diff --git a/README.txt b/README.txt
new file mode 100644
index 0000000..90a8d94
--- /dev/null
+++ b/README.txt
@@ -0,0 +1,126 @@
+
+  Pseudo-Random Number Generators in VHDL
+ =========================================
+
+This library contains a number of pseudo-random number generators (PRNG)
+in synthesizable VHDL code.
+
+The PRNGs in this library are useful to generate noise in digital signal
+processing and to generate random numbers for monte-carlo simulations or
+for games.
+
+The VHDL code is portable and should run on any FPGA/synthesis platform,
+but the designs are somewhat optimized for Xilinx FPGAs.
+
+These PRNGs are a good alternative to linear feedback shift registers (LFSR).
+Although LFSRs are commonly used, their output exhibits strong correlations.
+Furthermore, correctly generating multi-bit random words with LFSRs is tricky.
+
+NOTE: None of the RNGs in this package are cryptographic random number
+      generators. These generators are not suitable for cryptography.
+
+
+  Xoroshiro128+ RNG
+  ------------------
+
+Xoroshiro128+ is an RNG algorithm developed in 2016 by David Blackman
+and Sebastiano Vigna. It is a further development of the Xorshift algorithm
+invented by George Marsaglia
+
+See also http://xoroshiro.di.unimi.it/
+
+This RNG passes many statistical tests. Its period, 2**128 - 1, is long
+compared to a typical LFSR, but much shorter than the Mersenne Twister.
+The output is 1-dimensionally equidistributed.
+
+The VHDL implementation produces 64 new random bits on every (enabled)
+clock cycle. It is quite efficient in terms of FPGA resources, but it
+does require a 64-bit adder.
+
+Output word length: 64 bits
+Seed length:        128 bits
+Period:             2**128 - 1
+
+FPGA resources:     general logic and 64-bit adder
+Sythesis results:   194 LUTs, 192 registers on Spartan-6
+Timing results:     333 MHz on Spartan-6 LX45-3
+
+
+  Mersenne Twister RNG
+  --------------------
+
+The Mersenne Twister is an RNG algorithm developed in 1997 by
+Makoto Matsumoto and Takuji Nishimura. This library implements
+the most common variant of the algorithm, MT19937.
+
+See also M. Matsumoto, T. Nishimura, "Mersenne Twister: a 623-dimensionally
+equidistributed uniform pseudorandom number generator", ACM TOMACS, vol. 8,
+no. 1, 1998.
+
+This RNG is very popular in software applications because it is relatively
+fast while the output passes many statistical tests.
+
+The VHDL implementation produces 32 new random bits on every (enabled)
+clock cycle. It uses a RAM block, 32 bits wide, 1024 elements deep,
+to store the RNG state.
+
+The most demanding part of the implementation is the seeding procedure.
+Seeding involves repeated multiplication by a 32-bit constant. This
+multiplication is implemented as a hand-coded series of shifts and adds
+when force_const_mul = true; the multiplication is left to the synthesizer
+when force_const_mul = false. The hand-coded variant is much more efficient
+on Xilinx Spartan-6, so the recommended setting is force_const_mul = true.
+
+After reset and after each reseeding, the RNG needs 2496 clock cycles
+to initialize its state. The RNG can not provide random data during this time.
+
+Output word length: 32 bits
+Seed length:        32 bits
+Period:             2**19937 - 1
+
+FPGA resources:     RAM block, 32 bits x 1024 elements
+Sythesis results:   279 LUTs, 297 registers, 2x RAMB16 on Spartan-6
+Timing results:     300 MHz on Spartan-6 LX45-3
+
+
+  Code organization
+  -----------------
+
+ rtl/                           Synthesizable VHDL code
+ rtl/rng_xoroshiro128plus.vhdl  Implementation of Xoroshiro128+ RNG
+ rtl/rng_mt19937.vhdl           Implementation of Mersenne Twister RNG
+
+ sim/                           Test benches
+ sim/Makefile                   Makefile for building test benches with GHDL
+ sim/tb_xoroshiro128plus.vhdl   Test bench for Xoroshiro128+ RNG
+ sim/tb_mt19937.vhdl            Test bench for Mersenne Twister RNG
+
+ refimpl/                       Reference software implementations of RNGs
+
+ synth/                         Top-level wrappers for synthesis testruns
+
+
+  License
+  -------
+
+Copyright (C) 2016 Joris van Rantwijk
+
+This VHDL library is free software; you can redistribute it and/or
+modify it under the terms of the GNU Lesser General Public License
+s published by the Free Software Foundation; either version 2.1
+of the License, or (at your option) any later version.
+
+This library is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+Lesser General Public License for more details.
+
+You should have received a copy of the GNU Lesser General Public
+License along with this library; if not, see
+  <https://www.gnu.org/licenses/old-licenses/lgpl-2.1.html>
+
+This package contains a few support files which are distributed
+under Creative Commons CC0. This is explicitly and clearly marked
+in the files to which it applies.
+
+--