Hi guys, i’m a student currently researching on a particle swarm optimization algorithm for exam timetabling and was wondering if I should try cuda. I’m currently still reading through the programming guide but one main issue that I foresee is how best to allocate memory or what data structures are best to store the information from my sql database. I’m most worried about how ‘coalesced’ the data access will be and how it will affect performance
For context here are some relevant parts of tables that I have
Table: STUDENT {
matriculation number;
name;
course;
}
Table 2: Exam {
Exam Code;
Title;
number of students;
Duration;
}
Table 3: Venue {
Location ID;
Name;
Capacity;
}
Table 4: Schedule {
Exam code;
Time slot;
Venue;
}
Table 5: Enrollment {
matriculation number;
Exam code;
}
Here are some example constraints:
- Ensure that a student is not taking 2 exams at the same time
- Ensure that a student is not taking more than a certain number of exams a day
- Ensure that all the exams in each time slot don’t exceed the capacity of rooms available
As can be seen the data is relational and the constraint calculations for each particle while straightforward, draws data from all over the place. Hence it would be great if someone can get me started on how best to go about storing this data and reading it.
My specs are:
i7-6700k
32gb Ram
RTX 2080ti