Linear Algebra and Matrices Flashcards

(15 cards)

1
Q

Why is linear algebra important to HPC?

A

Many scientific and engineering applications use more advanced numerical methods which require solving system of linear equations.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Which method is used for complex meshes?

A

Finite Volume methods

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Which additional consideration does the finite volume method have?

A

It considers fluxes, the flow of mass, momentum, and energy, between adjacent cells.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

How does the finite element method approximate the solution?

A

It uses expansion functions where are non-zero over a small region of the domain.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What does BLAS stand for?

A

Basic Linear Algebra Subprograms

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What does BLAS do?

A

BLAS defines a standardised interface for basic vector and matrix operations.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What are the three levels of BLAS routines?

A
  1. Level 1: vector-vector
  2. Level 2: matrix-vector
  3. Level3 : matrix-matrix
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What are the arithmetic intensities of the BLAS levels?

A

Level 1 and 2 have O(1). Level 3 has O(N)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

How does Compressed Sparse Row store a matrix?

A

The matrix is stored using three one-dimensional arrays.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Why is CSR used?

A

It can significantly reduce the side of a sparse matrix by not storing any unnecessary elements.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Which three linear algebra benchmarks are used to asses a systems speed?

A
  1. Linpack
  2. High Performance Linpack
  3. High Performance Conjugate Gradients
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

How does Linpack assess performance?

A

It measures Flops with a dense linear algebra workload, asking the system to solve Ax = b using a lower-upper factorisation.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How does HPL assess performance?

A

It acts the same as Linpack but can change the size of matrices so that floating point performance is close to the peak value.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What does HPL measure?

A

It measures actual HPL performance which is usually a significant fraction of the theoretical maximum performance.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

How does HPCG assess performance?

A

HPCG is representative of sparse algebra workloads having the system solve a 3D Poisson’s equation using a 27-point stencil.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly