Getting Started with RTXMU SDK

Reducing Memory Consumption with an Open Source Solution

Download RTXMU SDK

System Requirements

v1.0
Operating System Windows 10 SDK v1809 (17763), or higher
Supported GPUsNVIDIA GeForce and Quadro products with Pascal and newer generation GPUs
Supported DriversLatest NVIDIA Display Driver
Development EnvironmentCMake 3.15, or higher Visual Studio 2017 or 2019

RTXMU SDK

RTXMU combines both compaction and suballocation techniques to optimize and reduce memory consumption of acceleration structures for any DXR or Vulkan Ray Tracing application.

Access GitHub

Release Notes

RTXMU SDK v1.0

  • Save Time - RTXMU reduces the time it takes for a developer to integrate compaction and suballocation into an RTX title.
  • Eliminate Wasted Memory - For applications using RTXMU, NVIDIA RTX cards get a ~50% reduction in memory footprint.
  • Prevent Fragmentation - Scenes with thousands of small unique BLAS benefit greatly from suballocation.
  • Open Source - To ensure we are able to support as many developers as possible, RTXMU will be made available as open source on GitHub.



FAQ

A: RTXMU (RTX Memory Utility) combines both compaction and suballocation techniques to optimize and reduce memory consumption of acceleration structures for any DXR or Vulkan Ray Tracing application.


A: RTXMU provides a simple interface that abstracts away the low level details of compaction and suballocation.


A: Works for both DXR 1.0, 1.1 along with Vulkan Ray Tracing. RTXMU SDK does not shoot any rays, it only manages the acceleration structures.


A: NVIDIA: GTX 1000, RTX 2000 and RTX 3000 series GPUs. AMD: Radeon RX 6000 series GPUs.




A: Compaction reduces the size of acceleration structures and suballocation allows smaller memory alignment requirements when packing multiple acceleration structures into a single allocation.


A: The open source initiative of RTXMU allows console developers to implement their own version. NVIDIA only provides a Vulkan and DX RT sample implementation.


A: Disabled compaction is the most likely culprit. All acceleration structures builds must have the Allow Compaction flag for RTXMU to prepare compaction workloads.




Resources