uniform signed/unsigned int8/16

builtins.cpp

@@ -488,12 +488,21 @@ lSetInternalFunctions(llvm::Module *module) {
         "__num_cores",
         "__packed_load_active",
         "__packed_store_active",
+        "__padds_i8",
+        "__padds_i16",
+        "__vpadds_i8",
+        "__paddus_i8",
+        "__paddus_i16",
         "__popcnt_int32",
         "__popcnt_int64",
         "__prefetch_read_uniform_1",
         "__prefetch_read_uniform_2",
         "__prefetch_read_uniform_3",
         "__prefetch_read_uniform_nt",
+        "__psubs_i8",
+        "__psubs_i16",
+        "__psubus_i8",
+        "__psubus_i16",
         "__rcp_uniform_float",
         "__rcp_varying_float",
         "__rdrand_i16",

builtins/target-sse2-common.ll

@@ -35,6 +35,64 @@ define_shuffles()
 aossoa()
 rdrand_decls()
 
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;; saturation arithmetic
+declare <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8>, <16 x i8>) nounwind readnone
+define <16 x i8> @__vpadds_i8(<16 x i8> %a0, <16 x i8> %a1) {
+  ; CHECK: vpaddsb
+  %res = call <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8> %a0, <16 x i8> %a1) ; <<16 x i8>> [#uses=1]
+  ret <16 x i8> %res
+}
+
+
+;;declare <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8>, <16 x i8>) nounwind readnone
+define i8 @__padds_i8(i8 %a0, i8 %a1) {
+  sse_binary_scalar(ret, 16, i8, @llvm.x86.sse2.padds.b, %a0, %a1)
+  ret i8 %ret
+}
+
+declare <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16>, <8 x i16>) nounwind readnone
+define i16 @__padds_i16(i16 %a0, i16 %a1) {
+  sse_binary_scalar(ret, 8, i16, @llvm.x86.sse2.padds.w, %a0, %a1)
+  ret i16 %ret
+}
+
+declare <16 x i8> @llvm.x86.sse2.paddus.b(<16 x i8>, <16 x i8>) nounwind readnone
+define i8 @__paddus_i8(i8 %a0, i8 %a1) {
+  sse_binary_scalar(ret, 16, i8, @llvm.x86.sse2.paddus.b, %a0, %a1)
+  ret i8 %ret
+}
+
+declare <8 x i16> @llvm.x86.sse2.paddus.w(<8 x i16>, <8 x i16>) nounwind readnone
+define i16 @__paddus_i16(i16 %a0, i16 %a1) {
+  sse_binary_scalar(ret, 8, i16, @llvm.x86.sse2.paddus.w, %a0, %a1)
+  ret i16 %ret
+}
+
+declare <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8>, <16 x i8>) nounwind readnone
+define i8 @__psubs_i8(i8 %a0, i8 %a1) {
+  sse_binary_scalar(ret, 16, i8, @llvm.x86.sse2.psubs.b, %a0, %a1)
+  ret i8 %ret
+}
+
+declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone
+define i16 @__psubs_i16(i16 %a0, i16 %a1) {
+  sse_binary_scalar(ret, 8, i16, @llvm.x86.sse2.psubs.w, %a0, %a1)
+  ret i16 %ret
+}
+
+declare <16 x i8> @llvm.x86.sse2.psubus.b(<16 x i8>, <16 x i8>) nounwind readnone
+define i8 @__psubus_i8(i8 %a0, i8 %a1) {
+  sse_binary_scalar(ret, 16, i8, @llvm.x86.sse2.psubus.b, %a0, %a1)
+  ret i8 %ret
+}
+
+declare <8 x i16> @llvm.x86.sse2.psubus.w(<8 x i16>, <8 x i16>) nounwind readnone
+define i16 @__psubus_i16(i16 %a0, i16 %a1) {
+  sse_binary_scalar(ret, 8, i16, @llvm.x86.sse2.psubus.w, %a0, %a1)
+  ret i16 %ret
+}
+
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;; rcp
 

stdlib.ispc

@@ -4257,6 +4257,41 @@ static inline void fastmath() {
     __fastmath();
 }
 
+///////////////////////////////////////////////////////////////////////////
+// saturation arithmetic
+
+static inline uniform int8 padds(uniform int8 a, uniform int8 b) {
+    return __padds_i8(a, b);
+}
+
+static inline uniform int16 padds(uniform int16 a, uniform int16 b) {
+    return __padds_i16(a, b);
+}
+
+static inline uniform unsigned int8 paddus(uniform unsigned int8 a, uniform unsigned int8 b) {
+    return __paddus_i8(a, b);
+}
+
+static inline uniform unsigned int16 paddus(uniform unsigned int16 a, unsigned uniform int16 b) {
+    return __paddus_i16(a, b);
+}
+
+static inline uniform int8 psubs(uniform int8 a, uniform int8 b) {
+    return __psubs_i8(a, b);
+}
+
+static inline uniform int16 psubs(uniform int16 a, uniform int16 b) {
+    return __psubs_i16(a, b);
+}
+
+static inline uniform unsigned int8 psubus(uniform unsigned int8 a, uniform unsigned int8 b) {
+    return __psubus_i8(a, b);
+}
+
+static inline uniform unsigned int16 psubus(uniform unsigned int16 a, unsigned uniform int16 b) {
+    return __psubus_i16(a, b);
+}
+
 ///////////////////////////////////////////////////////////////////////////
 // rdrand