uctz isn't added because it will implemented in the GLSL path and the
SPIR-V path using other pre-existing instructions.
v2: Avoid signed integer overflow for uabs_isub(0, INT_MIN). Noticed by
Caio.
v3: Alternate fix for signed integer overflow for abs_sub(0, INT_MIN).
I tried the previous methon in a small test program with -ftrapv, and it
failed.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com> [v1]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/767>
+unop("uclz", tuint32, """
+int bit;
+for (bit = bit_size - 1; bit >= 0; bit--) {
+ if ((src0 & (1u << bit)) != 0)
+ break;
+}
+dst = (unsigned)(31 - bit);
+""")
+
unop("ifind_msb", tint32, """
dst = -1;
for (int bit = 31; bit >= 0; bit--) {
unop("ifind_msb", tint32, """
dst = -1;
for (int bit = 31; bit >= 0; bit--) {
}
""")
binop("isub", tint, "", "src0 - src1")
}
""")
binop("isub", tint, "", "src0 - src1")
+binop_convert("uabs_isub", tuint, tint, "", """
+ src1 > src0 ? (uint64_t) src1 - (uint64_t) src0
+ : (uint64_t) src0 - (uint64_t) src1
+""")
+binop("uabs_usub", tuint, "", "(src1 > src0) ? (src1 - src0) : (src0 - src1)")
binop("fmul", tfloat, _2src_commutative + associative, """
if (nir_is_rounding_mode_rtz(execution_mode, bit_size)) {
binop("fmul", tfloat, _2src_commutative + associative, """
if (nir_is_rounding_mode_rtz(execution_mode, bit_size)) {
dst = ((uint64_t)src0 & mask) * ((uint64_t)src1 & mask);
""")
dst = ((uint64_t)src0 & mask) * ((uint64_t)src1 & mask);
""")
+# Multiply 32-bits with low 16-bits.
+binop("imul_32x16", tint32, "", "src0 * (int16_t) src1")
+binop("umul_32x16", tuint32, "", "src0 * (uint16_t) src1")
binop("fdiv", tfloat, "", "src0 / src1")
binop("idiv", tint, "", "src1 == 0 ? 0 : (src0 / src1)")
binop("fdiv", tfloat, "", "src0 / src1")
binop("idiv", tint, "", "src1 == 0 ? 0 : (src0 / src1)")