High resolution performance results

24 processors, SHC

  • 5 levels: base level 180x40 cells with 4 more levels refined 2,2,2,4 times respectively, x=(-.005, .085), y=(0, 0.02)

Method whole time integration time time steps cost/timestep (sec)
Roe- van Albada: 2x Refined, levels refined 2,2,4,4 15085 5079 444 11.4
Weno-TCD-bandwidth 4th order: pressure only dcflag, pcurv=1e6 14349 5551 430 12.9
Weno-TCD-bandwidth 2nd order: pressure only dcflag, pcurv=5e6 10895 4232 468 9.0
Weno-TCD-bandwidth 4th order: pressure only dcflag, pcurv=5e6 29595 6706 498 13.5
Weno-TCD-bandwidth 4th order: pressure only dcflag, pcurv=1e7 29857 12081 639 18.9

64 processors, SHC

  • 5 levels: base level 100x40 cells with 4 more levels refined 2,2,4,4 times respectively, x = [-0.005,0.095], y = [0, 0.04]
  • The "normalized cost/timestep" is adjusted for the differing base grid resolutions

Method whole time integration time time steps cost/timestep (sec) normalized cost/timestep
Roe- Min-Mod 4118 448 212 2.1 1.1
Roe- van Albada 4364 490 216 2.3 1.15
Roe- van Albada 2x refined 180x40 base grid (-.005, .085)(0, 0.02) 10246 1875 443 4.2 4.2
Roe- van Albada: inviscid 3510 415 175 2.4 1.2
Weno-sym-bandwidth 2nd order 25479 3300 207 16 8
Weno-sym-bandwidth 4th order 34272 4854 213 23 11.5
Weno-sym-order 4th order 26831 3465 208 17 8.5
Weno-sym-order 6th order 39001 5035 212 24 12
Weno-TCD-bandwidth 2nd order 17198 2002 207 9.7 4.85
Weno-TCD-bandwidth 4th order 29232 3752 223 17 8.5
Weno-TCD-bandwidth 4th order: pressure only dcflag, 180x40 base grid (-.005, .085)(0, 0.02) 10844 2574 567 4.5 4.5
Weno-TCD-order 4th order 17990 2139 207 10 5
Weno-TCD-order 6th order 25517 3091 214 14 7

Methods

  • Roe solver with Min-Mod Limiting: Method(1) = 1 (for viscous terms), Method(2) = 2, Method(3) = 0, Limiter(1)= 1, Limiter(2) = 1, Limiter(3) = 1
  • Roe solver with van Albada Limiting: Method(1) = 1 (for viscous terms), Method(2) = 2, Method(3) = 0, Limiter(1)= 5, Limiter(2) = 5, Limiter(3) = 5

(Note: current version of code was locally modified to enable the use of these input parameters)

  • WENO - 2nd order Symmetric bandwidth optimized, Method = 0, Optimize = 1, Order = 2, GhostCells = 3
  • WENO - 4th order Symmetric bandwidth optimized, Method = 0, Optimize = 1, Order = 4, GhostCells = 4
  • WENO - 4th order Symmetric order optimized, Method = 0, Optimize = 0, Order = 3, GhostCells = 3
  • WENO - 6th order Symmetric order optimized, Method = 0, Optimize = 0, Order = 5, GhostCells = 4
  • WENO-TCD - 2nd order Symmetric bandwidth optimized, Method = 1, Optimize = 1, Order = 2, GhostCells = 3
  • WENO-TCD - 4th order Symmetric bandwidth optimized, Method = 1, Optimize = 1, Order = 4, GhostCells = 4
  • WENO-TCD - 4th order Symmetric order optimized, Method = 1, Optimize = 0, Order = 4, GhostCells = 3
  • WENO-TCD - 6th order Symmetric order optimized, Method = 1, Optimize = 0, Order = 6, GhostCells = 4
(Note that for the use of WENO-TCD a problem dependent cles_dcflag_curv2d.f function was specifed in order to control the "discontinuity" flagging of where the WENO solution is used instead of TCD.)

DoubleMachReflectionStudy

-- JackZiegler? - 11 Aug 2008

Amroc > DoubleMachReflectionStudy > ConstantPropPerformanceDMR
Copyright © 1997-2024 California Institute of Technology.