cl3 0.9.1

// Copyright (c) 2020-2022 Via Technology Ltd.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//    http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

// This ia a CL_VERSION_1_2 so allow use of deprecated function: create_command_queue
#![allow(deprecated)]

extern crate cl3;

use cl3::command_queue::{
    create_command_queue, enqueue_nd_range_kernel, enqueue_read_buffer, enqueue_write_buffer,
    finish, release_command_queue, CL_QUEUE_PROFILING_ENABLE,
};
use cl3::context::{create_context, release_context};
use cl3::device::{
    get_device_ids, get_device_info, CL_DEVICE_TYPE_GPU, CL_DEVICE_VENDOR, CL_DEVICE_VENDOR_ID,
};
use cl3::event::{
    get_event_profiling_info, release_event, wait_for_events, CL_PROFILING_COMMAND_END,
    CL_PROFILING_COMMAND_START,
};
use cl3::kernel::{create_kernel, release_kernel, set_kernel_arg};
use cl3::memory::{create_buffer, release_mem_object, CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY};
use cl3::platform::{get_platform_ids, get_platform_info, CL_PLATFORM_NAME};
use cl3::program::{build_program, create_program_with_source, release_program};
use cl3::types::{cl_event, cl_float, cl_mem, CL_BLOCKING, CL_NON_BLOCKING};
use libc::{c_void, size_t};
use std::ffi::CString;
use std::mem;
use std::ptr;

const PROGRAM_SOURCE: &str = r#"
kernel void saxpy_float (global float* z,
    global float const* x,
    global float const* y,
    float a)
{
size_t i = get_global_id(0);
z[i] = a*x[i] + y[i];
}"#;

const KERNEL_NAME: &str = "saxpy_float";

#[test]
#[ignore]
fn test_opencl_1_2_example() {
    let platform_ids = get_platform_ids().unwrap();
    assert!(0 < platform_ids.len());

    // Choose the first platform
    let platform_id = platform_ids[0];
    let platform_name = get_platform_info(platform_id, CL_PLATFORM_NAME).unwrap();
    println!("Platform Name: {}", platform_name);

    let device_ids = get_device_ids(platform_id, CL_DEVICE_TYPE_GPU).unwrap();
    assert!(0 < device_ids.len());

    // Choose the first GPU device
    let device_id = device_ids[0];
    let vendor_name = get_device_info(device_id, CL_DEVICE_VENDOR).unwrap();
    println!("OpenCL device vendor name: {}", vendor_name);
    let vendor_id = get_device_info(device_id, CL_DEVICE_VENDOR_ID).unwrap();
    println!("OpenCL device vendor id: {:X}", u32::from(vendor_id));

    /////////////////////////////////////////////////////////////////////
    // Set up OpenCL compute environment

    // Create OpenCL context from the first device
    let device_ids = [device_id];
    let context = create_context(&device_ids, ptr::null(), None, ptr::null_mut()).unwrap();

    // Create the OpenCL program source
    let sources = [PROGRAM_SOURCE];
    let program = create_program_with_source(context, &sources).unwrap();

    // Build the OpenCL program for the device
    let build_options = CString::default();
    build_program(program, &device_ids, &build_options, None, ptr::null_mut()).unwrap();

    // Create the OpenCL kernel from the program
    let kernel_name = CString::new(KERNEL_NAME).unwrap();
    let kernel = create_kernel(program, &kernel_name).unwrap();

    // Create a command_queue for the device
    let queue =
        unsafe { create_command_queue(context, device_id, CL_QUEUE_PROFILING_ENABLE).unwrap() };

    /////////////////////////////////////////////////////////////////////
    // Process some data

    // The input data
    const ARRAY_SIZE: usize = 1000;
    let ones: [cl_float; ARRAY_SIZE] = [1.0; ARRAY_SIZE];
    let mut sums: [cl_float; ARRAY_SIZE] = [0.0; ARRAY_SIZE];
    for i in 0..ARRAY_SIZE {
        sums[i] = 1.0 + 1.0 * i as cl_float;
    }

    // Create OpenCL device buffers for input and output data
    let x = unsafe {
        create_buffer(
            context,
            CL_MEM_WRITE_ONLY,
            ARRAY_SIZE * mem::size_of::<cl_float>(),
            ptr::null_mut(),
        )
        .unwrap()
    };
    let y = unsafe {
        create_buffer(
            context,
            CL_MEM_WRITE_ONLY,
            ARRAY_SIZE * mem::size_of::<cl_float>(),
            ptr::null_mut(),
        )
        .unwrap()
    };
    let z = unsafe {
        create_buffer(
            context,
            CL_MEM_READ_ONLY,
            ARRAY_SIZE * mem::size_of::<cl_float>(),
            ptr::null_mut(),
        )
        .unwrap()
    };

    // Blocking write to OpenCL device buffer
    let x_write_event = unsafe {
        enqueue_write_buffer(
            queue,
            x,
            CL_BLOCKING,
            0,
            ones.len() * mem::size_of::<cl_float>(),
            ones.as_ptr() as cl_mem,
            0,
            ptr::null(),
        )
        .unwrap()
    };
    // Non-blocking write to OpenCL device buffer
    let y_write_event = unsafe {
        enqueue_write_buffer(
            queue,
            y,
            CL_NON_BLOCKING,
            0,
            sums.len() * mem::size_of::<cl_float>(),
            sums.as_ptr() as cl_mem,
            0,
            ptr::null(),
        )
        .unwrap()
    };

    // wait for y_write_event
    let mut events: Vec<cl_event> = Vec::default();
    events.push(y_write_event);
    wait_for_events(&events).unwrap();

    // a value for the kernel function
    let a: cl_float = 300.0;

    unsafe {
        // Set up the arguments to call the OpenCL kernel function
        // i.e. the x, y & z buffers and the constant value, a
        set_kernel_arg(
            kernel,
            0,
            mem::size_of::<cl_mem>(),
            &z as *const _ as *const c_void,
        )
        .unwrap();
        set_kernel_arg(
            kernel,
            1,
            mem::size_of::<cl_mem>(),
            &x as *const _ as *const c_void,
        )
        .unwrap();
        set_kernel_arg(
            kernel,
            2,
            mem::size_of::<cl_mem>(),
            &y as *const _ as *const c_void,
        )
        .unwrap();
        set_kernel_arg(
            kernel,
            3,
            mem::size_of::<cl_float>(),
            &a as *const _ as *const c_void,
        )
        .unwrap();
    }

    // Enqueue the OpenCL kernel for execution
    let global_work_sizes: [size_t; 1] = [ARRAY_SIZE];
    let kernel_event = unsafe {
        enqueue_nd_range_kernel(
            queue,
            kernel,
            1,
            ptr::null(),
            global_work_sizes.as_ptr(),
            ptr::null(),
            0,
            ptr::null(),
        )
        .unwrap()
    };

    // Push the kernel_event to the events wait list so that enqueue_read_buffer
    // can wait on it
    events.clear();
    events.push(kernel_event);

    // Create a results array to hold the results from the OpenCL device z buffer
    // and enqueue a read command to read the device buffer into the array
    // after the kernel event completes.
    let results: [cl_float; ARRAY_SIZE] = [0.0; ARRAY_SIZE];
    let read_event = unsafe {
        enqueue_read_buffer(
            queue,
            z,
            CL_NON_BLOCKING,
            0,
            results.len() * mem::size_of::<cl_float>(),
            results.as_ptr() as cl_mem,
            1,
            events.as_ptr(),
        )
        .unwrap()
    };
    events.clear();

    // Block until all commands on the queue (i.e. the read_event) have completed
    finish(queue).unwrap();

    // Test and print the results from OpenCL
    assert_eq!(1300.0, results[ARRAY_SIZE - 1]);
    println!("results back: {}", results[ARRAY_SIZE - 1]);

    let start_time = get_event_profiling_info(kernel_event, CL_PROFILING_COMMAND_START).unwrap();
    let end_time = get_event_profiling_info(kernel_event, CL_PROFILING_COMMAND_END).unwrap();
    let duration = u64::from(end_time) - u64::from(start_time);
    println!("kernel execution duration (ns): {}", duration);

    /////////////////////////////////////////////////////////////////////
    // Release OpenCL objects

    unsafe {
        release_event(x_write_event).unwrap();
        release_event(y_write_event).unwrap();
        release_event(kernel_event).unwrap();
        release_event(read_event).unwrap();
        release_mem_object(z).unwrap();
        release_mem_object(y).unwrap();
        release_mem_object(x).unwrap();

        // Release the OpenCL compute environment
        release_kernel(kernel).unwrap();
        release_program(program).unwrap();
        release_command_queue(queue).unwrap();
        release_context(context).unwrap();
    }
}