OpenFHE-WASM is the official web-assembly port of the OpenFHE library. OpenFHE-WASM currently supports a subset of BGV, BFV, and CKKS API available in the OpenFHE C++ version.

All versions of OpenFHE starting with v1.3.0 are supported.

Table of Contents

• Build instructions from source
  • Building the OpenFHE library with Emscripten
  • Running web-assembly unit tests
  • Running web-assembly benchmarks
  • Running automatically converted C++ examples
  • Building OpenFHE-WASM
  • Running OpenFHE-WASM Examples
• Notes specific to OpenFHE WebAssmebly

Build instructions from source

Building the OpenFHE library with Emscripten

1. Install emscripten using the instructions at https://emscripten.org/docs/getting_started/downloads.html. We tested v3.1.30 through 4.0.8.
2. Install NodeJs if not already installed. We suggest installing NodeJS 20 or later for best runtime results. Check the version using nodejs -v.
3. Clone OpenFHE-development
4. cd into the cloned directory and create embuild directory.
5. Run

export PREFIX=~/install/location
mkdir embuild
cd embuild
emcmake cmake .. -DCMAKE_INSTALL_PREFIX=${PREFIX}
emmake make -jN
emmake make install

Note:

• N is number of cores available on your system.
• ~/install/location can be any empty directory location where openfhe binaries should be installed.
• To include the unit tests, examples, or benchmarks, the corresponding cmake flags can be set to "ON" instead of "OFF".

Running web-assembly unit tests

Compile OpenFHE in the embuild directory using the following CMake flags

emcmake cmake .. -DBUILD_UNITTESTS=ON -DCMAKE_INSTALL_PREFIX=~/install/location

Run unit tests using nodejs:

nodejs unittest/binfhe_tests.js
nodejs unittest/core_tests.js
nodejs unittest/pke_tests.js

Running web-assembly benchmarks

Compile OpenFHE in the embuild directory using the following CMake flags

emcmake cmake .. -DBUILD_BENCHMARKS=ON -DCMAKE_INSTALL_PREFIX=~/install/location

Run benchmarks using nodejs, e.g.,

nodejs bin/benchmark/lib-benchmark.js

Running automatically converted C++ examples

Compile OpenFHE in the embuild directory using the following CMake flags

emcmake cmake .. -DBUILD_EXAMPLES=ON -DCMAKE_INSTALL_PREFIX=~/install/location

Run an exame using nodejs, e.g.,

nodejs bin/examples/pke/simple-integers

Building OpenFHE-WASM

1. Clone the openfhe-wasm repository and cd into it

2. Run the following commands to build the NodeJS bindings.

mkdir build
cd build
emcmake cmake .. -DOpenFHE_DIR=${PREFIX}/lib/OpenFHE
emmake make

This should install emscripten libraries in openfhe-wasm/lib directory.

Now run the examples in the following directories using nodejs

• examples/js/pke/

Running OpenFHE-WASM Examples

OpenFHE-WASM comes with the following examples, which are the JS versions of selected C++ PALISADE examples:

• pre_buffer.js: demonstrates use of OpenFHE for encryption, proxy re-encryption and decryption of packed vector of binary data
• simple_integer.js: simple example showing homomorphic additions, multiplications, and rotations for vectors of integers using BFV
• simple_integer_bgvrns.js: simple example showing homomorphic additions, multiplications, and rotations for vectors of integers using BGV
• simple_integer_serialization.js: simple example with serialization showing homomorphic additions, multiplications, and rotations for vectors of integers using BFV
• simple_real_number.js: simple example showing homomorphic additions, multiplications, and rotations for vectors of real numbers using CKKS
• threshold_fhe_bfv.js: example of threshold BFV

Notes specific to OpenFHE WebAssembly

• We have managed to compile OpenFHE-WASM using emscripten 3.1.30 through 4.0.8. A more recent version of nodejs (20 or later) should be used to achieve the best performance.
• The OpenFHE-WASM port is somewhat slower (typically 1.5 to 3.x depending on the operation) than the native C++ version of OpenFHE (in g++ or clang++) due to a normal slowdown incurred in web assembly builds (typically 2x) and additional slow-down due to the use of 64-bit arithmetic in PALISADE (64-bit arithmetic is emulated in WASM).
• Web assembly running environment is typically limited to 4GB of RAM.
• OpenFHE-WASM does not currently support multi-threading.