You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+3-3Lines changed: 3 additions & 3 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -65,12 +65,12 @@ The benchmark case is the 2D taylor Green, the time reported here are intended f
65
65
66
66
### Strong Parallelization
67
67
Strong scalability evaluates how efficiently a parallel code reduces execution time when the problem size remains fixed, but the number of processing units increases. There is a total of `400` elements on each side, leading to `160000` elements and `1920000` dofs in total.
68
-
{ width=50% }
68
+

69
69
70
70
71
71
### Weak Parallelization
72
-
Weak scalability measures how well a parallel code maintains performance when the problem size is kept constant per processor, and the number of processors increases. On each processor there are `50x50` elements, the number of dofs is kept constant at ``\approx```30K dfos/procs`.
73
-
{ width=50% }
72
+
Weak scalability measures how well a parallel code maintains performance when the problem size is kept constant per processor, and the number of processors increases. On each processor there are `50x50` elements, the number of dofs is kept constant at `30K dfos/procs`.
Copy file name to clipboardExpand all lines: docs/src/Cases/taylorgreen3D.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -79,7 +79,7 @@ at each time-step, a new line on a `.csv` file will be written with: timestep-Ki
79
79
The Linearized and Segregated VariationalMultiScale method has been validated with test case. The time-step is constant ``\Delta t =0.01s``, the matrices are re-computed every 5 time-steps. The Reynolds number is ``1600``, it is used to compute the viscosity: ``\nu = U_0 k / Re``. Order of interpolation of velocity and pressure is the same. The benchmark is the DNS result. The flow patterns generated by the Taylor-Green vortex are quite complex. The resolution of the most complex case tested: 64x64x64 Q2Q2, takes 40 hours on a 64CORE machine.
80
80
81
81
We note that the kinetic energy decay does not match that of the reference for Q1 elements, there is significant numerical dissipation since the enstrophy does not match the kinetic energy decay.
82
-

82
+

83
83
84
84
Increasing the order from Q1 to Q2 yield the following results which are better.
0 commit comments