JavaCPP Presets

To convert row-wise data (like a list of objects) into an Apache Arrow Table, use specialized ArrayBuilder classes.

Key steps:

1. Initialize a MemoryPool (e.g., default_memory_pool()).
2. Create builders for each type (e.g., Int64Builder, DoubleBuilder, ListBuilder).
3. For nested structures like lists, use a ListBuilder and a nested value builder (e.g., new DoubleBuilder(components_builder.value_builder())).
4. Iterate through your data, calling .Append() or .AppendValues() on the builders.
5. Finalize the arrays using .Finish(Array) and combine them into a Table using Table.Make(schema, ArrayVector).

// Example snippet for building a table
MemoryPool pool = default_memory_pool();
Int64Builder id_builder = new Int64Builder(int64(), pool);
DoubleBuilder cost_builder = new DoubleBuilder(float64(), pool);
ListBuilder components_builder = new ListBuilder(pool, new DoubleBuilder(float64(), pool));
DoubleBuilder cost_components_builder = new DoubleBuilder(components_builder.value_builder());

// ... loop and append data ...

Array id_array = new Array(null);
THROW_ON_FAILURE(id_builder.Finish(id_array));
// ... finish other arrays ...

FieldVector schema_vector = new FieldVector(
    new Field("id", int64()), 
    new Field("cost", float64()),
    new Field("cost_components", list(float64()))
);
Schema schema = new Schema(schema_vector);
table[0] = Table.Make(schema, new ArrayVector(id_array, cost_array, cost_components_array));

You can use Maven 3 to automatically download and install all necessary class files and native binaries. To run a sample project, create a pom.xml with the required dependencies for the CUDA platform and its redistributable components (cuBLAS, cuDNN, NCCL, etc.).

<project>
    <modelVersion>4.0.0</modelVersion>
    <groupId>org.bytedeco.cuda</groupId>
    <artifactId>mnistcudnn</artifactId>
    <version>1.5.14-SNAPSHOT</version>
    <properties>
        <exec.mainClass>MNISTCUDNN</exec.mainClass>
    </properties>
    <dependencies>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>

        <!-- Additional dependencies to use bundled CUDA, cuDNN, and NCCL -->
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-cublas</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-cudnn</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-cusolver</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-cusparse</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-npp</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-nccl</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist-nvcomp</artifactId>
            <version>13.3-9.24-1.5.14-SNAPSHOT</version>
        </dependency>
    </dependencies>
    <build>
        <sourceDirectory>.</sourceDirectory>
    </build>
</project>

After configuring the pom.xml and your Java source files, execute the following command to compile and run:

$ mvn compile exec:java

You can use Maven 3 to automatically download and install all necessary Java class files and native binaries for Caffe.

To include Caffe with GPU support and bundled CUDA/cuDNN, use the following dependencies in your pom.xml:

1. caffe-platform for core functionality.
2. caffe-platform-gpu for GPU support.
3. cuda-platform-redist for bundled CUDA and cuDNN binaries.

To run a compiled Java application (like the sample caffe.java), use the following command:

mvn compile exec:java -Dexec.args="..."

<project>
    <modelVersion>4.0.0</modelVersion>
    <groupId>org.bytedeco.caffe</groupId>
    <artifactId>caffe</artifactId>
    <version>1.5.8</version>
    <properties>
        <exec.mainClass>caffe</exec.mainClass>
    </properties>
    <dependencies>
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>caffe-platform</artifactId>
            <version>1.0-1.5.8</version>
        </dependency>

        <!-- Additional dependencies required to use CUDA and cuDNN -->
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>caffe-platform-gpu</artifactId>
            <version>1.0-1.5.8</version>
        </dependency>

        <!-- Additional dependencies to use bundled CUDA and cuDNN -->
        <dependency>
            <groupId>org.bytedeco</groupId>
            <artifactId>cuda-platform-redist</artifactId>
            <version>11.8-8.6-1.5.8</version>
        </dependency>

    </dependencies>
    <build>
        <sourceDirectory>.</sourceDirectory>
    </build>
</project>