Tools for people envious of nvidia's blob driver.
C C++ Python Perl CMake Yacc Other


envytools - Tools for people envious of nvidia's blob driver.

The canonical repo is at: Pushing anywhere else will result in a de-facto fork!



  • docs: plain-text documentation of the GPUs, nVidia binary driver, and the tools (in-sync HTML version at
  • envydis: Disassembler and assembler for various ISAs found on nvidia GPUs
  • rnn: Tools and libraries for the rules-ng-ng XML register database format
  • rnndb: rnn database of nvidia MMIO registers, FIFO methods, and memory structures.
  • nvbios: Tools to decode the card description structures found in nvidia VBIOS
  • nva: Tools to directly access the GPU registers
  • vstream: Tools to decode and encode raw video bitstreams
  • vdpow: A tool aiding in VP3 reverse engineering
  • easm: Utility code dealing with assembly language parsing & printing.
  • util: Misc utility code shared between envytools modules

Building, installing


  • cmake
  • libxml2
  • flex
  • bison
  • pkg-config

Optional dependencies needed by nva:

  • libpciaccess

Optional dependencies needed by vdpow:

  • vdpau
  • libx11

Optional dependencies needed for ninja build (faster recompile times):

  • ninja

If your distribution has -dev or -devel packages, you'll also need ones corresponding to the dependencies above.

On ubuntu it can be done like this:

apt-get install cmake flex libpciaccess-dev bison libx11-dev libxext-dev libxml2-dev libvdpau-dev python3-dev

To build using ninja (recommended if you work on envytools):

$ cmake . -G Ninja
$ ninja

If you prefer to stay with make, use

$ cmake .
$ make

To install [which is optional], use

$ ninja install


$ make install

If you want to install to a non-default directory, you'll also need to pass it as an option to cmake before building, eg.:

$ cmake -D CMAKE_INSTALL_PREFIX:PATH=/usr/local .

Cmake options

If you don't want to compile some parts of envytools, you can pass the following options to cmake: