Add vpickev/vpickod

README.md

@@ -30,10 +30,6 @@ TODO List:
 
 ### vssrarn.bu.h/hu.w/wu.d
 
-### vpickev.b/h/w/d
-
-### vpickod.b/h/w/d
-
 ### vreplve.b/h/w/d
 
 ### vfrstp.b/h

code/gen_impl.py

@@ -421,6 +421,20 @@
             file=f,
         )
         print(f"}}", file=f)
+    with open(f"vpickev_{width}.h", "w") as f:
+        print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
+        print(
+            f"  dst.{m}[i] = (i < {64 // w}) ? b.{m}[i * 2] : a.{m}[(i - {64 // w}) * 2];",
+            file=f,
+        )
+        print(f"}}", file=f)
+    with open(f"vpickod_{width}.h", "w") as f:
+        print(f"for (int i = 0;i < {128 // w};i++) {{", file=f)
+        print(
+            f"  dst.{m}[i] = (i < {64 // w}) ? b.{m}[i * 2 + 1] : a.{m}[(i - {64 // w}) * 2 + 1];",
+            file=f,
+        )
+        print(f"}}", file=f)
 
 for width in ["s", "d"]:
     m = members_fp[width]

code/gen_tb.py

@@ -77,6 +77,8 @@
     "vpackev": (widths_signed, "v128 a, v128 b"),
     "vpackod": (widths_signed, "v128 a, v128 b"),
     "vpcnt": (widths_signed, "v128 a"),
+    "vpickev": (widths_signed, "v128 a, v128 b"),
+    "vpickod": (widths_signed, "v128 a, v128 b"),
     "vrotr": (widths_signed, "v128 a, v128 b"),
     "vsadd": (widths_all, "v128 a, v128 b"),
     "vssub": (widths_all, "v128 a, v128 b"),

code/vpickev_b.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickev_b(v128 a, v128 b) {
+  v128 dst;
+#include "vpickev_b.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickev_b); }

code/vpickev_b.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 16; i++) {
+  dst.byte[i] = (i < 8) ? b.byte[i * 2] : a.byte[(i - 8) * 2];
+}

code/vpickev_d.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickev_d(v128 a, v128 b) {
+  v128 dst;
+#include "vpickev_d.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickev_d); }

code/vpickev_d.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 2; i++) {
+  dst.dword[i] = (i < 1) ? b.dword[i * 2] : a.dword[(i - 1) * 2];
+}

code/vpickev_h.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickev_h(v128 a, v128 b) {
+  v128 dst;
+#include "vpickev_h.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickev_h); }

code/vpickev_h.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 8; i++) {
+  dst.half[i] = (i < 4) ? b.half[i * 2] : a.half[(i - 4) * 2];
+}

code/vpickev_w.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickev_w(v128 a, v128 b) {
+  v128 dst;
+#include "vpickev_w.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickev_w); }

code/vpickev_w.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 4; i++) {
+  dst.word[i] = (i < 2) ? b.word[i * 2] : a.word[(i - 2) * 2];
+}

code/vpickod_b.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickod_b(v128 a, v128 b) {
+  v128 dst;
+#include "vpickod_b.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickod_b); }

code/vpickod_b.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 16; i++) {
+  dst.byte[i] = (i < 8) ? b.byte[i * 2 + 1] : a.byte[(i - 8) * 2 + 1];
+}

code/vpickod_d.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickod_d(v128 a, v128 b) {
+  v128 dst;
+#include "vpickod_d.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickod_d); }

code/vpickod_d.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 2; i++) {
+  dst.dword[i] = (i < 1) ? b.dword[i * 2 + 1] : a.dword[(i - 1) * 2 + 1];
+}

code/vpickod_h.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickod_h(v128 a, v128 b) {
+  v128 dst;
+#include "vpickod_h.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickod_h); }

code/vpickod_h.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 8; i++) {
+  dst.half[i] = (i < 4) ? b.half[i * 2 + 1] : a.half[(i - 4) * 2 + 1];
+}

code/vpickod_w.cpp

@@ -0,0 +1,9 @@
+#include "common.h"
+
+v128 vpickod_w(v128 a, v128 b) {
+  v128 dst;
+#include "vpickod_w.h"
+  return dst;
+}
+
+void test() { FUZZ2(vpickod_w); }

code/vpickod_w.h

@@ -0,0 +1,3 @@
+for (int i = 0; i < 4; i++) {
+  dst.word[i] = (i < 2) ? b.word[i * 2 + 1] : a.word[(i - 2) * 2 + 1];
+}

docs/lsx/misc.md

@@ -24,3 +24,13 @@
 {{ vpackod('h') }}
 {{ vpackod('w') }}
 {{ vpackod('d') }}
+
+{{ vpickev('b') }}
+{{ vpickev('h') }}
+{{ vpickev('w') }}
+{{ vpickev('d') }}
+
+{{ vpickod('b') }}
+{{ vpickod('h') }}
+{{ vpickod('w') }}
+{{ vpickod('d') }}

main.py

@@ -715,4 +715,22 @@ def vpackod(name):
             intrinsic=f"__m128i __lsx_vpackod_{name} (__m128i a, __m128i b)",
             instr=f"vpackod.{name} vr, vr, vr",
             desc=f"Collect and pack odd-positioned {width}-bit elements in `a` and `b` and store `dst`.",
+        )
+
+    @env.macro
+    def vpickev(name):
+        width = widths[name]
+        return instruction(
+            intrinsic=f"__m128i __lsx_vpickev_{name} (__m128i a, __m128i b)",
+            instr=f"vpickev.{name} vr, vr, vr",
+            desc=f"Pick even-positioned {width}-bit elements in `b` first, then pick even-positioned {width}-bit elements in `a`.",
+        )
+
+    @env.macro
+    def vpickod(name):
+        width = widths[name]
+        return instruction(
+            intrinsic=f"__m128i __lsx_vpickod_{name} (__m128i a, __m128i b)",
+            instr=f"vpickod.{name} vr, vr, vr",
+            desc=f"Pick odd-positioned {width}-bit elements in `b` first, then pick odd-positioned {width}-bit elements in `a`.",
         )