mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-04-22 16:45:31 +00:00
Code Activity

port logarithmic encoding to shr, add review comment on rotate

Browse source 
Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com>
This commit is contained in:
Nikolaj Bjorner 2014-10-02 09:41:53 -07:00
parent 8929c578c1
commit 2c70e8d79f
1 changed files with 30 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

51
src/ast/rewriter/bit_blaster/bit_blaster_tpl_def.h
View file

@ -914,6 +914,7 @@ void bit_blaster_tpl<Cfg>::mk_shl(unsigned sz, expr * const * a_bits, expr * con
out_bits.append(sz, a_bits);
for (unsigned i = 0; i < sz; ++i) {
checkpoint();
expr_ref_vector new_out_bits(m());
unsigned shift_i = 1 << i;
for (unsigned j = 0; j < sz; ++j) {
@ -942,19 +943,21 @@ void bit_blaster_tpl<Cfg>::mk_lshr(unsigned sz, expr * const * a_bits, expr * co
out_bits.push_back(m().mk_false());
}
else {
expr_ref_vector eqs(m());
mk_eqs(sz, b_bits, eqs);
out_bits.resize(sz);
for (unsigned i = 0; i < sz; i++) {
out_bits.append(sz, a_bits);
for (unsigned i = 0; i < sz; ++i) {
checkpoint();
expr_ref out(m());
mk_ite(eqs.get(i), a_bits[sz-1], m().mk_false(), out);
for (unsigned j = 1; j <= i; j++) {
expr_ref_vector new_out_bits(m());
unsigned shift_i = 1 << i;
for (unsigned j = 0; j < sz; ++j) {
expr_ref new_out(m());
mk_ite(eqs.get(i - j), a_bits[sz - j - 1], out, new_out);
out = new_out;
expr* a_j = m().mk_false();
if (shift_i + j < sz) a_j = a_bits[j+shift_i];
mk_ite(b_bits[i], a_j, out_bits[j].get(), new_out);
new_out_bits.push_back(new_out);
}
out_bits.set(sz - i - 1, out);
out_bits.reset();
out_bits.append(new_out_bits);
if (shift_i >= sz) break;
}
}
}
@ -971,20 +974,21 @@ void bit_blaster_tpl<Cfg>::mk_ashr(unsigned sz, expr * const * a_bits, expr * co
out_bits.push_back(a_bits[sz-1]);
}
else {
expr_ref_vector eqs(m());
mk_eqs(sz, b_bits, eqs);
out_bits.resize(sz);
for (unsigned i = 0; i < sz; i++) {
out_bits.append(sz, a_bits);
for (unsigned i = 0; i < sz; ++i) {
checkpoint();
expr_ref out(m());
out = a_bits[sz-1];
for (unsigned j = 1; j <= i; j++) {
expr_ref_vector new_out_bits(m());
unsigned shift_i = 1 << i;
for (unsigned j = 0; j < sz; ++j) {
expr_ref new_out(m());
mk_ite(eqs.get(i - j), a_bits[sz - j - 1], out, new_out);
out = new_out;
expr* a_j = a_bits[sz-1];
if (shift_i + j < sz) a_j = a_bits[j+shift_i];
mk_ite(b_bits[i], a_j, out_bits[j].get(), new_out);
new_out_bits.push_back(new_out);
}
TRACE("bit_blaster", tout << (sz - i - 1) << " :\n" << mk_pp(out, m()) << "\n";);
out_bits.set(sz - i - 1, out);
out_bits.reset();
out_bits.append(new_out_bits);
if (shift_i >= sz) break;
}
}
}
@ -1000,6 +1004,11 @@ void bit_blaster_tpl<Cfg>::mk_ext_rotate_left_right(unsigned sz, expr * const *
mk_rotate_right(sz, a_bits, static_cast<unsigned>(k.get_uint64()), out_bits);
}
else {
//
// Review: a better tuned implementation is possible by using shifts by power of two.
// e.g., looping over the bits of b_bits, then rotate by a power of two depending
// on the bit-position. This would get rid of the mk_urem.
//
expr_ref_vector sz_bits(m());
expr_ref_vector masked_b_bits(m());
expr_ref_vector eqs(m());