gaspersic/triton
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[CODEGEN] Fixed bug in atomic_add

Browse Source
This commit is contained in:
Philippe Tillet
2020-11-19 18:19:55 -05:00
parent 7d095ec686
commit baa858aa74
5 changed files with 37 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
lib/codegen/analysis/align.cc
View File

@@ -312,6 +312,7 @@ std::vector<unsigned> align::populate_max_contiguous_gep(ir::getelementptr_inst*
if(rhs_cst_info[d].num_cst)
rvalue = lhs_max_contiguous[d];
result[d] = std::max(lvalue, rvalue);
// std::cout << "max contiguous: " << x->get_name() << " " << d << " " << result[d] << std::endl;
}
return add_to_cache(x, result, max_contiguous_);
}
@@ -416,8 +417,10 @@ std::vector<unsigned> align::populate_starting_multiple_gep(ir::getelementptr_in
auto lhs = populate_starting_multiple(x->get_operand(0));
auto rhs = populate_starting_multiple(x->get_operand(1));
std::vector<unsigned> result(lhs.size(), 1);
for(size_t d = 0; d < lhs.size(); d++)
for(size_t d = 0; d < lhs.size(); d++){
result[d] = gcd(lhs[d], rhs[d]);
// std::cout << "starting multiple: " << x->get_name() << " " << d << " " << result[d] << std::endl;
}
return add_to_cache(x, result, starting_multiple_);
}