Initial SVP64 attempt to vpx_get4x4sse_cs_svp64_real()

diff --git a/media/video/libvpx/vpx_get4x4sse_cs_svp64_real.s b/media/video/libvpx/vpx_get4x4sse_cs_svp64_real.s
index 1d7c6e5d4ac570511f45f5b3cef3e337dcb7459f..cac95eee7f4f5e4236c038c833283ba02a95204d 100644 (file)
--- a/media/video/libvpx/vpx_get4x4sse_cs_svp64_real.s
+++ b/media/video/libvpx/vpx_get4x4sse_cs_svp64_real.s
@@ -1,3 +1,15 @@
+.set src_ptr, 3
+.set src_stride, 4
+.set ref_ptr, 5
+.set ref_stride, 6
+.set sum, 7
+.set src, 10
+.set ref, 36
+.set diff, 52
+.set prod, 68
+.set ctr, 9
+
+       .machine libresoc
        .file   "vpx_get4x4sse_cs_svp64_real.c"
        .abiversion 2
        .section        ".text"
@@ -7,29 +19,36 @@
 vpx_get4x4sse_cs_svp64_real:
 .LFB0:
        .cfi_startproc
-       addi 5,5,-1
-       addi 3,3,3
-       li 12,4
-       li 8,0
-.L2:
-       addi 7,3,-4
-       mr 11,5
-       subf 9,7,3
-       mtctr 9
-.L3:
-       lbzu 9,1(7)
-       lbzu 10,1(11)
-       subf 9,10,9
-       mullw 9,9,9
-       add 9,9,8
-       extsw 8,9
-       bdnz .L3
-       addi 9,12,-1
-       add 5,5,6
-       add 3,3,4
-       rldicl. 12,9,0,32
-       bne 0,.L2
-       rldicl 3,8,0,32
+       # Set sum to zero
+       li sum, 0                               # Set sum to zero
+       li ctr, 4                               # Need 4 iterations of 4 elements
+       mtctr ctr                               # Set counter special register
+       # Load 16 elements from src_ptr and ref_ptr, at groups of 4 with stride
+       setvl   0,0,4,0,1,1                     # Set VL to 4 elements
+       sv.lha  *src, 0(src_ptr)                # Load 4 ints from (src_ptr)
+       sv.lha  *ref, 0(ref_ptr)                # Load 4 ints from (ref_ptr)
+       add     src_ptr, src_ptr, src_stride    # Advance src_ptr by src_stride
+       add     ref_ptr, ref_ptr, ref_stride    # Advance ref_ptr by ref_stride
+       sv.lha  *{src + 4}, 0(src_ptr)
+       sv.lha  *{ref + 4}, 0(ref_ptr)
+       add     src_ptr, src_ptr, src_stride
+       add     ref_ptr, ref_ptr, ref_stride
+       sv.lha  *{src + 8}, 0(src_ptr)
+       sv.lha  *{ref + 8}, 0(ref_ptr)
+       add     src_ptr, src_ptr, src_stride
+       add     ref_ptr, ref_ptr, ref_stride
+       sv.lha  *{src + 12}, 0(src_ptr)
+       sv.lha  *{ref + 12}, 0(ref_ptr)
+
+       # now our values are in consecutive registers and we can set VL to 16 elements
+       setvl 0,0,16,0,1,1
+       # equivalent to: for (i = 0; i < 16; i++) diff[i] = src[i] - ref[i];
+       sv.sub          *diff, *src, *ref
+       # equivalent to: for (i = 0; i < 16; i++) prod[i] = diff[i] * diff[i];
+       sv.mulld        *prod, *diff, *diff
+       # equivalent to: for (i = 0; i < 32; i++) sum += prod[i];
+       sv.add/mr       sum, *prod, sum
+       mr 3, sum                               # Set r3 to sum
        blr
        .long 0
        .byte 0,0,0,0,0,0,0,0