If you've never heard of Zstd before, don't feel bad; most haven't. Developed at Facebook (now Meta), Zstd is a modern compression algorithm designed to strike a balance between compression ratio and decompression speed. It's already widely used in places like Linux distributions and cloud infrastructure because it compresses data well but also decompresses extremely quickly, which is a very important trait when you're trying to stream gigabytes of textures and geometry off an SSD while a game is running.
So, DirectStorage 1.4 adds native support for Zstd-compressed assets on both CPU and GPU paths, allowing developers to decompress data either on the CPU or through GPU compute shaders. The update does not replace or deprecate Microsoft's earlier GDeflate compression format, which debuted alongside GPU decompression in DirectStorage 1.1. Instead, Zstd simply becomes another supported option for developers building DirectStorage pipelines.
You may like Shader Execution Reordering part of new DirectX update boosts FPS by 90% in Intel Arc B-series GPUs Nvidia introduces DLSS 4.5 and Multi Frame Generation 6X at CES 2026 AMD FSR Redstone press Q&A roundtable CES 2026 transcript DirectStorage itself has had a somewhat underwhelming run so far. Despite early demos promising dramatic improvements to loading and asset streaming, only a handful of PC games have meaningfully adopted the technology, and in practice, DirectStorage sometimes improves load times, but it can actually cost a bit of frame rate. The reason is simple: GPU decompression isn't free . When the GPU is spending time unpacking texture data, that's compute time and power budget it can't spend on rendering frames. Since GPU time is usually the most precious resource in a modern game, the benefits of DirectStorage depend heavily on how much spare GPU compute capacity a system has.
Another technical change in the update may help with that. DirectStorage 1.4 adds global D3D12 CreatorID support to the EnqueueRequests function. CreatorIDs allow different GPU workloads to identify themselves to the driver, enabling smarter scheduling decisions when multiple queues are competing for GPU resources. In practical terms, this could help drivers better coordinate DirectStorage decompression work with traditional rendering and compute tasks, reducing the risk that asset streaming workloads interfere with frame rendering.
Arguably, the more interesting addition, though, is GACL, the Game Asset Conditioning Library. GACL is an attempt to standardize the asset pipeline around DirectStorage so developers can ship more compact game data that still streams quickly. Specifically, it's a build-time tool that preprocesses game assets before they're compressed.
By reorganizing and transforming asset data using techniques like data shuffling and entropy reduction, it can significantly improve how well the data compresses with Zstd. Microsoft says the conditioning step can dramatically improve compression efficiency for certain asset types while keeping decompression costs low. DirectStorage then reverses those transformations automatically when the data is loaded.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/video-games/pc-gaming/SPONSORED_LINK_URL
- https://www.tomshardware.com/video-games/pc-gaming/microsoft-debuts-directstorage-1-4-at-gdc-2026-with-zstandard-compression-and-gacl-update-promises-developers-improved-compression-ratios-faster-loading-and-more#main
- https://www.tomshardware.com
- NVIDIA and ComfyUI Streamline Local AI Video Generation for Game Developers and Creators at GDC
- ROG Xbox Ally X gaming handheld to get Auto Super Resolution boost in April — Microsoft touts 30% performance boost thanks to AI-powered image upscaling
- How to use Precision Boost Overdrive and Curve Optimizer to improve Ryzen CPU performance
- IBM and Lam Research team up on High NA EUV dry resist to push chip scaling past 1nm
- IBM and Lam's new partnership paves the way toward sub-1nm logic using High-NA EUV — Albany lab to pioneer dry resist process integration
Informational only. No financial advice. Do your own research.