This might be a specific problem related to my overclock but may be related to Ubuntu or AMD GPU driver. I am not sure.
System Specs:
Intel i7 6800K (overclocked to 4.2Ghz running at 1.40V)
96 GB DDR4 RAM (at 2400 Mhz, not overclocked)
Ubuntu 16.04 LTS
AMD RX 470 GPU
amdgpu-pro-16.50 driver (proprietary) , tried 16.40 / 16.60 also.
- I installed 16.04 WITHOUT any overclock in the beginning and the system ran normal. No issues at all. I ran several benchmarks to test my setup since the system would be used for calculations. It ran for a two days like this with 7-10 reboots.
- Next, I moved to overclocking. Overclocked to 4.0 then 4.1 and then 4.2 Ghz. At each of these frequencies it ran stable and completed the benchmarks, the temperature not going above 70 C). The system was deployed for work. After a few days, I could not login via GUI but only by ssh login. So the screen stays blank all the way, like no video output. Next, I reduced the overclock to 4.0 Ghz and still this problem came up after a day(basically after a reboot). Otherwise, the system runs without any issues (since I can work via ssh).
Issue: The GUI login goes away and the screen stays blank (permanently) after a couple of reboots. The system performs normally, and benchmarks and temperature show that it is stable.
Benchmarks ran:
stress
for 12 hrs (mix of CPU and memory).
prime95
for total of 12 hrs.
unigine valley
and unigine heaven
full tests for graphics card.
memtest
for memory.
All benchmarks were cleared without any issues.
Following is a temperature profile for CPU when using stress
for 5.5 hrs using all cores then following by 30 mins of (memory + 3 cores). As clear from the temperature profile cooling looks adequate.
Updates:
- Removed the
amdgpu-pro driver 16.50
, and installed16.60
version. Also triednomodeset
as the grub parameter. Problem persists. amggpu-pro driver 16.40
is also tried. Problem persists.- overclock removed and problem persists.