riscv64/itx: Add 4x16 8bpc eob test

Kendryte K230                                        Before         After

inv_txfm_add_4x16_adst_adst_0_8bpc_rvv:           429.9 (7.45x)  381.3 (8.58x)
inv_txfm_add_4x16_adst_adst_1_8bpc_rvv:           430.0 (7.45x)  381.3 (8.57x)
inv_txfm_add_4x16_adst_dct_0_8bpc_rvv:            381.0 (8.01x)  332.5 (9.19x)
inv_txfm_add_4x16_adst_dct_1_8bpc_rvv:            381.0 (8.00x)  332.5 (9.19x)
inv_txfm_add_4x16_adst_flipadst_0_8bpc_rvv:       432.8 (7.42x)  384.5 (8.52x)
inv_txfm_add_4x16_adst_flipadst_1_8bpc_rvv:       432.8 (7.42x)  384.4 (8.52x)
inv_txfm_add_4x16_adst_identity_0_8bpc_rvv:       304.6 (7.32x)  249.8 (9.18x)
inv_txfm_add_4x16_adst_identity_1_8bpc_rvv:       304.5 (7.32x)  249.8 (9.18x)
inv_txfm_add_4x16_dct_adst_0_8bpc_rvv:            407.2 (7.68x)  371.4 (8.57x)
inv_txfm_add_4x16_dct_adst_1_8bpc_rvv:            407.1 (7.68x)  371.5 (8.58x)
inv_txfm_add_4x16_dct_dct_0_8bpc_rvv:             357.9 (1.27x)  323.1 (1.41x)
inv_txfm_add_4x16_dct_dct_1_8bpc_rvv:             357.9 (8.29x)  322.9 (9.16x)
inv_txfm_add_4x16_dct_flipadst_0_8bpc_rvv:        410.0 (7.62x)  376.6 (8.45x)
inv_txfm_add_4x16_dct_flipadst_1_8bpc_rvv:        410.0 (7.62x)  376.5 (8.47x)
inv_txfm_add_4x16_dct_identity_0_8bpc_rvv:        275.2 (7.79x)  240.5 (9.21x)
inv_txfm_add_4x16_dct_identity_1_8bpc_rvv:        275.3 (7.78x)  240.6 (9.19x)
inv_txfm_add_4x16_flipadst_adst_0_8bpc_rvv:       430.5 (7.51x)  382.6 (8.60x)
inv_txfm_add_4x16_flipadst_adst_1_8bpc_rvv:       430.1 (7.52x)  382.8 (8.60x)
inv_txfm_add_4x16_flipadst_dct_0_8bpc_rvv:        381.1 (8.09x)  333.8 (9.21x)
inv_txfm_add_4x16_flipadst_dct_1_8bpc_rvv:        381.0 (8.08x)  333.7 (9.21x)
inv_txfm_add_4x16_flipadst_flipadst_0_8bpc_rvv:   433.0 (7.48x)  385.7 (8.55x)
inv_txfm_add_4x16_flipadst_flipadst_1_8bpc_rvv:   433.0 (7.48x)  385.7 (8.55x)
inv_txfm_add_4x16_flipadst_identity_0_8bpc_rvv:   298.6 (7.57x)  250.8 (9.28x)
inv_txfm_add_4x16_flipadst_identity_1_8bpc_rvv:   298.6 (7.57x)  250.9 (9.27x)
inv_txfm_add_4x16_identity_adst_0_8bpc_rvv:       361.5 (7.93x)  347.3 (8.35x)
inv_txfm_add_4x16_identity_adst_1_8bpc_rvv:       361.4 (7.93x)  347.4 (8.35x)
inv_txfm_add_4x16_identity_dct_0_8bpc_rvv:        310.9 (8.69x)  297.8 (9.02x)
inv_txfm_add_4x16_identity_dct_1_8bpc_rvv:        311.0 (8.69x)  297.8 (9.02x)
inv_txfm_add_4x16_identity_flipadst_0_8bpc_rvv:   364.1 (7.88x)  350.5 (8.29x)
inv_txfm_add_4x16_identity_flipadst_1_8bpc_rvv:   364.2 (7.88x)  350.4 (8.31x)
inv_txfm_add_4x16_identity_identity_0_8bpc_rvv:   229.7 (8.22x)  211.4 (9.11x)
inv_txfm_add_4x16_identity_identity_1_8bpc_rvv:   229.7 (8.21x)  211.2 (9.12x)

1 files changed, 30 insertions, 17 deletions

diff --git a/src/riscv/64/itx.S b/src/riscv/64/itx.S
index 4ba4c1a..5ed3c90 100644
--- a/src/riscv/64/itx.S
+++ b/src/riscv/64/itx.S
@@ -1567,6 +1567,8 @@ function inv_txfm_\variant\()add_4x16_rvv, export=1, ext=v
 
   vsetivli zero, 8, e16, m1, ta, ma
 
+  blt a3, a6, 1f
+
   addi t0, a2, 16
   vle16.v v0, (t0)
   addi t0, t0, 32
@@ -1595,6 +1597,14 @@ function inv_txfm_\variant\()add_4x16_rvv, export=1, ext=v
   vssra.vi v7, v3, 1
 .endif
 
+  j 2f
+
+1:
+.irp i, 4, 5, 6, 7
+  vmv.v.x v\i, zero
+.endr
+
+2:
   vle16.v v0, (a2)
   addi t0, a2, 32
   vle16.v v1, (t0)
@@ -1805,7 +1815,7 @@ endfunc
 def_fn_416_base identity_
 def_fn_416_base
 
-.macro def_fn_416 w, h, txfm1, txfm2
+.macro def_fn_416 w, h, txfm1, txfm2, eob_half
 function inv_txfm_add_\txfm1\()_\txfm2\()_\w\()x\h\()_8bpc_rvv, export=1
 .if \w == 4 && (\txfm1 == adst || \txfm1 == flipadst)
   la a4, inv_\txfm1\()_e16_x\w\()w_rvv
@@ -1817,6 +1827,9 @@ function inv_txfm_add_\txfm1\()_\txfm2\()_\w\()x\h\()_8bpc_rvv, export=1
 .else
   la a5, inv_\txfm2\()_e16_x\h\()_rvv
 .endif
+.if \w == 4
+  li a6, \eob_half
+.endif
 .ifc \txfm1, identity
   j inv_txfm_identity_add_\w\()x\h\()_rvv
 .else
@@ -1826,22 +1839,22 @@ endfunc
 .endm
 
 .macro def_fns_416 w, h
-def_fn_416 \w, \h, dct, dct
-def_fn_416 \w, \h, identity, identity
-def_fn_416 \w, \h, dct, adst
-def_fn_416 \w, \h, dct, flipadst
-def_fn_416 \w, \h, dct, identity
-def_fn_416 \w, \h, adst, dct
-def_fn_416 \w, \h, adst, adst
-def_fn_416 \w, \h, adst, flipadst
-def_fn_416 \w, \h, flipadst, dct
-def_fn_416 \w, \h, flipadst, adst
-def_fn_416 \w, \h, flipadst, flipadst
-def_fn_416 \w, \h, identity, dct
-def_fn_416 \w, \h, adst, identity
-def_fn_416 \w, \h, flipadst, identity
-def_fn_416 \w, \h, identity, adst
-def_fn_416 \w, \h, identity, flipadst
+def_fn_416 \w, \h, dct, dct, 29
+def_fn_416 \w, \h, identity, identity, 29
+def_fn_416 \w, \h, dct, adst, 29
+def_fn_416 \w, \h, dct, flipadst, 29
+def_fn_416 \w, \h, dct, identity, 8
+def_fn_416 \w, \h, adst, dct, 29
+def_fn_416 \w, \h, adst, adst, 29
+def_fn_416 \w, \h, adst, flipadst, 29
+def_fn_416 \w, \h, flipadst, dct, 29
+def_fn_416 \w, \h, flipadst, adst, 29
+def_fn_416 \w, \h, flipadst, flipadst, 29
+def_fn_416 \w, \h, identity, dct, 32
+def_fn_416 \w, \h, adst, identity, 8
+def_fn_416 \w, \h, flipadst, identity, 8
+def_fn_416 \w, \h, identity, adst, 32
+def_fn_416 \w, \h, identity, flipadst, 32
 .endm
 
 def_fns_416 4, 16