port royale custom 1440p, async on, reflection filtering medium (non demo), reflection mode RT, all off next 5 options, max reflection samples 1, bloom on, filtering mode AF, max AF 16, max TAA 16, TAA on for tests, vsync and fixed fps off
stock - profile 1
320w
1905 but drops to 1845 at end 100% power limit isnt enough
score 11331 52.46fps
custom - profile 2 - be profile for games that require max performance mode
142w - power limit 80%
1500 (1515 below 35C) 0.731v - old 1500 0.718v
score 8613 39.88fps
custom - profile 3 - probably be day to day profile.
240w - 255w at end - power limit 80%
1830 (1845 at end) 0.850v - old 1860 0.850v
score 11118 51.48fps
custom - profile 4 - this profile maintains same clocks as stock configuration, but unlike stock can hold the clock at 100% power limit, possibly be profile for new games where profile 3 cannot reach 60fps, although it only gains 1fps.
280w - at end 290w - power limit 100%
1905 0.882v
score 11454 53.03 fps
timespy custom, 1440p, graphics test 1,2 no cpu test, borderless full screen windowed, demo vram, auto, async on, max af 16, max tesselation 32, tesselation scale 1, vsync triple buffer fixed fps all off
profile 1 80% power limit
250-255w (clocks, averaging about 1780mhz dropping to 1690)
score 16155, 105.91fps, 92.15fps
profile 1 100% power limit
315-320w (clock speed avaraged 1845, dropping to 1770)
score 17068, 111.81fps, 97.42fps
profile 2
135-180w
score 13300, 83.64fps, 78.77fps
profile 3
240-260w (hit limit) clock bottomed out at 1785
score 16886, 109.82fps, 97.00fps
profile 4
270w-318w didnt hit limit
score 17642, 113.12fps, 100.66fps
next is timespy capped 60fps tests, settings otherwise same as above, to mimic my typical gaming config.
This is to see the point of using lower profiles when card is not under full load. Also useful for testing curve stability as full load tests keeps clocks at top end of curve only. These peak load figures might also help people who want to use lower end psu's.
profile 1 100% power limit
106w-314w, score 9793
profile 2
105w-153w, score 9760 (in one 1-2 second part of run fps did dip to 56)
profile 3
105w-230w, score 9789
profile 4
105w-240w, score 9800
all tests gysnc disabled. clock multiplier left at 0, voltage left at 0 on slider as well.
conclusion is only time will tell if profiles 2-4 are stable given profile 5 cannot run timespy reliably. profiles 2-4 had at least 10 runs of 10 loops each of timespy and port royale, but I only checked osd data on first run of each, rest were unattended just to see if would complete.
profile 3 isnt quite at profile 1 performance but runs at 80% limit
profile 4 exceeds profile 1 performance with same limit but average power usage lower.
If playing games at 60fps that dont fully load hardware, profiles 3 and 4 are very close, profile 2 peaks much lower but average power consumption is similar, stock remains inefficient.