Merge pull request #5 from ratschlab/fixes

Authored-by: Harun Mustafa <[email protected]>
jermp · Jun 20, 2024 · 072f2cd · 072f2cd
2 parents 2a66378 + 29b6c6b
commit 072f2cd
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 4 deletions.
diff --git a/external/pthash b/external/pthash
diff --git a/include/minimizers.hpp b/include/minimizers.hpp
@@ -13,7 +13,7 @@ struct minimizers {
         mphf_config.seed = util::get_seed_for_hash_function(build_config);
         mphf_config.minimal_output = true;
         mphf_config.verbose_output = false;
-        mphf_config.num_threads = std::thread::hardware_concurrency();
+        mphf_config.num_threads = build_config.num_threads;
         mphf_config.num_partitions = 4 * mphf_config.num_threads;
 
         if (size / mphf_config.num_partitions < pthash::constants::min_partition_size) {
@@ -29,6 +29,10 @@ struct minimizers {
 
         mphf_config.ram = 4 * essentials::GB;
         mphf_config.tmp_dir = build_config.tmp_dirname;
+        if (size <= 1) {
+            throw std::runtime_error(
+                "each partition must contain more than one key: use less partitions");
+        }
         m_mphf.build_in_external_memory(begin, size, mphf_config);
     }
 
@@ -49,4 +53,4 @@ struct minimizers {
     minimizers_pthash_type m_mphf;
 };
 
-}  // namespace sshash
+}  // namespace sshash
diff --git a/include/util.hpp b/include/util.hpp
@@ -3,6 +3,7 @@
 #include <array>
 #include <cassert>
 #include <fstream>
+#include <thread>
 #include <cmath>  // for std::ceil on linux
 
 #include "hash_util.hpp"
@@ -74,6 +75,7 @@ struct build_configuration {
         : k(31)
         , m(17)
         , seed(constants::seed)
+        , num_threads(std::thread::hardware_concurrency())
 
         , l(constants::min_l)
         , c(constants::c)
@@ -87,6 +89,7 @@ struct build_configuration {
     uint64_t k;  // kmer size
     uint64_t m;  // minimizer size
     uint64_t seed;
+    uint64_t num_threads;
 
     uint64_t l;  // drive dictionary trade-off
     double c;    // drive PTHash trade-off
@@ -302,4 +305,4 @@ struct buffered_lines_iterator {
     uint64_t m_read_chars;
 };
 
-}  // namespace sshash
+}  // namespace sshash
+4 −0		README.md
+12 −4		include/builders/internal_memory_builder_partitioned_phf.hpp
+17 −27		include/builders/internal_memory_builder_single_phf.hpp
+22 −1		include/builders/util.hpp
+15 −0		include/partitioned_phf.hpp
+12 −1		include/single_phf.hpp
+6 −2		include/utils/bucketers.hpp
+4 −3		src/example.cpp