use the zstd version 1.5.5

cwida · Nov 18, 2024 · 009c37b · 009c37b
1 parent ba708f9
commit 009c37b
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 22 deletions.
diff --git a/publication/compression_ratio_result/float/zstd.csv b/publication/compression_ratio_result/float/zstd.csv
@@ -1,5 +1,5 @@
 dataset,size
-Dino-Vitb16,39.57
-GPT2,39.76
-Grammarly-lg,39.37
-WAV2VEC,43.15
+Dino-Vitb16,14.87
+GPT2,14.85
+Grammarly-lg,14.82
+WAV2VEC,9.55
diff --git a/publication/source_code/CMakeLists.txt b/publication/source_code/CMakeLists.txt
@@ -7,8 +7,9 @@ endif ()
 
 FetchContent_Declare(
         zstd
-        GIT_REPOSITORY https://github.com/facebook/zstd
-        GIT_TAG 794ea1b0afca0f020f4e57b6732332231fb23c70)
+        GIT_REPOSITORY https://github.com/facebook/zstd.git
+        GIT_TAG v1.5.5
+)
 
 FetchContent_MakeAvailable(zstd)
 

diff --git a/publication/source_code/bench_compression_ratio/CMakeLists.txt b/publication/source_code/bench_compression_ratio/CMakeLists.txt
@@ -13,7 +13,7 @@ gtest_discover_tests(publication_bench_alp32_compression_ratio)
 
 # Test ZSTD: ----------------------------------------------------------------------------------------------------------
 add_executable(publication_bench_zstd_compression_ratio zstd.cpp)
-target_link_libraries(publication_bench_zstd_compression_ratio PRIVATE gtest_main libzstd)
+target_link_libraries(publication_bench_zstd_compression_ratio PRIVATE gtest_main libzstd_shared)
 gtest_discover_tests(publication_bench_zstd_compression_ratio)
 
 # Test Chimp: ----------------------------------------------------------------------------------------------------------

diff --git a/publication/source_code/bench_compression_ratio/zstd.cpp b/publication/source_code/bench_compression_ratio/zstd.cpp
@@ -10,26 +10,23 @@ class zstd_test : public ::testing::Test {
 	void*  dec_dbl_arr;
 	size_t zstd_vector_size =
 	    alp::config::ROWGROUP_SIZE; // For Zstd we compress rowgroups since it would be unfair to compress small vectors
-	size_t enc_size_upper_bound = zstd_vector_size * 8;
-	size_t input_size           = zstd_vector_size * 8;
-	size_t dec_size             = input_size;
 
-	void SetUp() override {
-		enc_dbl_arr = malloc(input_size);
-		dec_dbl_arr = malloc(input_size);
-	}
+	void SetUp() override {}
 
-	~zstd_test() override {
-		free(enc_dbl_arr);
-		free(dec_dbl_arr);
-	}
+	~zstd_test() override {}
 
 	template <typename T, int N_DATASETS>
 	void bench_compression_ratio(const std::array<alp_bench::Column, N_DATASETS>& datasets, const std::string& path) {
 		if (const auto v = std::getenv("ALP_DATASET_DIR_PATH"); v != nullptr) {
 			alp_bench::get_paths().alp_dataset_binary_dir_path = *v;
 		}
 
+		size_t enc_size_upper_bound = zstd_vector_size * sizeof(T);
+		size_t input_size           = zstd_vector_size * sizeof(T);
+		size_t dec_size             = input_size;
+		enc_dbl_arr                 = malloc(input_size);
+		dec_dbl_arr                 = malloc(input_size);
+
 		std::ofstream ofile(path, std::ios::out);
 		ofile << "dataset,size\n";
 
@@ -48,8 +45,8 @@ class zstd_test : public ::testing::Test {
 
 			if (tuples_count < zstd_vector_size) {
 				zstd_vector_size     = tuples_count;
-				input_size           = zstd_vector_size * 8;
-				enc_size_upper_bound = zstd_vector_size * 8;
+				input_size           = zstd_vector_size * sizeof(T);
+				enc_size_upper_bound = zstd_vector_size * sizeof(T);
 			}
 
 			/* Encode - Decode - Validate. */
@@ -67,7 +64,7 @@ class zstd_test : public ::testing::Test {
 				    ZSTD_compress(enc_dbl_arr, enc_size_upper_bound, dbl_arr, input_size, 3); // Level 3
 
 				// SUM COMPRESSED SIZE
-				compressed_data_size += ENC_SIZE * 8;
+				compressed_data_size += ENC_SIZE * sizeof(T);
 
 				// Decode
 				ZSTD_decompress(dec_dbl_arr, dec_size, enc_dbl_arr, ENC_SIZE);
@@ -83,10 +80,12 @@ class zstd_test : public ::testing::Test {
 
 			auto compression_ratio = (double)compressed_data_size / processed_tuples;
 
-			ofile << std::fixed << std::setprecision(2) << dataset.name << "," << compression_ratio << "\n";
+			ofile << std::fixed << std::setprecision(2) << dataset.name << "," << compression_ratio << std::endl;
 		}
 
 		delete[] dbl_arr;
+		free(enc_dbl_arr);
+		free(dec_dbl_arr);
 	}
 };