added notes from April 22

builtins.cpp

@@ -985,7 +985,7 @@ DefineStdlib(SymbolTable *symbolTable, llvm::LLVMContext *ctx, llvm::Module *mod
     case Target::NVPTX: 
       {
         if (runtime32) {
-            fprintf(stderr, "Unforetunatly 32bit targets are supported at the moment .. \n");
+            fprintf(stderr, "Unfortunatly 32bit targets are not supported at the moment .. \n");
             assert(0);
         }
         else {

ctx.cpp

@@ -1375,7 +1375,7 @@ FunctionEmitContext::None(llvm::Value *mask) {
 llvm::Value *
 FunctionEmitContext::LaneMask(llvm::Value *v) 
 {
-#if 1 /* this makes mandelbrot example slower, why ?!? */
+#if 1 /* this makes mandelbrot example slower with "nvptx" target. Need further investigation. */
   const char *__movmsk = g->target->getISA() == Target::NVPTX ? "__movmsk_ptx" : "__movmsk";
 #else
   const char *__movmsk = "__movmsk";