Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

further optimization fixes #54

Merged
merged 2 commits into from
Jun 2, 2024
Merged

further optimization fixes #54

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6d65456
Makefile.am: adding warnings and ignoring unknown attributes for the …
andrewdavidsmith Jun 2, 2024
a4810f0
AbismalAlign.hpp: Changing the way loop vectorization is turned off f…
andrewdavidsmith Jun 2, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions Makefile.am
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,7 @@ install installdirs: SUBDIRS := $(filter-out src/smithlab_cpp, $(SUBDIRS))
AM_CPPFLAGS = -I $(top_srcdir)/src/smithlab_cpp -I $(top_srcdir)/src/bamxx

AM_CXXFLAGS = $(OPENMP_CXXFLAGS)
AM_CXXFLAGS += -Wall -Wextra -Wpedantic -Wno-unknown-attributes
if ENABLE_SHORT
AM_CXXFLAGS += -DENABLE_SHORT
endif
Expand Down
22 changes: 7 additions & 15 deletions src/AbismalAlign.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -206,21 +206,15 @@ get_best_score(const std::vector<score_t> &table, const size_t n_cells,
return *best_cell_itr;
}

// ADS: it seems like with g++-13, on macos ventura on intel hardware
// the dynamic vectorized optimization of -O3 might be too aggressive
// and makes this function have strange behavior. Placing this pragma
// here helps, and below we restore it to the `-O3` default. Probably
// should move to attribute syntax soon.

#ifdef __APPLE__
#pragma GCC push_options
// ADS: below this won't make sense if the user wants no optimizations
// at all...
#pragma GCC optimize ("O2")
#endif
// ADS: it seems like with some versions of GCC, the optimization from
// `-O3`, which seems to include loop vectorizations, breaks the
// function below and the other overload named `from_diag`
// below. Using the __attribute__ helps with GCC, and it should be
// ignored if not supported.

template<score_t (*scr_fun)(const uint8_t, const uint8_t), class T,
class QueryConstItr>
__attribute__((optimize("no-tree-loop-vectorize")))
void
from_diag(T next_row, const T next_row_end, T cur_row, QueryConstItr query_seq,
uint8_t ref_base) {
Expand Down Expand Up @@ -251,6 +245,7 @@ from_left(T left_itr, T target, const T target_end) {
/********* SAME FUNCTIONS AS ABOVE BUT WITH TRACEBACK ********/
template<score_t (*scr_fun)(const uint8_t, const uint8_t), class T,
class QueryConstItr, class U>
__attribute__((optimize("no-tree-loop-vectorize")))
void
from_diag(T next_row, const T next_row_end, T cur_row, QueryConstItr query_seq,
uint8_t ref_base, U traceback) {
Expand Down Expand Up @@ -289,9 +284,6 @@ from_left(T left_itr, T target, const T target_end, U traceback) {
}
}

#ifdef __APPLE__
#pragma GCC pop_options
#endif

inline void
make_default_cigar(const uint32_t len, std::string &cigar) {
Expand Down
Loading