update text in parallel-computing.rst

content/parallel-computing.rst

@@ -23,9 +23,7 @@
 The performance of a single CPU core has stagnated over the last ten years
 and most of the speed-up in modern CPUs is coming from using multiple
 CPU cores, i.e. parallel processing. Parallel processing is normally based
-either on multiple threads or multiple processes. 
-
-There are two main models of parallel computing:
+either on multiple threads or multiple processes: 
 
 - **Shared memory parallelism (multithreading):** 
 
@@ -47,7 +45,7 @@
    **"Embarrassingly" parallel**: If you can run multiple instances of a program and do not need to synchronize/communicate with other instances, 
    i.e. the problem at hand can be easily decomposed into independent tasks or datasets and there is no need to control access to shared resources, 
    it is known as an embarrassingly parallel program. A few examples are listed here:
      - Monte Carlo analysis
      - Ensemble calculations of numerical weather prediction
      - Discrete Fourier transform 
      - Convolutional neural networks
@@ -62,7 +60,7 @@
 
 In the Python world, it is common to see the word `concurrency` denoting any type of simultaneous 
 processing, including *threads*, *tasks* and *processes*. 
   - Concurrent tasks can be executed in any order but with the same final results
   - Concurrent tasks can be but need not to be executed in parallel
   - ``concurrent.futures`` module provides implementation of thread and process-based executors for managing resources pools for running concurrent tasks
   - Concurrency is difficult: Race condition and Deadlock may arise in concurrent programs