memMapIpc: Resolve build-time warnings and minor potential issues (#329)

* Fix compute performance calculation type casting in gpuGetMaxGflopsDeviceIdDRV() for #109

* 3_CUDA_Features/memMapIPCDrv: Increase procIdx buffer size to prevent potential buffer overflow

* memMapIPCDrv: Fix memory leaks and improve header inclusion

- Remove redundant string.h header
- Add memory cleanup for dynamically allocated JIT options and log buffer
- Fix printf format specifier for unsigned long long

Samples/3_CUDA_Features/memMapIPCDrv/memMapIpc.cpp

@@ -31,7 +31,6 @@
  */
 
 #include <stdio.h>
-#include <string.h>
 #include <cstring>
 #include <iostream>
 #include "cuda.h"
@@ -293,6 +292,11 @@ static void memMapGetDeviceFunction(char **argv) {
                                        jitNumOptions, jitOptions,
                                        (void **)jitOptVals));
     printf("> PTX JIT log:\n%s\n", jitLogBuffer);
+
+    // Clean up dynamically allocated memory
+    delete[] jitOptions;
+    delete[] jitOptVals;
+    delete[] jitLogBuffer;
   } else {
     checkCudaErrors(cuModuleLoad(&cuModule, module_path.c_str()));
   }
@@ -379,7 +383,7 @@ static void childProcess(int devId, int id, char **argv) {
     // deterministic.
     barrierWait(&shm->barrier, &shm->sense, (unsigned int)procCount);
     if (id == 0) {
-      printf("Step %lld done\n", (unsigned long long)i);
+      printf("Step %llu done\n", (unsigned long long)i);
     }
   }
 
@@ -550,7 +554,7 @@ static void parentProcess(char *app) {
   // Launch the child processes!
   for (i = 0; i < nprocesses; i++) {
     char devIdx[10];
-    char procIdx[10];
+    char procIdx[12];
     char *const args[] = {app, devIdx, procIdx, NULL};
     Process process;