add key-value file!

.gitignore

@@ -0,0 +1,35 @@
+# Prerequisites
+*.d
+
+# Compiled Object files
+*.slo
+*.lo
+*.o
+*.obj
+
+# Precompiled Headers
+*.gch
+*.pch
+
+# Compiled Dynamic libraries
+*.so
+*.dylib
+*.dll
+
+# Fortran module files
+*.mod
+*.smod
+
+# Compiled Static libraries
+*.lai
+*.la
+*.a
+*.lib
+
+# Executables
+*.exe
+*.out
+*.app
+
+**/.vscode
+

Makefile

@@ -24,4 +24,4 @@ bench: $(BENCHDIR)/main.cpp $(SRCS)
 		$(CXX) $(BENCHDIR)/main.cpp $(CXXFLAGS) -march=icelake-client -O3 -o benchexe
 
 clean:
-		rm -f $(TESTDIR)/*.o testexe benchexe
+		rm -f $(TESTDIR)/*.o testexe benchexe

src/avx512-16bit-qsort.hpp

@@ -686,7 +686,7 @@ replace_inf_with_nan(uint16_t *arr, int64_t arrsize, int64_t nan_count)
 }
 
 template <>
-void avx512_qsort(int16_t *arr, int64_t arrsize)
+inline void avx512_qsort(int16_t *arr, int64_t arrsize)
 {
     if (arrsize > 1) {
         qsort_16bit_<zmm_vector<int16_t>, int16_t>(
@@ -695,7 +695,7 @@ void avx512_qsort(int16_t *arr, int64_t arrsize)
 }
 
 template <>
-void avx512_qsort(uint16_t *arr, int64_t arrsize)
+inline void avx512_qsort(uint16_t *arr, int64_t arrsize)
 {
     if (arrsize > 1) {
         qsort_16bit_<zmm_vector<uint16_t>, uint16_t>(

src/avx512-32bit-qsort.hpp

@@ -682,7 +682,7 @@ replace_inf_with_nan(float *arr, int64_t arrsize, int64_t nan_count)
 }
 
 template <>
-void avx512_qsort<int32_t>(int32_t *arr, int64_t arrsize)
+inline void avx512_qsort<int32_t>(int32_t *arr, int64_t arrsize)
 {
     if (arrsize > 1) {
         qsort_32bit_<zmm_vector<int32_t>, int32_t>(
@@ -691,7 +691,7 @@ void avx512_qsort<int32_t>(int32_t *arr, int64_t arrsize)
 }
 
 template <>
-void avx512_qsort<uint32_t>(uint32_t *arr, int64_t arrsize)
+inline void avx512_qsort<uint32_t>(uint32_t *arr, int64_t arrsize)
 {
     if (arrsize > 1) {
         qsort_32bit_<zmm_vector<uint32_t>, uint32_t>(
@@ -700,7 +700,7 @@ void avx512_qsort<uint32_t>(uint32_t *arr, int64_t arrsize)
 }
 
 template <>
-void avx512_qsort<float>(float *arr, int64_t arrsize)
+inline void avx512_qsort<float>(float *arr, int64_t arrsize)
 {
     if (arrsize > 1) {
         int64_t nan_count = replace_nan_with_inf(arr, arrsize);