Types and utility functions

Integer types sct_int and sct_uint

The bit-accurate integer types which support any number of bits:

• sct_int<N> is signed integer with N bits, where N is zero or positive
• sct_uint<N> is unsigned integer with N bits, where N is zero or positive

SystemC contains sc_int/sc_uint integer types which are limited to 64 bit. Also SystemC contains sc_bigint/sc_biguint which can be more than 64bits, but are slow in simulation. Also all these integer types do not support 0 width. sct_int and sct_uint automatically substitute sc_int/sc_bigint(sc_uint/sc_biguint) depends on bit width. These types are implemented in sct_sel_types.h.

To initialize a variable of sct_uint type with all bits 0 or 1 there are special templates:

• sct_zeros<N> is N-bit zeros literal of sct_uint type,
• sct_ones<N> is N-bit ones literal of sct_uint type.

// Variable declaration with 0/1 initialization
sct_uint<12> a = sct_zeros<12>;
sct_uint<66> b = sct_ones<66>;
auto c  = sct_zeros<90>;

Zero width integer types

Zero width integer types sct_int<0> and sct_uint<0> intended to represent optional variables, signals, ports and record fields. In the generated SV such variables are not declared, assignment to such variable is not generated, using such variable as RValue replaced with 0. The same works for zero width signals, ports and record fields.

template <unsigned N>
struct MyModule : public sc_module {
  sct_uint<N> optVar;
  sc_in<sct_uint<N>> optInPort{"optInPort"};
  sc_signal<sct_uint<N>> optSig{"optSig"};
  struct Rec {
     sct_uint<N> optField;     
  };
}

Utility functions

• sct_log2<X> / sct_floor_log2<X> compile-time floor(log2(X)) value, e.g. sct_log2<1> == 0, sct_log2<2> == 1, sct_log2<3> == 1, sct_log2<4> == 2
• sct_ceil_log2<X> compile-time ceil(log2(X)) value, e.g. sct_ceil_log2<1> == 0, sct_ceil_log2<2> == 1, sct_ceil_log2<3> == 2, sct_ceil_log2<4> == 2
• sct_nbits<X> number of bits needed to store the value X: zero if X is zero; otherwise 1 + floor(log2(X)). E.g. sct_nbits<0> == 0, sct_nbits<1> == 1, sct_nbits<2> == 2, sct_nbits<3> == 2, sct_nbits<4> == 3
• sct_addrbits<N> address/index width for N elements, same as ceil(log2(N)). E.g. sct_addrbits<0> == 0, sct_addrbits<1> == 0, sct_addrbits<2> == 1, sct_addrbits<3> == 2, sct_addrbits<4> == 2
• sct_addrbits1<N> address/index width for N elements, same as sct_addrbits but not less than one. E.g. sct_addrbits1<0> == 1, sct_addrbits1<1> == 1, sct_addrbits1<2> == 1, sct_addrbits1<3> == 2, sct_addrbits1<4> == 2
• sct_is_pow2<X> checks if a number X is power of 2, numbers 0 and 1 considered as power of 2. E.g. sct_is_pow2<0> == 1, sct_is_pow2<1> == 1, sct_is_pow2<8> == 1, sct_is_pow2<9> == 0
• sct_next_pow2<X> returns power of 2 number greater than X. E.g. sct_next_pow2<0> == 1, sct_next_pow2<1> == 2, sct_next_pow2<5> == 8, sct_is_pow2<8> == 16