Merge pull request #14061 from JosephTremoulet/BenchmarkReadme

Update BenchmarksGames README.txt

Author: JosephTremoulet

tests/src/JIT/Performance/CodeQuality/BenchmarksGame/README.TXT

@@ -5,9 +5,13 @@ The benchmarks in these sub-directories are based on
 
 See the adjoining LICENSE.TXT file for license terms.
 
-Our intention with these tests is to provide interesting test cases
-for jit developers to use in daily development practice -- not to
-produce variants that give the maximum possible performance.
+Our intention with these tests is twofold:
+ 1 - To track .NET Core's performance on these benchmarks in the
+     same benchmarking system used for other internal .NET Core
+     performance benchmarks.
+ 2 - To make these available for daily JIT (and runtime) development,
+     as a factor in assessing the performance impact of compiler
+     (and runtime) changes.
 
 The benchmarks have been modified to fit into the CoreCLR test and
 performance test framework, as follows:
@@ -19,19 +23,15 @@ performance test framework, as follows:
    xunit-performance iteration is approximately 1 second on modern x64
    hardware
  - reducing verbosity when run as a benchmark
- - reformatting (via the codeformatter tool)
+ - calling different APIs in a few places to allow compiling against
+   netstandard1.4
+ - reformatting
+ - in the case of pidigits, implementing on top of .NET's BigInteger
+   type rather than p/invokes to the native GMP library
 
 These benchmarks are just a subset of the benchmarks available in C# from
-the Benchmarks Game site. We've selected variants that do not rely on
-multiple threads to ensure relative benchmark stability across a
-variety of machines. 
-
-We've excluded two benchmarks that are inherently multitheaded:
-chamenosredux and threadring. We may revisit this as we improve our
-ability to harness threading tests in a stable way.
-
-We've also excluded benchmarks that read in large input files:
-knucleotide, regexdna, revcomp.
-
-
-
+the Benchmarks Game site. The highest-scoring C# .NET Core variant of each
+benchmark is included, and in the (common) case of benchmarks where the
+best-scoring variant uses multiple threads, we've also selected variants
+that do not rely on multiple threads, to ensure relative benchmark stability
+across a variety of machines.