mp3_0 initial sv

diff --git a/media/audio/mp3/mp3_0_apply_window_float_basicsv.s b/media/audio/mp3/mp3_0_apply_window_float_basicsv.s
index f41bbe725864306de68966a11ecb0402462e0660..f248e1b0c4df5643945a435a652e50b687271c46 100644 (file)
--- a/media/audio/mp3/mp3_0_apply_window_float_basicsv.s
+++ b/media/audio/mp3/mp3_0_apply_window_float_basicsv.s
@@ -10,9 +10,19 @@
 .set i, 8
 .set out2, 10
 
+# SV ints, so we don't have to play with the stack
+.set win2, 32
+
+# SV floats
+.set fv0, 32
+.set fv1, 40
+.set fv2, 48
+
 # floats
 .set sum, 0
 .set sum2, 1
+.set tmpsum, 2
+.set tmpsum2, 2
 
        .machine power9
        .abiversion 2
@@ -34,12 +44,27 @@ ff_mpadsp_apply_window_float_sv:
        mulli 0, incr, 31
        add out2, out, 0
 
-       # w2 = window + 31; TODO
+       # sv.setvli 0, MVL=8, VL=8
+       # sv.addi win2, win, 124
+
        lfiwax sum, 0, 9 # zero it
        addi p, buf, 64
-       # SUM8(MACS, sum, w, p); TODO
+
+       # SUM8(MACS, sum, w, p)
+       # sv.lfs/els fv0.v, 256(win)
+       # sv.lfs/els fv1.v, 256(p)
+       # sv.fmuls/mr sum, fv0.v, fv1.v
+
+       lfiwax tmpsum, 0, 9 # zero it
        addi p, buf, 192
-       # SUM8(MLSS, sum, w + 32, p); TODO
+       addi win, win, 128
+       # SUM8(MLSS, sum, w + 32, p)
+       # sv.lfs/els fv0.v, 256(win)
+       # sv.lfs/els fv1.v, 256(p)
+       # sv.fmuls/mr tmpsum, fv0.v, fv1.v
+       fsubs sum, sum, tmpsum
+       subi win, win, 128
+
        stfs sum, 0(out)
        add out, out, incr
        addi win, win, 4
@@ -50,12 +75,36 @@ ff_mpadsp_apply_window_float_sv:
        li i, 4
 .Lloop:
                lfiwax sum, 0, 9 # zero it
+               lfiwax sum2, 0, 9 # zero it
                addi p, buf, 64
                add p, p, i
-               # SUM8P2(sum, MACS, sum2, MLSS, w, w2, p);TODO
+
+               # SUM8P2(sum, MACS, sum2, MLSS, w, w2, p)
+               # sv.lfs/els fv0.v, 256(p)
+               # sv.lfs/els fv1.v, 256(win)
+               # sv.lfs/els fv2.v, 256(win2)
+               # sv.fmuls/mr sum, fv0.v, fv1.v
+               # sv.fmuls/mr sum2, fv0.v, fv2.v
+               fneg sum2, sum2
+
                addi p, buf, 192
                subf p, i, p
-               # SUM8P2(sum, MLSS, sum2, MLSS, w + 32, w2 + 32, p); TODO
+               lfiwax tmpsum, 0, 9 # zero it
+               lfiwax tmpsum2, 0, 9 # zero it
+               addi win, win, 128
+               # sv.addi win2, win2, 128
+
+               # SUM8P2(sum, MLSS, sum2, MLSS, w + 32, w2 + 32, p)
+               # sv.lfs/els fv0.v, 256(p)
+               # sv.lfs/els fv1.v, 256(win)
+               # sv.lfs/els fv2.v, 256(win2)
+               # sv.fmuls/mr tmpsum, fv0.v, fv1.v
+               # sv.fmuls/mr tmpsum2, fv0.v, fv2.v
+               fsubs sum, sum, tmpsum
+               fsubs sum2, sum2, tmpsum2
+
+               subi win, win, 128
+               # sv.subi win2, win2, 128
 
                stfs sum, 0(out)
                add out, out, incr
@@ -64,11 +113,18 @@ ff_mpadsp_apply_window_float_sv:
 
                addi i, i, 4
                addi win, win, 4
-               subi win2, win2, 4
+               # sv.subi win2, win2, 4
        bdnz .Lloop
 
        addi p, buf, 128
-       # SUM8(MLSS, sum, w + 32, p); TODO
+       addi win, win, 128
+       lfiwax sum, 0, 9 # zero it
+       # SUM8(MLSS, sum, w + 32, p)
+       # sv.lfs/els fv0.v, 256(win)
+       # sv.lfs/els fv1.v, 256(p)
+       # sv.fmuls/mr sum, fv0.v, fv1.v
+       fneg sum, sum
+
        stfs sum, 0(out)
 
        blr