tfloat32 = "float32"
tint32 = "int32"
tuint32 = "uint32"
+tuint64 = "uint64"
tfloat64 = "float64"
commutative = "commutative "
unop("flog2", tfloat, "log2f(src0)")
unop_convert("f2i", tint32, tfloat32, "src0") # Float-to-integer conversion.
unop_convert("f2u", tuint32, tfloat32, "src0") # Float-to-unsigned conversion
+unop_convert("d2i", tint32, tfloat64, "src0") # Double-to-integer conversion.
+unop_convert("d2u", tuint32, tfloat64, "src0") # Double-to-unsigned conversion.
unop_convert("i2f", tfloat32, tint32, "src0") # Integer-to-float conversion.
+unop_convert("i2d", tfloat64, tint32, "src0") # Integer-to-double conversion.
# Float-to-boolean conversion
unop_convert("f2b", tbool, tfloat32, "src0 != 0.0f")
+unop_convert("d2b", tbool, tfloat64, "src0 != 0.0")
# Boolean-to-float conversion
unop_convert("b2f", tfloat32, tbool, "src0 ? 1.0f : 0.0f")
# Int-to-boolean conversion
unop_convert("i2b", tbool, tint32, "src0 != 0")
unop_convert("b2i", tint32, tbool, "src0 ? 1 : 0") # Boolean-to-int conversion
unop_convert("u2f", tfloat32, tuint32, "src0") # Unsigned-to-float conversion.
+unop_convert("u2d", tfloat64, tuint32, "src0") # Unsigned-to-double conversion.
+# double-to-float conversion
+unop_convert("d2f", tfloat32, tfloat64, "src0") # Single to double precision
+unop_convert("f2d", tfloat64, tfloat32, "src0") # Double to single precision
# Unary floating-point rounding operations.
(src0.w << 24);
""")
+unop_horiz("pack_double_2x32", 1, tuint64, 2, tuint32,
+ "dst.x = src0.x | ((uint64_t)src0.y << 32);")
+
+unop_horiz("unpack_double_2x32", 2, tuint32, 1, tuint64,
+ "dst.x = src0.x; dst.y = src0.x >> 32;")
+
# Lowered floating point unpacking operations.
unop_horiz("unpack_half_2x16_split_y", 1, tfloat32, 1, tuint32,
"unpack_half_1x16((uint16_t)(src0.x >> 16))")
+unop_convert("unpack_double_2x32_split_x", tuint32, tuint64, "src0")
+unop_convert("unpack_double_2x32_split_y", tuint32, tuint64, "src0 >> 32")
# Bit operations, part of ARB_gpu_shader5.
binop_convert("usub_borrow", tuint, tuint, "", "src0 < src1")
-binop("fmod", tfloat, "", "src0 - src1 * floorf(src0 / src1)")
binop("umod", tuint, "", "src1 == 0 ? 0 : src0 % src1")
+# For signed integers, there are several different possible definitions of
+# "modulus" or "remainder". We follow the conventions used by LLVM and
+# SPIR-V. The irem opcode implements the standard C/C++ signed "%"
+# operation while the imod opcode implements the more mathematical
+# "modulus" operation. For details on the difference, see
+#
+# http://mathforum.org/library/drmath/view/52343.html
+
+binop("irem", tint, "", "src1 == 0 ? 0 : src0 % src1")
+binop("imod", tint, "",
+ "src1 == 0 ? 0 : ((src0 % src1 == 0 || (src0 >= 0) == (src1 >= 0)) ?"
+ " src0 % src1 : src0 % src1 + src1)")
+binop("fmod", tfloat, "", "src0 - src1 * floorf(src0 / src1)")
+binop("frem", tfloat, "", "src0 - src1 * truncf(src0 / src1)")
+
#
# Comparisons
#
binop_horiz("pack_half_2x16_split", 1, tuint32, 1, tfloat32, 1, tfloat32,
"pack_half_1x16(src0.x) | (pack_half_1x16(src1.x) << 16)")
+binop_convert("pack_double_2x32_split", tuint64, tuint32, "",
+ "src0 | ((uint64_t)src1 << 32)")
+
# bfm implements the behavior of the first operation of the SM5 "bfi" assembly
# and that of the "bfi1" i965 instruction. That is, it has undefined behavior
# if either of its arguments are 32.