I recently acquired a 3070 Ti and once I put it into my mining rig it started frequently hard rebooting. Up until this point the rig has been 24/7 stable with 4 cards for the past 6+ months. Running HiveOS and I checked all of the system and miner logs and there are no errors, the system just immediately restarts. I tried many things like using different risers and tweaking the BIOS but ultimately this seems like a power related issue.
Rig specs are:
EVGA 1300w G2
Intel Pentium G440
MSI Z170A Pro Carbon Gaming
2x4gb DDR4 2400 Memory
PCE164P-N03 VER 006 Risers
2x 3060 @ 125W
1x 3060ti @ 157W
1x 3070 @ 157W
1x 3070TI @ 250W (new)
Total PSU load with all 5 cards is only 814W and about 900W at the wall, well under the 1300W spec
I temporarily installed Windows 10 on the rig so I could get some more information using HWINFO like the voltages the cards are seeing. At idle the 12v PCIE slot and connectors are all 12.1-12.2V across the board. At load things get interesting, the slot power from a few risers were dropping to ~11.4. The connectors were around 11.86-12.04v. One 3060 and the 3070 riser was being powered by SATA to molex adapters so I removed those and recabled everything to run off of 4 pin molex with no more than 2 risers per string. This brought the PCIE slot voltages to the 11.736 to 11.893 range. The rig was more stable but still rebooted after an hour. I’m wondering if this version of risers are just crap for voltage regulation or if I need to up my PSU cable gauge which is currently at 18AWG. Maybe the PSU is getting tired and needs to be replaced?
Ultimately, I ended up paralleling a second power supply to provide power to only the 3070 TI. Since doing this the rig has been running without a hiccup for over 3 days. Any thoughts?