Benchmarks

This document details the performance benchmarks of homomorphic operations using TFHE-rs.

By their nature, homomorphic operations run slower than their cleartext equivalents. The following are the timings for basic operations, including benchmarks from other libraries for comparison.

Integer operations

The following tables benchmark the execution time of some operation sets using FheUint (unsigned integers). The FheInt (signed integers) performs similarly.

The next table shows the operation timings on CPU when all inputs are encrypted:

The next table shows the operation timings on CPU when the left input is encrypted and the right is a clear scalar of the same size:

All timings are based on parallelized Radix-based integer operations where each block is encrypted using the default parameters PARAM_MESSAGE_2_CARRY_2_KS_PBS. To ensure predictable timings, we perform operations in the default mode, which propagates the carry bit as needed. You can minimize operational costs by selecting from 'unchecked', 'checked', or 'smart' modes, each balancing performance and security differently.

For more details about parameters, see here. You can find the benchmark results on GPU for all these operations here.

Shortint operations

The next table shows the execution time of some operations using various parameter sets of tfhe-rs::shortint. Except for unchecked_add, we perform all the operations in the default mode. This mode ensures predictable timings along the entire circuit by clearing the carry space after each operation. The configuration is Concrete FFT + AVX-512.

Boolean operations

The next table shows the execution time of a single binary Boolean gate.

tfhe-rs::boolean

tfhe-lib

Using the same hpc7a.96xlarge machine as the one for tfhe-rs, the timings are as follows:

OpenFHE (v1.1.2)

Following the official instructions from OpenFHE, we use clang14 and the following command to setup the project: cmake -DNATIVE_SIZE=32 -DWITH_NATIVEOPT=ON -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DWITH_OPENMP=OFF ..

The following example shows how to initialize the configuration to use the HEXL library:

export CXX=clang++
export CC=clang

scripts/configure.sh
Release -> y
hexl -> y

scripts/build-openfhe-development-hexl.sh

Using the same hpc7a.96xlarge machine as the one for tfhe-rs, the timings are as follows:

Reproducing TFHE-rs benchmarks

TFHE-rs benchmarks can be easily reproduced from the source.

The following example shows how to reproduce TFHE-rs benchmarks:

#Boolean benchmarks:
make bench_boolean

#Integer benchmarks:
make bench_integer

#Shortint benchmarks:
make bench_shortint