Intel launches XeSS 3.0 SDK with multi frame generation and memory optimizations

Intel has released the XeSS 3.0 software development kit as a closed binary for Windows with a focus on multi frame generation and more efficient GPU memory use. The update lets developers boost frame rates and integrate XeSS more cleanly into existing engines.

Intel has released the official XeSS 3.0 software development kit, enabling game developers to integrate the latest XeSS 3.0 binaries into their titles and game engines. The new SDK arrives as a binary, pre compiled file instead of the long promised open source implementation, extending a four year period in which all XeSS versions have remained closed source under the Intel Simplified Software License as of the October 2022 revision. Distribution is limited to a DLL file for Windows operating systems, so Linux users cannot run this SDK natively and would need some form of translation layer. Developers updating older XeSS 2.x integrations are instructed to replace the existing libxess.dll, libxell.dll, and libxess_fg.dll files with those from the new XeSS 3.0 ZIP package.

Intel positions multi frame generation as the flagship capability of XeSS 3.0. The technology can integrate up to three generated frames between two rendered frames, resulting in up to a fourfold frame increase using MFG, which aligns conceptually with NVIDIA’s DLSS MFG approach. By inserting additional Artificial Intelligence generated frames, Intel is aligning XeSS 3.0 with a broader industry trend toward frame interpolation techniques aimed at higher perceived performance without a proportional increase in native rendering cost. The feature targets smoother gameplay on compatible hardware while leaning on the XeSS temporal reconstruction pipeline.

Alongside frame generation, XeSS 3.0 introduces support for external memory heaps, a change aimed at cleaner and more efficient engine integration. With this feature, the XeSS SDK can use GPU memory allocated directly by the game engine so that XeSS and the rendering pipeline can operate on the same VRAM blocks rather than reserving separate regions. This setup helps developers avoid duplicate buffers and memory fragmentation, while also providing more direct control over allocation and residency decisions. The result is a more streamlined path to adding XeSS 3.0 into existing render pipelines and potentially lower memory overhead in complex game engines.

52

Impact Score

Fujitsu debuts ‘Monaka’ Armv9 CPU sample with 3.5D packaging

Fujitsu has unveiled early silicon and an engineering sample of its ‘Monaka’ Armv9 CPU, built on TSMC’s 2 nm node and Broadcom’s 3.5D XDSiP packaging, ahead of a planned 2027 launch. The 144 core design targets Artificial Intelligence inference, simulation, and large scale data processing workloads.

ABB and NVIDIA bring industrial-grade physical artificial intelligence to factory robotics

ABB Robotics is integrating NVIDIA Omniverse into its RobotStudio suite to deliver physically accurate simulation for industrial robots, aiming to close the long-standing sim-to-real gap and cut deployment time and cost. Early pilots with Foxconn and Workr highlight how synthetic data and unified workflows could accelerate automation in complex manufacturing environments.

Artificial intelligence dashboards turn Iran conflict into real-time spectacle

New wartime intelligence dashboards built with artificial intelligence tools promise real-time insight into the Iran conflict but risk turning it into entertainment and spreading confusion. Open-source feeds, prediction markets, and synthetic imagery are colliding to create more noise than understanding.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.