Updated documentation to link to the current source code release.

README.md

@@ -1,5 +1,5 @@
 # LIBXSMM
-Library for small matrix-matrix multiplications targeting Intel Architecture (x86). The library generates code for the following instruction set extensions: Intel SSE3, Intel AVX, Intel AVX2, IMCI (KNCni) for Intel Xeon Phi coprocessors ("KNC"), and Intel AVX-512 as found in the Intel Xeon Phi processor family ("KNL") and future Intel Xeon processors. Historically the library was solely targeting the Intel Many Integrated Core Architecture "MIC") using intrinsic functions, however meanwhile optimized assembly code is generated for the fore mentioned instruction set extensions. [[pdf](https://github.com/hfp/libxsmm/raw/master/documentation/libxsmm.pdf)] [[src](https://github.com/hfp/libxsmm/archive/0.8.6.zip)]
+Library for small matrix-matrix multiplications targeting Intel Architecture (x86). The library generates code for the following instruction set extensions: Intel SSE3, Intel AVX, Intel AVX2, IMCI (KNCni) for Intel Xeon Phi coprocessors ("KNC"), and Intel AVX-512 as found in the Intel Xeon Phi processor family ("KNL") and future Intel Xeon processors. Historically the library was solely targeting the Intel Many Integrated Core Architecture "MIC") using intrinsic functions, however meanwhile optimized assembly code is generated for the fore mentioned instruction set extensions. [[pdf](https://github.com/hfp/libxsmm/raw/master/documentation/libxsmm.pdf)] [[src](https://github.com/hfp/libxsmm/archive/0.9.0.zip)]
 
 **What is a small matrix-matrix multiplication?** When characterizing the problem size using the M, N, and K parameters, a problem size suitable for LIBXSMM falls approximately within (M N K)^(1/3) <= 60 (which illustrates that non-square matrices or even "tall and skinny" shapes are covered as well). However the code generator only generates code up to the specified [threshold](#auto-dispatch). Raising the threshold may not only generate excessive amounts of code (due to unrolling), but also missing to implement a tiling scheme to effectively utilize the L1 cache. For problem sizes above the configurable threshold, LIBXSMM is falling back to BLAS.