파일
rocm-systems/samples/jpegDecode/README.md
T
Aryan Salmanpour bb085a9bf1 Add a new sample (jpegDecodeMultiThreads) for jpeg decoding using multiple threads (#18)
* Add a new sample for jpeg decoding using multiple threads

* code clean up

* code clean up

* code cleanup

* remove extra line

* code clean up - change some variable names

* code clean up

* Move common functions to a new header for samples

* move additional functions to the common samples header

* move the common functions to a new header file and modify the readme

* modify the sample's README

* Add a CTEST for the jpegDecodeThreads sample

* Add a samples overview README

* modify the jpeg decode threads sample

* add finding the threads

* rename jpegDecodeThreads to jpegDecodeMultiThreads

* Make changes based on the reviewers comments

* use one instance of the rocjpeg_utils

* code cleanup
2024-05-03 12:16:43 -04:00

1.4 KiB

JPEG decode sample

The jpeg decode sample illustrates decoding a JPEG images using rocJPEG library to get the individual decoded images in one of the supported output format (i.e., native, yuv, y, rgb, rgb_planar). This sample can be configured with a device ID and optionally able to dump the output to a file.

Prerequisites:

Build

mkdir jpeg_decode_sample && cd jpeg_decode_sample
cmake ../
make -j

Run

./jpegdecode -i     <[input path] - input path to a single JPEG image or a directory containing JPEG images - [required]>
             -be    <[backend] - select rocJPEG backend (0 for hardware-accelerated JPEG decoding using VCN,
                                                         1 for hybrid JPEG decoding using CPU and GPU HIP kernels (currently not supported)) [optional - default: 0]>
             -fmt   <[output format] - select rocJPEG output format for decoding, one of the [native, yuv, y, rgb, rgb_planar] [optional - default: native]>
             -o     <[output path] - path to an output file or a path to a directory - write decoded images to a file or directory based on selected output format [optional]>
             -d     <[device id] - specify the GPU device id for the desired device (use 0 for the first device, 1 for the second device, and so on); [optional - default: 0]>