image/docs/usecase4.md

Go to the documentation of this file.

# Copyright (c) 2017-2019, NVIDIA CORPORATION.  All rights reserved.

@page image_usecase4 Image Streamer Multi-Thread
@tableofcontents

@note SW Release Applicability: This tutorial is applicable to modules in both **NVIDIA DriveWorks** and **NVIDIA DRIVE Software** releases.

In addition to using the ImageStreamer module in a [single thread](@ref image_usecase2), you can also use it across multiple threads.

This workflow is similar to the single thread case since the same APIs are thread safe.<br>
For details on how to stream an image from CPU to CUDA memory, please refer to \n @ref image_usecase2. All other details are left out of this tutorial for simplicity.

@section image_usecase4_threadA Initializing the ImageStreamer Module in Thread A

In the following example, the ImageStreamer is initialized in Thread A before it is used to send images:

```{.cpp}
    // Thread A:
    // streamer from CPU to CUDA
    dwImageStreamerHandle_t streamerCPU2CUDA = DW_NULL_HANDLE;

    dwImageStreamer_initialize(streamerCPU2CUDA, &cpuProp, DW_IMAGE_CUDA, dwContext);
    while(true)
    {
        // CODE: Get an image to stream
        dwImageStreamer_producerSend(imgCPU, streamerCPU2CUDA);
        dwImageStreamer_producerReturn(nullptr, TIMEOUT, streamerCPU2CUDA);
    }
```

@section image_usecase4_retrieving_output Retrieving the Output from Thread B

After sending the images from Thread A, retrieve the output from Thread B:

```{.cpp}
    // Thread B:
    while(true)
    {
        dwImageStreamer_consumerReceive(&imageCuda, TIMEOUT, streamerCPU2CUDA);
        // CODE: Use the streamed image
        dwImageStreamer_consumerReturn(&imageCuda, streamerCPU2CUDA);
    }
```

@section image_usecase4_releasing_handle Releasing the Handle

```
dwImageStreamer_release(...);
```

To see a full demonstration of this workflow, please refer to @ref dwx_image_streamer_multi_sample.