- Nov 23, 2001
- 6,712
- 142
- 106
So looking at my Vega Fe and going over the powerplay table the defaults can be seen below:
sclk:
p0 852
p1 991
p2 1138
p3 1269
p4 1348
p5 1440
p6 1528
p7 1600
mclk:
p0 167
p1 500
p2 800
p3 945
SoC clock:
p0 600
p1 720
p2 847
p3 960
p4 1028
p5 1107
Disp clock:
p0 282
p1 515
p2 686
p3 800
p4 900
p5 1029
p6 1108
p7 1200
Has anyone done research into the voltages and state transition behavior for these different clock domains ?
I think many of you already know that there is NO memory voltage for vega, it just decides to transition the mclk if the core voltage is above the needed threshold, thus using the current core voltage.
The behavior for SoC and Disp should be similar, but having such a large number of clock states makes it difficult to know what is going on here.
The Disp clock being tied to the sclk, since they both have 7 states, would seem to make sense.
Having 2 fewer SoC states makes that difficult to figure out.
What do you all think ?
I'd love to get some thoughts from anyone else that has pondered this.
sclk:
p0 852
p1 991
p2 1138
p3 1269
p4 1348
p5 1440
p6 1528
p7 1600
mclk:
p0 167
p1 500
p2 800
p3 945
SoC clock:
p0 600
p1 720
p2 847
p3 960
p4 1028
p5 1107
Disp clock:
p0 282
p1 515
p2 686
p3 800
p4 900
p5 1029
p6 1108
p7 1200
Has anyone done research into the voltages and state transition behavior for these different clock domains ?
I think many of you already know that there is NO memory voltage for vega, it just decides to transition the mclk if the core voltage is above the needed threshold, thus using the current core voltage.
The behavior for SoC and Disp should be similar, but having such a large number of clock states makes it difficult to know what is going on here.
The Disp clock being tied to the sclk, since they both have 7 states, would seem to make sense.
Having 2 fewer SoC states makes that difficult to figure out.
What do you all think ?
I'd love to get some thoughts from anyone else that has pondered this.