Originally Posted by gupsterg @TwoBeers
, too many members with 1700 disable it to gain "realistic" temps, so I thought I'd mention it
1st R7 1700 needed +137mV for 3.8GHz, stability tested back to back for ~36hrs+. 3.9GHz was only tested for 10 loops x264, +181mV with LLC3 as info was not out on what variation LLC did as overshoot, this was ~1.465V on ProbIt point.
2nd R7 1700 needed +162mV for 3.8GHz, again back to back high hours stress testing. 3.9GHz it needs +250mV with LLC: [Auto], ~1.475V on ProbeIt, x264 20 loops.
Both did 4.0GHz bench stable with 3.9GHz voltage, not stress tested.
1st only did 2933MHz with stock VBOOT/VDIMM for RAM (1.35V), SOC: 0.900V in UEFI. 2nd 3200MHz with stock VBOOT/VDIMM for RAM (1.35V), SOC: 0.975V in UEFI after lengthy testing.
At stock 1st loaded with x264 ProbeIt point ~1.089V, 2nd ~1.119V, SOC also differed, 1st ~0.838V and 2nd ~0.893V on ProbeIt points. These stock voltage differences translated into similar voltage requirement for OC. etc. Idle differed for both, 2nd is higher, will look at my notepad (yeah paper & pen tools
3.7GHz ACB is a sweet spot IMO, I need only +50mV for 48 loops x264. Will be testing further. I was toying with getting a 3rd, but after seeing a lot of the OC result data I think I may pass on it.
Nice to read your enjoying your exploits