haraldwolff
/
CLBlast
mirror of https://github.com/CNugteren/CLBlast.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fixed an overflow bug on 32-bit systems when chosing a GEMM kernel

pull/176/head
Cedric Nugteren 2017-06-18 20:51:11 +02:00
parent 33ed1e5a06
commit 3070b502b5
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/routines/level3/xgemm.cpp
View File

@ -102,7 +102,9 @@ void Xgemm<T>::DoGemm(const Layout layout,
TestMatrixC(c_one, c_two, c_buffer, c_offset, c_ld);
// Selects which version of GEMM to run
const auto do_gemm_direct = (m * n * k < db_["XGEMM_MIN_INDIRECT_SIZE"]);
const auto m_n_k = static_cast<unsigned long>(m) * static_cast<unsigned long>(n) *
static_cast<unsigned long>(k);
const auto do_gemm_direct = (m_n_k < static_cast<unsigned long>(db_["XGEMM_MIN_INDIRECT_SIZE"]));
if (do_gemm_direct) { // for small sizes (single kernel)
GemmDirect(m, n, k, alpha,
a_buffer, a_offset, a_ld, b_buffer, b_offset, b_ld, beta,