Merge pull request #12 from aliddell/add-acquire-again

Add Acquire Zarr to benchmark suite

.gitignore

@@ -4,4 +4,5 @@ __pycache__/
 .vscode/
 example_data/
 build/
-test.*
+test.*
+venv/

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "acquire-zarr"]
+	path = external/acquire-zarr
+	url = [email protected]:aliddell/acquire-zarr.git

CMakeLists.txt

@@ -1,22 +1,48 @@
-cmake_minimum_required(VERSION 3.2)
-project(cpp-zarr)
+cmake_minimum_required(VERSION 3.23)
+project(zarr-benchmarks)
 cmake_policy(SET CMP0057 NEW)
+cmake_policy(SET CMP0079 NEW)
 
-set(CMAKE_CXX_STANDARD 11)
-set(CMAKE_CXX_STANDARD_REQUIRED True)
+set(CMAKE_CXX_STANDARD 20)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
 
-# finding the cppZarr lib
+# Common dependencies
+find_package(Python3 REQUIRED COMPONENTS Interpreter Development)
+find_package(pybind11 REQUIRED)
+
+# Dependencies for acquire-zarr
+find_package(nlohmann_json CONFIG REQUIRED)
+find_package(blosc CONFIG REQUIRED)
+find_package(miniocpp CONFIG REQUIRED)
+
+include_directories(
+    /usr/local/include
+    ${Python3_INCLUDE_DIRS}
+    ${pybind11_INCLUDE_DIR}
+)
+
+# construct bindings for cppZarr
 find_library(cppZarrLib cppZarr REQUIRED)
+pybind11_add_module(pyCppZarr zarr_libraries/cpp_zarr/cpp_zarr.cpp)
+target_link_libraries(pyCppZarr PRIVATE ${cppZarrLib})
 
-# setting include dir for dependencies
-include_directories(/usr/local/include)
+# build acquire-zarr
+set(BUILD_PYTHON ON CACHE BOOL "Build Python bindings for acquire-zarr")
+set(BUILD_TESTING OFF CACHE BOOL "Disable acquire-zarr tests")
+add_subdirectory(external/acquire-zarr)
 
-# pybind11
-find_package(Python3 REQUIRED COMPONENTS Interpreter Development)
-include_directories(${Python3_INCLUDE_DIRS})
+add_custom_target(install-acquire-zarr-python
+    COMMAND ${Python3_EXECUTABLE} -m pip install .
+    WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/external/acquire-zarr
+    DEPENDS acquire-zarr
+    COMMENT "Installing acquire-zarr Python package"
+)
 
-find_package(pybind11 REQUIRED)
-include_directories(${pybind11_INCLUDE_DIR})
+# Make the Python installation part of the default build
+add_dependencies(pyCppZarr install-acquire-zarr-python)
 
-pybind11_add_module(pyCppZarr zarr_libraries/cpp_zarr/cpp_zarr.cpp)
-target_link_libraries(pyCppZarr PRIVATE ${cppZarrLib})
+# Installation rules
+install(TARGETS pyCppZarr
+        LIBRARY DESTINATION ${Python3_SITEARCH}/zarr_benchmarks
+        RUNTIME DESTINATION ${Python3_SITEARCH}/zarr_benchmarks)

CMakePresets.json

@@ -0,0 +1,12 @@
+{
+  "version": 3,
+  "configurePresets": [
+    {
+      "name": "default",
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_TOOLCHAIN_FILE": "$env{VCPKG_ROOT}/scripts/buildsystems/vcpkg.cmake"
+      }
+    }
+  ]
+}

benchmark.py

@@ -11,10 +11,11 @@ def __init__(self, shape: list, chunks: list) -> None:
         self.__chunks = chunks
         self.__average_bandwidth = {}
         self.__zarr_writers = {
-            "TensorStore" : Tensorstore(),
-            "Zarr Python" : Zarr_Python(),
-            "OME Zarr"    : Ome_Zarr(),
-            "Cpp Zarr"    : Cpp_Zarr()
+            "TensorStore"  : Tensorstore(),
+            "Zarr Python"  : Zarr_Python(),
+            "OME Zarr"     : Ome_Zarr(),
+            "Cpp Zarr"     : Cpp_Zarr(),
+            "Acquire Zarr" : AcquireZarr()
         }
 
 
@@ -46,7 +47,7 @@ def run_write_tests(self, num_of_gigabytes: int, show_results: bool,
                         avg_graph: Optional[matplotlib.axes._axes.Axes] = None) -> None:
 
         # error checking to see if chosen lib exists in test
-        if choose_lib != None and choose_lib not in set(self.__zarr_writers.keys()):
+        if choose_lib is not None and choose_lib not in set(self.__zarr_writers.keys()):
             raise ValueError(f"There is no library of name \"{choose_lib}\".") 
 
         gb_in_bytes = 1073741824         # represents number of bytes in a GB
@@ -77,11 +78,11 @@ def run_write_tests(self, num_of_gigabytes: int, show_results: bool,
                     shutil.rmtree(writer.data_path)
 
                 # if a specified library is chosen for testing, skip any that isn't that test
-                if choose_lib != None and choose_lib != lib_name: 
+                if choose_lib is not None and choose_lib != lib_name: 
                     continue 
 
                 # store time taken to write zarr
-                if lib_name == "TensorStore" or lib_name == "Zarr Python":
+                if lib_name in ("TensorStore", "Zarr Python", "Acquire Zarr"):
                     total_time = writer.write_zarr(shape=new_shape, chunks=self.chunks, zarr_data=zarr_data)
                 elif lib_name == "OME Zarr":
                     total_time = writer.write_zarr(chunks=self.chunks, zarr_data=zarr_data)
@@ -102,7 +103,7 @@ def run_write_tests(self, num_of_gigabytes: int, show_results: bool,
         # plot the data and clean up the folders
         for lib_name, writer in self.__zarr_writers.items():
             # if a specified library is chosen for testing, skip any that isn't that test
-            if choose_lib != None and choose_lib != lib_name: 
+            if choose_lib is not None and choose_lib != lib_name: 
                 continue 
 
             # cleans up data left behind
@@ -125,11 +126,11 @@ def run_append_tests(self, num_of_gigabytes: int, show_results: bool,
                         avg_graph: Optional[matplotlib.axes._axes.Axes] = None) -> None:
 
         # error checking to see if chosen lib exists in test
-        if choose_lib != None and choose_lib not in set(self.__zarr_writers.keys()):
+        if choose_lib is not None and choose_lib not in set(self.__zarr_writers.keys()):
             raise ValueError(f"There is no library of name \"{choose_lib}\".") 
 
         # these are the only libraries that allow for appending of data
-        if choose_lib != None and choose_lib != "TensorStore" and choose_lib != "Zarr Python":
+        if choose_lib is not None and choose_lib != "TensorStore" and choose_lib != "Zarr Python":
             return
 
         gb_in_bytes = 1073741824         # represents number of bytes in a GB
@@ -157,7 +158,7 @@ def run_append_tests(self, num_of_gigabytes: int, show_results: bool,
 
             for lib_name, writer in self.__zarr_writers.items():    
                 # if a specified library is chosen for testing, skip any that isn't that test
-                if choose_lib != None and choose_lib != lib_name: 
+                if choose_lib is not None and choose_lib != lib_name: 
                     continue 
 
                 # store time taken to append data
@@ -184,7 +185,7 @@ def run_append_tests(self, num_of_gigabytes: int, show_results: bool,
                 continue
 
             # if a specified library is chosen for testing, skip any that isn't that test
-            if choose_lib != None and choose_lib != lib_name: 
+            if choose_lib is not None and choose_lib != lib_name: 
                 continue 
 
             shutil.rmtree(writer.data_path) 
@@ -204,7 +205,7 @@ def run_all_tests(self, append_test_gigabytes: int, write_test_gigabytes: int,
                       append_graph: Optional[matplotlib.axes._axes.Axes] = None, append_avg_graph: Optional[matplotlib.axes._axes.Axes] = None,
                       write_graph: Optional[matplotlib.axes._axes.Axes] = None, write_avg_graph: Optional[matplotlib.axes._axes.Axes] = None) -> None:
 
-        self.run_append_tests(num_of_gigabytes=append_test_gigabytes, show_results=False, choose_lib=choose_lib, graph=append_graph, avg_graph=append_avg_graph)
         self.run_write_tests(num_of_gigabytes=write_test_gigabytes, show_results=False, choose_lib=choose_lib, graph=write_graph, avg_graph=write_avg_graph)
+        self.run_append_tests(num_of_gigabytes=append_test_gigabytes, show_results=False, choose_lib=choose_lib, graph=append_graph, avg_graph=append_avg_graph)
         self.__print_results(additional_info=(f"Write Test GB Soft Cap: {write_test_gigabytes}GB | Append Test GB Soft Cap: {append_test_gigabytes}GB"))
-
+

setup.sh

@@ -18,7 +18,7 @@ echo "Directory '$BUILD_DIR' has been created successfully."
 
 # build cpp code 
 cd "$BUILD_DIR" || { echo "Failed to change directory to '$BUILD_DIR'"; exit 1; }
-cmake .. || { echo "Failed to run cmake '$BUILD_DIR'"; exit 1; }
-make || { echo "Failed to run make"; exit 1; }
+cmake --preset=default .. || { echo "Failed to run cmake '$BUILD_DIR'"; exit 1; }
+cmake --build . || { echo "Failed to run make"; exit 1; }
 
 echo "Build process completed successfully."

vcpkg-configuration.json

@@ -0,0 +1,14 @@
+{
+  "default-registry": {
+    "kind": "git",
+    "baseline": "f7423ee180c4b7f40d43402c2feb3859161ef625",
+    "repository": "https://github.com/microsoft/vcpkg"
+  },
+  "registries": [
+    {
+      "kind": "artifact",
+      "location": "https://github.com/microsoft/vcpkg-ce-catalog/archive/refs/heads/main.zip",
+      "name": "microsoft"
+    }
+  ]
+}

vcpkg.json

@@ -0,0 +1,16 @@
+{
+  "dependencies": [
+    {
+      "name": "blosc",
+      "version>=": "1.21.5"
+    },
+    {
+      "name": "nlohmann-json",
+      "version>=": "3.11.3"
+    },
+    {
+      "name": "minio-cpp",
+      "version>=": "0.3.0"
+    }
+  ]
+}

zarr_libraries/__init__.py

@@ -2,4 +2,5 @@
 from zarr_libraries.tensorstore.tensorstore_zarr import *
 from zarr_libraries.zarr_python.zarr_python import *
 from zarr_libraries.ome_ngff.ome_zarr import *
-from zarr_libraries.cpp_zarr.cpp_zarr import *
+from zarr_libraries.cpp_zarr.cpp_zarr import *
+from zarr_libraries.acquire_zarr import AcquireZarr

zarr_libraries/acquire_zarr.py

@@ -0,0 +1,62 @@
+import acquire_zarr as aqz
+
+from pathlib import Path
+import numpy as np
+import time
+
+
+class AcquireZarr:
+    def __init__(self) -> None:
+        aqz.set_log_level(aqz.LogLevel.Error)
+        base_dir = Path(__file__).parent
+        self.__path_to_data = (
+            base_dir / "example_data" / "acquire_zarr_data" / "test.zarr"
+        ).resolve()
+
+        if not self.__path_to_data.parent.exists():
+            self.__path_to_data.parent.mkdir(parents=True)
+
+    @property
+    def data_path(self) -> str:
+        return str(self.__path_to_data)
+
+    def write_zarr(self, shape: list, chunks: list, zarr_data: np.ndarray) -> float:
+        settings = aqz.StreamSettings()
+        settings.store_path = self.data_path
+        settings.compression = aqz.CompressionSettings(
+            compressor=aqz.Compressor.BLOSC1,
+            codec=aqz.CompressionCodec.BLOSC_LZ4,
+            level=1,
+            shuffle=0
+        )
+        settings.data_type=aqz.DataType.UINT8
+        settings.dimensions.extend(
+            [
+                aqz.Dimension(
+                    name="t",
+                    kind=aqz.DimensionType.TIME,
+                    array_size_px=shape[0],
+                    chunk_size_px=chunks[0],
+                ),
+                aqz.Dimension(
+                    name="y",
+                    kind=aqz.DimensionType.SPACE,
+                    array_size_px=shape[1],
+                    chunk_size_px=chunks[1],
+                ),
+                aqz.Dimension(
+                    name="x",
+                    kind=aqz.DimensionType.SPACE,
+                    array_size_px=shape[2],
+                    chunk_size_px=chunks[2],
+                ),
+            ]
+        )
+        settings.version = aqz.ZarrVersion.V2
+        stream = aqz.ZarrStream(settings)
+
+        t = time.perf_counter()
+        stream.append(zarr_data)
+        total_time = time.perf_counter() - t
+
+        return total_time

zarr_libraries/common.py

@@ -3,6 +3,8 @@
 
 # getting size of zarr folder recursively 
 def folder_size_in_bytes(folder: str) -> int:
+    if not os.path.exists(folder):
+        return 0
     total_size = os.path.getsize(folder)
 
     for item in os.listdir(folder):