AMD Unveils DGFS Super Compression Technology to Further Optimize Storage Efficiency

On May 8, AMD officially launched the DGF Super Compression Technology, named DGFS, through the GPU Open platform. This technology is designed to further compress DGF data, reduce storage costs, while maintaining compatibility with non-DGF hardware.

Adopting DGFS for asset packages has two key advantages over storing DGF blocks directly. Compressed file size is reduced by up to 22% while DGF and non-DGF devices can be targeted using a common asset format.

According to AMD's test data, in terms of raw storage usage, DGFS saves an additional 17% to 31% of disk space compared to the standard DGF format across five test models: Crab, Dragon, Little Statue, Buddha, and Bike. In gaming scenarios, DGFS data is not persistently stored in memory. When GDeflate compression is applied to the data, DGFS is roughly 20% smaller than DGF.

In terms of decoding speed, a single CPU core can decode large-scale models in a very short time. Test results show that CPU-based streaming decoding is fast enough. AMD notes that a GPU-based decoder is also feasible, and the vectorized version of the DGFS decoder delivers good performance.