JPEG Image Compression SDK

The technology that lies behind JPEG Image Compression on GPU allows us to do JPEG compression and decompression extremely fast. That solution is based on parallel and highly optimized implementation of JPEG Baseline algorithm for NVIDIA GPUs. The new technology enables consumers and businesses to compress and decompress JPEG images much faster than ever. This is unigue approach for high speed and heavy-loaded applications to handle thousands of images per second.

We have created the fastest JPEG encoder and decoder based on NVIDIA CUDA technology. On NVIDIA GeForce GTX 1080 and for JPEG quality 90% we achieved compression rate more than 30 GByte/s (without host-to-device and device-to-host transfers) due to highly optimized parallel implementation of Baseline JPEG algorithm. Our JPEG compression software is much more fast in comparison with the best commercial multithreaded JPEG codecs on multicore CPUs.

We released JPEG Compression SDK to give our customers opportunity to use the power of NVIDIA GPUs in various applications.

JPEG SDK Breakthrough FeaturesJPEG logo

  • It's increadibly fast
  • Baseline JPEG Compression and Decompression on NVIDIA GPU
  • Highly optimized parallel implementation of Baseline JPEG algorithm

Software Features

  • Implementation is 100% compliant with JPEG Baseline standard
  • Baseline JPEG compression and decompression for grayscale (8-bit) and color (24-bit) images with arbitrary width and height
  • 12-bit JPEG compression on GPU
  • Extremely fast lossy image encoding and decoding with variable compression ratio
  • Subsampling modes: 4:4:4, 4:2:2, 4:2:0
  • Minimum input image size 1×1 for grayscale and color images with any subsampling
  • Maximum input image size is 12000×12000 or even more (optional)
  • JPEG image quality in the range from 1 to 100%
  • Read/write any EXIF section
  • Data input: 8/24-bit images from RAM/HDD/RAID/SSD
  • Data output: final compressed/uncompressed image in RAM/HDD/RAID/SSD
  • Continuous data mode (input one image after another)
  • Standard set of computations for parallel implementation of Baseline JPEG encoding and decoding
    • JPEG Encoding: Input data parcing, Color Transform, 2D DCT, Quantization, Zig-zag, AC/DC, DPCM, RLE, Huffman coding, Byte stuffing, JFIF formatting
    • JPEG Decoding: JFIF parcing, Restart marker search, Inverse Huffman Decoding, Inverse RLE, Inverse DPCM, AC/DC, Inverse Zig-zag, Inverse Quantization, IDCT, Inverse Color Transform, Output formatting
  • Optimized for the latest NVIDIA GPUs (Fermi, Kepler, Maxwell and Pascal)
  • Compatibility with FFmpeg to read/write MJPEG streams
  • Optional integration with OpenGL
  • Compatible with Windows-7/8/10 and Linux (32/64)

We have succeeded to make parallel all stages of JPEG algorithm including entropy encoding and decoding. There was a widespread opinion that Huffman algorithm could be only serial. In our JPEG SDK, Huffman coding is not a bottleneck anymore and it's fully parallel. Now we don't offload anything from GPU to CPU to make JPEG codec faster. JPEG codec is extremely fast and is does everything completely on GPU.

Additional Info


  • Full technical support up to successful client integration
  • SDK, documentation

For any further information concerning JPEG Image Compression SDK or free trial please contact us via email.

     Home                   Contacts                 Site Map
GPU Image Processing