jackshendrikov / parallel-calc

Parallel and Distributed Calculations Labs

Parallel and Distributed Calculations

Lab 1

Semaphores, mutexes, events, critical sections in WinAPI

Task:

1. Develop a parallel algorithm for solving a mathematical problem MU = MD * MC * d + max (Z) * MR using WinAPI library in C++;
2. Identify shared resources;
3. Describe the algorithm of each thread (T1 - Ti) with the definition of critical areas and synchronization points (Wij, Sij);
4. Develop the structural scheme of interaction of threads where to apply all specified means of interaction of processes;
5. Develop a program;
6. Perform program debugging;
7. Get the correct calculation results;
8. Use Windows Task Manager to monitor CPU kernel load.

Means of organizing interaction: semaphores, mutexes, critical sections, events;

Means of interaction: semaphores.

Block diagram of the interaction of threads

Symbols in the block diagram:

• CS - to access the shared resource d, m;
• M - to access the shared resource MC;
• E1 - for synchronization with the completion of input in T1;
• E3 - for synchronization with the completion of input in T3;
• E4 - for synchronization with the completion of input in T4;
• Sm1, Sm2, Sm3, Smax - to synchronize the calculations of the maximum Z;
• S.MA1, S.MA2, S.MA3 - to synchronize the rest of the calculations and output the result.

Lab 2

Semaphores, mutexes, events, critical sections in WinAPI

Task:

1. Develop a parallel algorithm for solving a mathematical problem Z = sort(D * (ME * MM)) + (B * C) * E * x in C#;
2. Identify shared resources;
3. Describe the algorithm of each thread (T1 - Ti) with the definition of critical areas and synchronization points (Wij, Sij);
4. Develop the structural scheme of interaction of threads where to apply all specified means of interaction of processes;
5. Develop a program;
6. Perform program debugging;
7. Get the correct calculation results;
8. Use Windows Task Manager to monitor CPU kernel load.

Problem: Z = sort (D * (ME * MM)) + (B * C) * E * x;

Programming language: C#;

Means of organizing interaction: semaphores, mutexes, events, critical sections, atomic variables (types);

Block diagram of the interaction of threads

Symbols in the block diagram:

• M - mutex for access to the shared resource b;
• volatile - keyword to access the shared resource x;
• Lock - lock to access the shared resource D;
• ME - semaphore for access to the shared resource ME;
  
  
• E0 - event for synchronization with the completion of input in T1;
• E1 - event for synchronization with the completion of input in T2;
• E2 - event for synchronization with the completion of input in T3;
• E3 - event for synchronization with the completion of input in T4;
  
  
• S0 - semaphore for synchronization with the completion of the merger K in the thread T1;
• S1 - semaphore to synchronize the completion of calculations ZH in the thread T2;
• S2 - semaphore to synchronize the completion of calculations ZH in the thread T3;
• S3 - semaphore to synchronize the completion of calculations ZH in the thread T4;
  
  
• SM0 - semaphore to synchronize the completion of the K2H merger in the thread T2;
• SM1 - semaphore to synchronize the completion of sorting KH in the thread T3;
• SM2 - semaphore to synchronize the completion of sorting KH in the thread T4.

Lab 3

Java. Monitors

Task:

1. Develop a parallel algorithm for solving a mathematical problem E = (B * MR) * (MM * MO) + min(B) * Q * d using monitors in Java;
2. Identify shared resources;
3. Describe the algorithm of each thread (T1 - Ti) with the definition of critical areas and synchronization points (Wij, Sij);
4. Develop the structural scheme of interaction of threads where to apply all specified means of interaction of processes;
5. Develop a program;
6. Perform program debugging;
7. Get the correct calculation results;
8. Use Windows Task Manager to monitor CPU kernel load.

Problem: E = (B * MR) * (MM * MO) + min(B) * Q * d;

Programming language: Java;

Means of organizing interaction: Java monitors, synchronized blocks;

Block diagram of the interaction of threads

Symbols in the block diagram:

• InputSignal - signal about the completion of input in threads T1, T2, T3;
• WaitForInput - waiting for input completion signals in threads T1, T2, T3;
  
  
• Signal - signal about the end of calculation E in threads T2, T3, T4;
• WaitForSignal - waiting for signals to complete the calculation of E in threads T2, T2, T3;
  
  
• SignalCalcM - signal about the completion of complete the calculation of m in threads T1, T2, T3, T4;
• WaitForCalcM - waiting for signals to complete the calculation of m in threads T1, T2, T3, T4;
  
  
• SignalCalcA - signal about the completion of complete the calculation of Aн in threads T1, T2, T3, T4;
• WaitForCalcA - waiting for signals to complete the calculation of Aн in threads T1, T2, T3, T4;
  
  
• copyM - copying a shared resource m by threads T1, T2, T3, T4;
• calcM - calculation of m = min[m, m(i)] by threads T1, T2, T3, T4;
  
  
• copyA - copying a shared resource A by threads T1, T2, T3, T4;
• calcA - calculation Aн;
  
  
• copyD - copying a shared resource d by threads T1, T2, T3, T4;
• setD - input of the shared resource d by thread T3;
  
  
• copyB - copying a shared resource B by threads T1, T2, T3, T4;
• setB - input of the shared resource B by thread T2;
  
  
• copyMO - copying a shared resource MO by threads T1, T2, T3, T4;
• setMO - input of the shared resource MO by thread T3;

Lab 4

OpenMP. Barriers, critical sections.

Task:

1. Develop a parallel algorithm for solving a mathematical problem Z = (B * C) * D + E * (MA * MB) * x using the OpenMP library in C ++;
2. Identify shared resources;
3. Describe the algorithm of each thread (T1 - Ti) with the definition of critical areas and barriers; (
4. Develop the structural scheme of interaction of threads where to apply all specified means of interaction of processes;
5. Develop a program;
6. Perform program debugging;
7. Get the correct calculation results;
8. Use Windows Task Manager to monitor CPU kernel load.

Problem: Z = (B * C) * D + E * (MA * MB) * x;

Programming language: C++;

Means of organizing interaction: barriers, locks, critical sections of OpenMP;;

Block diagram of the interaction of threads

Lab 5

ADA. Rendezvous.

Task:

1. Develop a parallel algorithm for solving a mathematical problem Z = (B * C) * D + E * (MA * MB) * x using the rendezvous mechanism in ADA;
2. Describe the algorithm of each thread;
3. Develop the structural scheme of interaction of threads where to apply all specified means of interaction of processes;
4. Develop a program, Perform program debugging, Get the correct calculation results;

Problem: Z = (B * C) * D + E * (MA * MB) * x;

Programming language: ADA;

Means of organizing interaction: rendezvous;

The structure of PKS LP: linear.

Block diagram of the interaction of threads