Sprinkled in the new SUBPAR_VECTORIZABLE macro to encourage autovecto…

…rization.
tatami-inc · Oct 24, 2024 · d714d7f · d714d7f
1 parent a19defa
commit d714d7f
Show file tree

Hide file tree

Showing 5 changed files with 22 additions and 1 deletion.
diff --git a/extern/CMakeLists.txt b/extern/CMakeLists.txt
@@ -9,7 +9,7 @@ FetchContent_Declare(
 FetchContent_Declare(
   subpar
   GIT_REPOSITORY https://github.com/LTLA/subpar
-  GIT_TAG master # ^0.3.0
+  GIT_TAG master # ^0.3.2
 )
 
 FetchContent_MakeAvailable(tatami)

diff --git a/include/tatami_stats/counts.hpp b/include/tatami_stats/counts.hpp
@@ -111,6 +111,7 @@ void apply(bool row, const tatami::Matrix<Value_, Index_>* p, Output_* output, i
 
                 for (Index_ x = 0; x < len; ++x) {
                     auto range = ext->fetch(xbuffer.data(), ibuffer.data());
+                    SUBPAR_VECTORIZABLE
                     for (Index_ j = 0; j < range.number; ++j) {
                         auto idx = range.index[j];
                         curoutput[idx] += condition(range.value[j]);
@@ -119,6 +120,7 @@ void apply(bool row, const tatami::Matrix<Value_, Index_>* p, Output_* output, i
                 }
 
                 if (count_zero) {
+                    SUBPAR_VECTORIZABLE
                     for (int d = 0; d < dim; ++d) {
                         curoutput[d] += len - nonzeros[d];
                     }
@@ -133,6 +135,7 @@ void apply(bool row, const tatami::Matrix<Value_, Index_>* p, Output_* output, i
 
                 for (Index_ x = 0; x < len; ++x) {
                     auto ptr = ext->fetch(xbuffer.data());
+                    SUBPAR_VECTORIZABLE
                     for (Index_ j = 0; j < dim; ++j) {
                         curoutput[j] += condition(ptr[j]);
                     }
@@ -142,6 +145,7 @@ void apply(bool row, const tatami::Matrix<Value_, Index_>* p, Output_* output, i
 
         for (int t = 1; t < num_threads; ++t) {
             auto curoutput = threaded_output_ptrs[t];
+            SUBPAR_VECTORIZABLE
             for (Index_ d = 0; d < dim; ++d) {
                 output[d] += curoutput[d];
             }

diff --git a/include/tatami_stats/ranges.hpp b/include/tatami_stats/ranges.hpp
@@ -190,6 +190,7 @@ class RunningDense {
             ::tatami_stats::internal::nanable_ifelse<Value_>(
                 my_skip_nan,
                 [&]() {
+                    SUBPAR_VECTORIZABLE
                     for (Index_ i = 0; i < my_num; ++i, ++ptr) {
                         auto val = *ptr;
                         if (std::isnan(val)) {
@@ -205,6 +206,7 @@ class RunningDense {
             );
 
         } else {
+            SUBPAR_VECTORIZABLE
             for (Index_ i = 0; i < my_num; ++i, ++ptr) {
                 auto val = *ptr;
                 if (internal::is_better<minimum_>(my_store[i], val)) { // this should implicitly skip NaNs, any NaN comparison will be false.
@@ -269,6 +271,7 @@ class RunningSparse {
             std::fill_n(my_store, my_num, internal::choose_placeholder<minimum_, Value_>());
 
             if (!my_skip_nan) {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < number; ++i, ++value, ++index) {
                     auto val = *value;
                     auto idx = *index - my_subtract;
@@ -280,6 +283,7 @@ class RunningSparse {
             }
         }
 
+        SUBPAR_VECTORIZABLE
         for (Index_ i = 0; i < number; ++i, ++value, ++index) {
             auto val = *value;
             auto idx = *index - my_subtract;
@@ -298,6 +302,7 @@ class RunningSparse {
      */
     void finish() {
         if (my_count) {
+            SUBPAR_VECTORIZABLE
             for (Index_ i = 0; i < my_num; ++i) {
                 if (my_count > my_nonzero[i]) {
                     auto& current = my_store[i];

diff --git a/include/tatami_stats/sums.hpp b/include/tatami_stats/sums.hpp
@@ -108,6 +108,7 @@ class RunningDense {
         internal::nanable_ifelse<Value_>(
             my_skip_nan,
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < my_num; ++i) {
                     auto val = ptr[i];
                     if (!std::isnan(val)) {
@@ -116,6 +117,7 @@ class RunningDense {
                 }
             },
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < my_num; ++i) {
                     my_sum[i] += ptr[i];
                 }
@@ -164,6 +166,7 @@ class RunningSparse {
         internal::nanable_ifelse<Value_>(
             my_skip_nan,
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < number; ++i) {
                     auto val = value[i];
                     if (!std::isnan(val)) {
@@ -172,6 +175,7 @@ class RunningSparse {
                 }
             },
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < number; ++i) {
                     my_sum[index[i] - my_subtract] += value[i];
                 }

diff --git a/include/tatami_stats/variances.hpp b/include/tatami_stats/variances.hpp
@@ -210,6 +210,7 @@ class RunningDense {
         ::tatami_stats::internal::nanable_ifelse<Value_>(
             my_skip_nan,
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < my_num; ++i, ++ptr) {
                     auto val = *ptr;
                     if (!std::isnan(val)) {
@@ -219,6 +220,7 @@ class RunningDense {
             },
             [&]() {
                 ++my_count;
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < my_num; ++i, ++ptr) {
                     internal::add_welford(my_mean[i], my_variance[i], *ptr, my_count);
                 }
@@ -233,6 +235,7 @@ class RunningDense {
         ::tatami_stats::internal::nanable_ifelse<Value_>(
             my_skip_nan,
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < my_num; ++i) {
                     auto ct = my_ok_count[i];
                     if (ct < 2) {
@@ -252,6 +255,7 @@ class RunningDense {
                         std::fill_n(my_mean, my_num, std::numeric_limits<Output_>::quiet_NaN());
                     }
                 } else {
+                    SUBPAR_VECTORIZABLE
                     for (Index_ i = 0; i < my_num; ++i) {
                         my_variance[i] /= my_count - 1;
                     }
@@ -308,6 +312,7 @@ class RunningSparse {
         ::tatami_stats::internal::nanable_ifelse<Value_>(
             my_skip_nan,
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < number; ++i) {
                     auto val = value[i];
                     auto ri = index[i] - my_subtract;
@@ -319,6 +324,7 @@ class RunningSparse {
                 }
             },
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < number; ++i) {
                     auto ri = index[i] - my_subtract;
                     internal::add_welford(my_mean[ri], my_variance[ri], value[i], ++(my_nonzero[ri]));
@@ -334,6 +340,7 @@ class RunningSparse {
         ::tatami_stats::internal::nanable_ifelse<Value_>(
             my_skip_nan,
             [&]() {
+                SUBPAR_VECTORIZABLE
                 for (Index_ i = 0; i < my_num; ++i) {
                     auto& curM = my_mean[i];
                     auto& curV = my_variance[i];
@@ -357,6 +364,7 @@ class RunningSparse {
                         std::fill_n(my_mean, my_num, std::numeric_limits<Output_>::quiet_NaN());
                     }
                 } else {
+                    SUBPAR_VECTORIZABLE
                     for (Index_ i = 0; i < my_num; ++i) {
                         auto& var = my_variance[i];
                         internal::add_welford_zeros(my_mean[i], var, my_nonzero[i], my_count);