simd_dna4

seqan · Jun 21, 2022 · 62bcff8 · 62bcff8
1 parent b11753c
commit 62bcff8
Show file tree

Hide file tree

Showing 10 changed files with 241 additions and 228 deletions.
diff --git a/...performance/alphabet/simdifyable_dna4.hpp → ...an3/test/performance/simdifyable_dna4.hpp b/...performance/alphabet/simdifyable_dna4.hpp → ...an3/test/performance/simdifyable_dna4.hpp
@@ -7,8 +7,6 @@
 
 #pragma once
 
-#include <vector>
-
 #include <seqan3/alphabet/nucleotide/nucleotide_base.hpp>
 
 // ------------------------------------------------------------------
@@ -24,18 +22,13 @@
 namespace seqan3
 {
 
-class rna4;
-
 class simdifyable_dna4 : public nucleotide_base<simdifyable_dna4, 256>
 {
 private:
     using base_t = nucleotide_base<simdifyable_dna4, 256>;
 
     friend base_t;
-    //!\cond \brief Befriend seqan3::alphabet_base.
     friend base_t::base_t;
-    //!\endcond
-    friend rna4;
 
 public:
     constexpr simdifyable_dna4() noexcept = default;
@@ -46,52 +39,58 @@ class simdifyable_dna4 : public nucleotide_base<simdifyable_dna4, 256>
     ~simdifyable_dna4() noexcept = default;
 
     using base_t::base_t;
-
-    template <std::same_as<rna4> t> // template parameter t to accept incomplete type
-    constexpr simdifyable_dna4(t const & r) noexcept
-    {
-        assign_rank(r.to_rank());
-    }
+    using base_t::to_rank;
+    using base_t::assign_rank;
 
     static constexpr uint8_t alphabet_size = 4;
 
-    constexpr simdifyable_dna4 & assign_rank(rank_type const c) noexcept
-    {
-        base_t::assign_rank(c == 0 ? 'A' : (c == 1 ? 'C' : (c == 2 ? 'G' : 'T')));
-        return *this;
-    }
-
     constexpr simdifyable_dna4 & assign_char(char_type const c) noexcept
     {
-        base_t::assign_rank(c);
-        return *this;
+        char_type const upper_char = c & 0b0101'1111; // to_upper
+        rank_type rank = (upper_char == 'T') * 3 + (upper_char == 'G') * 2 + (upper_char == 'C');
+        return assign_rank(rank);
     }
 
+    // Faster for single assign_char, slower for auto-vectorized
+    // constexpr simdifyable_dna4 & assign_char(char_type const c) noexcept
+    // {
+    //     rank_type rank{static_cast<rank_type>(c)};
+    //     rank |= (rank & 0b0001'0000) >> 1;
+    //     rank += 1u;
+    //     rank >>= 2;
+    //     rank &= 0b0000'0011;
+    //     return assign_rank(rank);
+    // }
+
     constexpr char_type to_char() const noexcept
     {
-        return base_t::to_rank();
+        rank_type const rank = to_rank();
+        switch (rank)
+        {
+        case 0u:
+            return 'A';
+        case 1u:
+            return 'C';
+        case 2u:
+            return 'G';
+        default:
+            return 'T';
+        }
     }
 
-    constexpr rank_type to_rank() const noexcept
+    constexpr simdifyable_dna4 complement() const noexcept
     {
-        char_type c{to_char()};
-        char_type upper_char = c & 0b0101'1111; // to_upper
-        rank_type rank{};
-        rank = (upper_char == 'T') * 3 + (upper_char == 'G') * 2 + (upper_char == 'C');
-        return rank;
+        rank_type rank{to_rank()};
+        rank ^= 0b11;
+        simdifyable_dna4 ret{};
+        return ret.assign_rank(rank);
     }
 
-    constexpr simdifyable_dna4 complement() const noexcept
+    // For alphabet tests
+    static constexpr bool char_is_valid(char_type const) noexcept
     {
-        char_type rank{to_char()};
-        rank ^= rank % 2 ? ('C' ^ 'G') : ('A' ^ 'T');
-
-        simdifyable_dna4 ret{};
-        static_cast<base_t &>(ret).assign_rank(rank);
-        return ret;
+        return false;
     }
 };
 
-using simdifyable_dna4_vector = std::vector<simdifyable_dna4>;
-
 } // namespace seqan3
diff --git a/test/performance/alphabet/CMakeLists.txt b/test/performance/alphabet/CMakeLists.txt
@@ -2,6 +2,6 @@ add_subdirectories ()
 
 seqan3_benchmark (alphabet_assign_char_benchmark.cpp)
 seqan3_benchmark (alphabet_assign_rank_benchmark.cpp)
-seqan3_benchmark (alphabet_complement.cpp)
+seqan3_benchmark (alphabet_complement_benchmark.cpp)
 seqan3_benchmark (alphabet_to_char_benchmark.cpp)
 seqan3_benchmark (alphabet_to_rank_benchmark.cpp)
diff --git a/test/performance/alphabet/alphabet_assign_char_benchmark.cpp b/test/performance/alphabet/alphabet_assign_char_benchmark.cpp
@@ -12,6 +12,7 @@
 #include <numeric>
 
 #include <seqan3/alphabet/all.hpp>
+#include <seqan3/test/performance/simdifyable_dna4.hpp>
 #include <seqan3/test/seqan2.hpp>
 
 #if SEQAN3_HAS_SEQAN2
@@ -38,6 +39,7 @@ void assign_char(benchmark::State & state)
 BENCHMARK_TEMPLATE(assign_char, seqan3::gap);
 BENCHMARK_TEMPLATE(assign_char, seqan3::dna4);
 BENCHMARK_TEMPLATE(assign_char, seqan3::rna4);
+BENCHMARK_TEMPLATE(assign_char, seqan3::simdifyable_dna4);
 BENCHMARK_TEMPLATE(assign_char, seqan3::dna5);
 BENCHMARK_TEMPLATE(assign_char, seqan3::rna5);
 BENCHMARK_TEMPLATE(assign_char, seqan3::dna15);

diff --git a/test/performance/alphabet/alphabet_assign_rank_benchmark.cpp b/test/performance/alphabet/alphabet_assign_rank_benchmark.cpp
@@ -11,6 +11,7 @@
 #include <cstring>
 
 #include <seqan3/alphabet/all.hpp>
+#include <seqan3/test/performance/simdifyable_dna4.hpp>
 #include <seqan3/test/seqan2.hpp>
 
 #if SEQAN3_HAS_SEQAN2
@@ -45,6 +46,7 @@ void assign_rank(benchmark::State & state)
 BENCHMARK_TEMPLATE(assign_rank, seqan3::gap);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::dna4);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::rna4);
+BENCHMARK_TEMPLATE(assign_rank, seqan3::simdifyable_dna4);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::dna5);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::rna5);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::dna15);

diff --git a/test/performance/alphabet/alphabet_complement.cpp b/test/performance/alphabet/alphabet_complement.cpp