Add vrotri/vsrari/vsrlri

README.md

@@ -132,12 +132,6 @@ TODO List:
 
 ### vftintrneh.l.s
 
-### vrotri.b/h/w/d
-
-### vsrlri.b/h/w/d
-
-### vsrari.b/h/w/d
-
 ### vpickve2gr.b/h/w/d/bu/hu/wu/du
 
 ### vsllwil.h.b/w.h/d.w

code/gen_impl.py

@@ -401,6 +401,13 @@
             file=f,
         )
         print(f"}}", file=f)
+    with open(f"vrotri_{width}.h", "w") as f:
+        print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
+        print(
+            f"  dst.{m}[i] = (a.{m}[i] >> imm) | (a.{m}[i] << ({w} - imm));",
+            file=f,
+        )
+        print(f"}}", file=f)
     with open(f"vsrlr_{width}.h", "w") as f:
         print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
         print(f"  if ((b.{m}[i] & 0x{w-1:x}) == 0) {{", file=f)
@@ -415,6 +422,20 @@
         )
         print(f"  }}", file=f)
         print(f"}}", file=f)
+    with open(f"vsrlri_{width}.h", "w") as f:
+        print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
+        print(f"  if (imm == 0) {{", file=f)
+        print(
+            f"    dst.{m}[i] = a.{m}[i];",
+            file=f,
+        )
+        print(f"  }} else {{", file=f)
+        print(
+            f"    dst.{m}[i] = (a.{m}[i] >> imm) + ((a.{m}[i] >> (imm - 1)) & 0x1);",
+            file=f,
+        )
+        print(f"  }}", file=f)
+        print(f"}}", file=f)
     with open(f"vsrar_{width}.h", "w") as f:
         print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
         print(f"  if ((b.{m}[i] & 0x{w-1:x}) == 0) {{", file=f)
@@ -429,6 +450,20 @@
         )
         print(f"  }}", file=f)
         print(f"}}", file=f)
+    with open(f"vsrari_{width}.h", "w") as f:
+        print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
+        print(f"  if (imm == 0) {{", file=f)
+        print(
+            f"    dst.{m}[i] = a.{m}[i];",
+            file=f,
+        )
+        print(f"  }} else {{", file=f)
+        print(
+            f"    dst.{m}[i] = ((s{w})a.{m}[i] >> imm) + (((s{w})a.{m}[i] >> (imm - 1)) & 0x1);",
+            file=f,
+        )
+        print(f"  }}", file=f)
+        print(f"}}", file=f)
     with open(f"vpackev_{width}.h", "w") as f:
         print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
         print(

code/gen_tb.py

@@ -86,6 +86,7 @@
     "vpickev": (widths_signed, "v128 a, v128 b"),
     "vpickod": (widths_signed, "v128 a, v128 b"),
     "vrotr": (widths_signed, "v128 a, v128 b"),
+    "vrotri": (widths_signed, "v128 a, int imm", [0, 7]),
     "vreplve": (widths_signed, "v128 a, int idx", [0, 1]),
     "vreplvei": (widths_signed, "v128 a, int idx", [0, 1]),
     "vreplgr2vr": (widths_signed, "int val", [0, 1, 256]),
@@ -103,8 +104,10 @@
     "vslei": (widths_all, "v128 a, int imm", [0, 15]),
     "vsrl": (widths_signed, "v128 a, v128 b"),
     "vsrlr": (widths_signed, "v128 a, v128 b"),
+    "vsrlri": (widths_signed, "v128 a, int imm", [0, 7]),
     "vsra": (widths_signed, "v128 a, v128 b"),
     "vsrar": (widths_signed, "v128 a, v128 b"),
+    "vsrari": (widths_signed, "v128 a, int imm", [0, 7]),
     "vsub": (widths_signed, "v128 a, v128 b"),
     "vsubwev": (widths_vsubw, "v128 a, v128 b"),
     "vsubwod": (widths_vsubw, "v128 a, v128 b"),

code/vrotri_b.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vrotri_b(v128 a, int imm) {
+  v128 dst;
+#include "vrotri_b.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vrotri_b, 0);
+  FUZZ1(vrotri_b, 7);
+}

code/vrotri_b.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 16; i++) {
+  dst.byte[i] = (a.byte[i] >> imm) | (a.byte[i] << (8 - imm));
+}

code/vrotri_d.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vrotri_d(v128 a, int imm) {
+  v128 dst;
+#include "vrotri_d.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vrotri_d, 0);
+  FUZZ1(vrotri_d, 7);
+}

code/vrotri_d.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 2; i++) {
+  dst.dword[i] = (a.dword[i] >> imm) | (a.dword[i] << (64 - imm));
+}

code/vrotri_h.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vrotri_h(v128 a, int imm) {
+  v128 dst;
+#include "vrotri_h.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vrotri_h, 0);
+  FUZZ1(vrotri_h, 7);
+}

code/vrotri_h.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 8; i++) {
+  dst.half[i] = (a.half[i] >> imm) | (a.half[i] << (16 - imm));
+}

code/vrotri_w.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vrotri_w(v128 a, int imm) {
+  v128 dst;
+#include "vrotri_w.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vrotri_w, 0);
+  FUZZ1(vrotri_w, 7);
+}

code/vrotri_w.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 4; i++) {
+  dst.word[i] = (a.word[i] >> imm) | (a.word[i] << (32 - imm));
+}

code/vsrari_b.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrari_b(v128 a, int imm) {
+  v128 dst;
+#include "vsrari_b.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrari_b, 0);
+  FUZZ1(vsrari_b, 7);
+}

code/vsrari_b.h

@@ -0,0 +1,7 @@
+for (int i = 0; i < 16; i++) {
+  if (imm == 0) {
+    dst.byte[i] = a.byte[i];
+  } else {
+    dst.byte[i] = ((s8)a.byte[i] >> imm) + (((s8)a.byte[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrari_d.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrari_d(v128 a, int imm) {
+  v128 dst;
+#include "vsrari_d.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrari_d, 0);
+  FUZZ1(vsrari_d, 7);
+}

code/vsrari_d.h

@@ -0,0 +1,8 @@
+for (int i = 0; i < 2; i++) {
+  if (imm == 0) {
+    dst.dword[i] = a.dword[i];
+  } else {
+    dst.dword[i] =
+        ((s64)a.dword[i] >> imm) + (((s64)a.dword[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrari_h.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrari_h(v128 a, int imm) {
+  v128 dst;
+#include "vsrari_h.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrari_h, 0);
+  FUZZ1(vsrari_h, 7);
+}

code/vsrari_h.h

@@ -0,0 +1,8 @@
+for (int i = 0; i < 8; i++) {
+  if (imm == 0) {
+    dst.half[i] = a.half[i];
+  } else {
+    dst.half[i] =
+        ((s16)a.half[i] >> imm) + (((s16)a.half[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrari_w.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrari_w(v128 a, int imm) {
+  v128 dst;
+#include "vsrari_w.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrari_w, 0);
+  FUZZ1(vsrari_w, 7);
+}

code/vsrari_w.h

@@ -0,0 +1,8 @@
+for (int i = 0; i < 4; i++) {
+  if (imm == 0) {
+    dst.word[i] = a.word[i];
+  } else {
+    dst.word[i] =
+        ((s32)a.word[i] >> imm) + (((s32)a.word[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrlri_b.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrlri_b(v128 a, int imm) {
+  v128 dst;
+#include "vsrlri_b.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrlri_b, 0);
+  FUZZ1(vsrlri_b, 7);
+}

code/vsrlri_b.h

@@ -0,0 +1,7 @@
+for (int i = 0; i < 16; i++) {
+  if (imm == 0) {
+    dst.byte[i] = a.byte[i];
+  } else {
+    dst.byte[i] = (a.byte[i] >> imm) + ((a.byte[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrlri_d.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrlri_d(v128 a, int imm) {
+  v128 dst;
+#include "vsrlri_d.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrlri_d, 0);
+  FUZZ1(vsrlri_d, 7);
+}

code/vsrlri_d.h

@@ -0,0 +1,7 @@
+for (int i = 0; i < 2; i++) {
+  if (imm == 0) {
+    dst.dword[i] = a.dword[i];
+  } else {
+    dst.dword[i] = (a.dword[i] >> imm) + ((a.dword[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrlri_h.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrlri_h(v128 a, int imm) {
+  v128 dst;
+#include "vsrlri_h.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrlri_h, 0);
+  FUZZ1(vsrlri_h, 7);
+}

code/vsrlri_h.h

@@ -0,0 +1,7 @@
+for (int i = 0; i < 8; i++) {
+  if (imm == 0) {
+    dst.half[i] = a.half[i];
+  } else {
+    dst.half[i] = (a.half[i] >> imm) + ((a.half[i] >> (imm - 1)) & 0x1);
+  }
+}

code/vsrlri_w.cpp

@@ -0,0 +1,12 @@
+#include "common.h"
+
+v128 vsrlri_w(v128 a, int imm) {
+  v128 dst;
+#include "vsrlri_w.h"
+  return dst;
+}
+
+void test() {
+  FUZZ1(vsrlri_w, 0);
+  FUZZ1(vsrlri_w, 7);
+}

code/vsrlri_w.h

@@ -0,0 +1,7 @@
+for (int i = 0; i < 4; i++) {
+  if (imm == 0) {
+    dst.word[i] = a.word[i];
+  } else {
+    dst.word[i] = (a.word[i] >> imm) + ((a.word[i] >> (imm - 1)) & 0x1);
+  }
+}

docs/lsx/shift.md

@@ -57,6 +57,11 @@ Compute 128-bit `a` shifted right by `imm * 8` bits.
 {{ vsrlr('w') }}
 {{ vsrlr('d') }}
 
+{{ vsrlri('b') }}
+{{ vsrlri('h') }}
+{{ vsrlri('w') }}
+{{ vsrlri('d') }}
+
 {{ vsra('b') }}
 {{ vsra('h') }}
 {{ vsra('w') }}
@@ -67,7 +72,17 @@ Compute 128-bit `a` shifted right by `imm * 8` bits.
 {{ vsrar('w') }}
 {{ vsrar('d') }}
 
+{{ vsrari('b') }}
+{{ vsrari('h') }}
+{{ vsrari('w') }}
+{{ vsrari('d') }}
+
 {{ vrotr('b') }}
 {{ vrotr('h') }}
 {{ vrotr('w') }}
 {{ vrotr('d') }}
+
+{{ vrotri('b') }}
+{{ vrotri('h') }}
+{{ vrotri('w') }}
+{{ vrotri('d') }}