Graphics Card: (2) Vega 64 in Crossfire (Sapphire reference design)
Operating System: Windows 10 Pro x64 v.1709
Driver Version: Adrenalin 18.2.3
Display Devices: Monoprice MP 27" Zero-G, 2560x1440 @ 120Hz, Freesync, Display Port
ASUS VZ229, 1920x1080 @ 60Hz, HDMI
Motherboard: MSI X370 XPower Gaming Titanium, Bios 7A31v19 (9/19/2017)
CPU: Ryzen 7 1800X (no OC)
Power Supply: Corsair AX1200i (set for single-rail mode)
RAM: GSkill Trident Z Series 16GB, model F4-3200C14D-16GTZKW
I've seen a lot of similar complaints about display loss but not real resolutions, and none that mentioned an accompanied sound.
Randomly my screen will just go black, I will hear a light click sound at the same time. The system is still on, fans/lights are on but no video output. Hard reset or power off/on and the system boots up normally, resetting Radeon settings to defaults - but without the normal 'settings have been reset' message in the corner of the screen. The issue happens under load and occasionally at idle. It usually occurs 0-3 times per day though it almost always occurs while I am using or was recently using the machine, only once or twice have I found the machine already in this state when I sit down to use it (system is on 24x7).
I don't OC the cards but I found they perform best undervolted and at low temperatures. My wattman settings lower the voltage settings for core and memory to 1075mv and 1050mv. Fan is set to auto with max speed increased to 3800RPM, target temp is lowered to 68c and power is left at 0%. Note: the problem occurs with wattman set to default settings also and increasing or decreasing power/heat doesn't effect how often this happens.
First off, I am running 4 separate 8-pin power cables from the PSU to the cards, I've seen a lot of people bring that up.
Case ventilation is good. Intake is 3 filtered 140mm and 3 filtered 120mm at front and side. Exhaust is 2 120mm at top (AIO), 1 120mm exhaust on back, and of course 2 Vega 64s exhausting out the back.
As the screen goes black I hear a click sound, like a relay or something is being tripped. My first though was a power supply problem so I switched to an HX1200i for a few days but the problem continued. Turning off Crossfire hasn't helped. The issue has been persistent across numerous driver versions.
I am in the [slow] process of testing each card individually. I ran just the top card (card 1) on it's own for about a week but the issue continued. I am currently testing the second card on it's own in the same slot. It's only been a day or two but so far no crashes.
So, for my question. Does anyone have any ideas OTHER than the possibility I just have a bad card?
Did you ever figure it out? Cause i have the same issue with my sapphire liquid cooled, I just use it for gaming and work. I RMA'd it and they said there was nothing wrong with it. Then I rebuilt my whole PC and the only parts that are the same are the hard drives and that one card. It still does it randomly even at idle. Black screen, the tachometer goes dark and I have to do a hard reset to get it to work again. Some days it works, others I just give up working on my PC cause it does it so often.
No. I have not been able to reliably reproduce the problem. I would RMA the card (I have 2 and only one has this problem) but I'm sure they'd send it back saying there's nothing wrong, as happened to you.
It's clearly a problem with the card, the problem occurs with card A&B installed, with only card A installed, but not with only card B installed, using the same slot for single card testing.
Like you, I've updated my system since my last post, new motherboard & chipset, CPU, memory, still happens on a fresh Windows 10 install.
I don't know what else to do except buy from another manufacturer. I've used Sapphire cards exclusively for 10 years, since my pair of ATI Radeon HD 4870s, and this is the first time I've had a failure. Considering this seems to be fairly "common" as far as Vega failures go, and their support team doesn't care to address it. On top of that only offering a 2 year warranty, it may be time to take my business elsewhere.
For comparison, a friend bought a liquid cooled 1080 Ti around the same time - it came with a 6 year manufacturer warranty. Two years is pathetic.
Is anyone having this problem with a non-Sapphire card?