I am using deep learning in my research and I added a second 1080ti to my computer, because using only one would not let me finish my experiments in time. Tensorflow will utilize almost 100% GPU and some CPU.
Problem: If I run tensorflow with both GPUs the system turns off after about 30s and will not POST. I have to remove the first gpu to power on again (afterwards I can add the first GPU again).
- 1x be quiet! Dark Power Pro 11 750W ATX 2.4 (BN252)
- 1x ASUS Prime X370-Pro (90MB0TD0-M0EAY0)
- 1x AMD Ryzen 5 1600 (TDP: 65W), 6x 3.20GHz, boxed (YD1600BBAEBOX)
- 2 x MSI GeForce GTX 1080 Ti (250 W) Gaming X 11G, 11GB GDDR5X, DVI, 2x HDMI, 2x DP (V360-001R)
- 1x Samsung SSD 850 EVO 250GB, SATA (MZ-75E250B)
- 3x Seagate IronWolf NAS HDD 10TB, SATA 6Gb/s (ST10000VN0004)
- 1x G.Skill Aegis DIMM Kit 16GB, DDR4-3000, CL16-18-18-38 (F4-3000C16D-16GISB)
I use PCIEX16_1 and PCIEX16_2 for the GPUs.
What I tried so far:
- Running tensorflow with each of the GPUs (100% GPU utilization) --> OK for both GPUs
- Checking the temperatur of both GPUs while using GPUs in parallel: --> OK max temp <80°C
- Double checked that nothing is overlocked --> OK
Can someone please guide me trough the next steps to figure out the problem?
Thank you all for your help.