minor text updates

Author: jrjohansson

Lecture-6B-HPC.ipynb

@@ -1,7 +1,7 @@
 {
  "metadata": {
   "name": "",
-  "signature": "sha256:57584155c2522841a684c3d37b8e0b8c40065febcc6846080ebe0132e7210f2e"
+  "signature": "sha256:7896fbc0a826538e5fd8e42d31783283070f4328cbbfa20fdd524b251121d762"
  },
  "nbformat": 3,
  "nbformat_minor": 0,
@@ -1220,7 +1220,7 @@
      "cell_type": "markdown",
      "metadata": {},
      "source": [
-      "Now, this implementation is much slower than numpy.dot for this problem size, because of overhead associated with OpenMP and thread, etc. But let's look at the how the different implementations scale with increased matrix size:"
+      "Now, this implementation is much slower than numpy.dot for this problem size, because of overhead associated with OpenMP and threading, etc. But let's look at the how the different implementations compare with larger matrix sizes:"
      ]
     },
     {
@@ -1332,7 +1332,7 @@
      "cell_type": "markdown",
      "metadata": {},
      "source": [
-      " Obviously one could do a better job more effort, since the theoretical limit of the speed-up should be:"
+      " Obviously one could do a better job with more effort, since the theoretical limit of the speed-up is:"
      ]
     },
     {