conferences/fosdem2024/fosdem2024_ddffirst/maxloc.s

   1 # while (i<n)
   2 setvl 2,0,4,0,1,1            # set MVL=4, VL=MIN(MVL,CTR)
   3 #    while (i<n and a[i]<=m) : i += 1
   4 sv.cmp/ff=gt/m=ge *0,0,*10,4 # truncates VL to min
   5 sv.creqv *16,*16,*16         # set mask on already-tested
   6 setvl 2,0,4,0,1,1            # set MVL=4, VL=MIN(MVL,CTR)
   7 mtcrf 128, 0                 # clear CR0 (in case VL=0?)
   8 #    while (i<n and a[i]>m):
   9 sv.minmax./ff=le/m=ge/mr 4,*10,4,1 # r4 accumulator
  10 crternlogi 0,1,2,127         # test greater/equal or VL=0
  11 sv.crand *19,*16,0           # clear if CR0.eq=0
  12 #      nm = i (count masked bits. could use crweirds here)
  13 sv.svstep/mr/m=so 1,0,6,1    # svstep: get vector dststep
  14 sv.creqv *16,*16,*16         # set mask on already-tested
  15 bc 12,0, -0x40               # CR0 lt bit clear, branch back