/2016/08/30/0008/1579/542251/51/43c407dbf2.jpg

First, they're very hot. Also, did you tried to reinstall Windows ? On a bad quality HDD, when I turned off the system, I would block like you. If you find this too painful, you can also try to run the GPUs in another rig, since you have several. All of this is to make sure that they're really dead.
Did you downclocked or overclocked it ?

they just die, replace it by warranty

GPU 1130mhz 900mv, memory 2200mhz 900mv
This is 100% not bad hdd. Cards died

5 graphics cards died in different configurations and in no way related. Different power supplies, different sdd drives. All died in the first 24 hours of mining.

Interesting, I got 1 died yesterday with same error as yours... It was Sapphire 480 Reference model...

Exactly the same behaviour as yours...

My all GPUs run at 65-70c max, so its not a temp issue...

EDIT: Card died after 5 days of mining

More info about motherboard and risers please. Were cards connected directly to the mobo or through risers (what kind of risers, molex connector, berg connector)?

Different configurations:
all rigs windows 10
motheboard - asrock b85 anniversary, H81 PRO BTC
psu - fsp aurum 1000w gold, COUGAR CMX700 700W bronze, Deepcool DA650 650W bronze
powered risers - http://risers.ru/images/riser_PCI_E_1x-16x_power_molex_onboard.jpg
                       https://ae01.alicdn.com/kf/HTB1EP0DLXXXXXa7XXXXq6xXFXXXa/2015-PCI-E-1X-16X-Flexible-Extension-PCI-Express-1-to-16-X-Adapter-PCI-E.jpg_220x220.jpg

A dead card is replaced by working in the same place and work well

Yup. Same here

H81 btc asrock, connected via USB Riser, 750W cooler master PSU.

This rig had 3x rx480 undervolted to:

800mA@1000mhz core clock with mem clocked at 2000

GPU Had artifacts on half of the screen when conected to monitor via hdmi. And with that card connected winodws eould always celrash during boot. RMAed waiting for replacement card to return.

Maybe the reason downvolt?

I have 30 pieces of work from a week to a month.
How many cards do you have?

9pcs

both of you have risers on the dead cards am I correct?


I have 20 cards all good all have no risers.

My guess is it is a riser issue.

And what were the vram timings? default? https://pp.vk.me/c630716/v630716944/4968c/QkL9uTMphN0.jpg

Number of rigs should be under 20 cards? What motherboard?

Yes. Default timings. All settings were done in wattmann with stock BIOS on all cards.

So funny Phil, this attack against risers :D. Though I'll listen to you for my next rig, by not using any riser and placing all the cards on a motherboard, for now all my cards except two cards are connected throught risers and I never had any problem.

As long as you run one riser per powerline you should be ok, two per powerline is a 100% fail, just a question about when it is time for a meltdown.
I now run my rigs with 4 x 480 (custom bios 30.5Mhs/gpu) on H81 BTC Pro boards, one riser per rail and they dont get that hot.

The riser was the only thing powered from that powerline so that's not an issue.

BTW nothing was melted or eaven hot to touch.

If the screen have "garbage" and GPU is booting up, defo its a memory problem

Could be a bad batch of memory used, try to find out which memory was used on those cards, there were reports on bad batch of micron chips with MSI Gaming X 1070 card, they were canceled

I had one MSI gaming 1070 die on my, but it just blown up, prolly VRM dies, bad components

They replace that kind of failure on warranty ?

Good hypothesis... but still not a good news.

Thank you very much for sharing OP. the reliability issue is very important...

And it seems that the issue is more widespread than I expected  :-\.

let me ask stupid two questions:

cards not "died" but you see only artifacts  right ?

did you trued to restore original bios ? - is the problem still presists ?

also those ribbon risers are very bad in all cards.
I am bying only USB risers - they are the best.

Cards do not work with the drivers,  BSOD "error thread stuck in device driver"
BIOS restored to stock using atiwinflash

it might be software issue ?
like card bios issue ?

no, I don't think because of this

i have this kind of artifcacts using risers,with 1 gpu it boots up, when i tried 2 or more this kind of bug appears and windows freezes ( tried ethos same error)
try plugin only one gpu to motherboard direct, without risers
It worked for me

And looking your riser, i think you should plug only one 12v rail, the 4 pin one, not the 6 pin

Put in different systems with different OS, without a riser

Doe it boot on security mode? without any display bug?

It has artifacts even in BIOS. So it's not OS related.

I am using 4 Ribbon risers applied with my 4 GPUs, the risers are not powered and working fine as of the moment. cross finger not to experience this one.  ;D