Details of run
10 days has been chosen to be a similar time to the run time
for the Bakward-Euler run at N96 (5,000s
before the improvements).
Calling tree
The times in the tables below are the total time, which is the time in each routine and the routines below it, except where `itself' is written to indicate time taken solely by that routine. Only the main routes - as given by time spent in them - are general shown.
| Routines | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UM_SHELL (4,969s) | ||||||||||||||||||
| U_MODEL_4A (4,965s) | ||||||||||||||||||
| ATM_STEP_4A* (3,010s) | UKCA_MAIN1 (1,611s) | |||||||||||||||||
| ATMOS _PHYS- ICS1 (1,047s) | EG_ COR- RECT_ TRAC- ERS (45s) | ATMOS _PHYS- ICS2 (470s) | EG_ SL_ HELM- HOLTZ (382s) | TR_ SET_ PHYS _4A* (63s) | EG_CORRECT _TRACERS _UKCA (208s) | SL_ TRAC- ER1_ 4A (157s) | EG_ SL_ MOI- STURE (88s) | EG_SL_ FULL_WIND (161s) | ⇓ | UP- DATE _M_ STAR (96s) | ATM_ STEP_ STASH (30s) | ⇓ | ⇓ | See profiling for UKCA_ MAIN1 below | ||||
| See profile for ATMOS_ PHYSICS1 and EG_CORRECT _TRACERS below | See profile for ATMOS_ PHYSICS2 and EG_SL_ HELMHOLTZ below | EG_SL_WIND_U, EG_SL_WIND_V & EG_SL_WIND_W (48 + 48 + 54 = 150s) | ⇓ | EG_Q_ TO_MIX (97s) | ⇓ | STASH (120s) | ||||||||||||
| Itself (144s) | EG_INTERPOLATION _ETA (283s) | DEP- ARTURE_ POINT _ETA (105s) | EG_SWAP_ BOUNDS_DP (219s) | STWORK (120s) | ||||||||||||||
| EG_ CUBIC_ LAG- RANGE (115s, itself) | EG_VERT_ WEIGHTS_ ETA (25s, itself) | MONO_ ENFORCE (22s, itself) | Itself (40s) | See profile for SWAP_ BOUNDS _DP below | SPA- TIAL (41s) | PP_ HEAD (38s) | EXP- PXI (24s, itself) | |||||||||||
After considering our fixes, the mean time EG_SL_HELMHOLTZ seems to be significantly larger and the mean time in ATMOS_PHYSICS1 seems to be significantly less than the N96 Backward-Euler run.
The profiling tables lower down suggest that around an extra 100s is spent in TRI_SOR_DP_DP at N216, which accounts for a large chunk of the extra time in EG_SL_HELMHOLTZ. While all the routines below RAD_CTL and MICROPHYS_CTL seem to be much quicker, which accounts for the reduced time in ATMOS_PHYSICS1.
| Routines | Total mean time | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| UKCA_MAIN* (1,611s) | 1,611s | ||||||||||
| UKCA_AERO_CTL (1,006s) | UKCA_ ACTIVATE (106s) | UKCA_ CHEMISTRY_CTL (266s) | UKCA_ EMISSION _CTL (42s) | 1,420s | |||||||
| UKCA_AERO_STEP (963s) | UKCA_ ABDULRAZZAK_ GHAN (97s) | ASAD_CDRIVE (252s) | 1,312s | ||||||||
| UKCA_COAGWITHNUCL (455s) | UKCA_ CONDEN (180s) | UKCA_ CHECK_ MD_ND (81s, itself) | UKCA_ CALCNUCRATE (79s) | UKCA_ VOLUME_ MODE (44s) | Itself (92s) | ASAD_SPMJPDRIV (244s) | 1,217s | ||||
| Itself (375s) | UKCA_ SOLVECOAGNUCL _V (80s, itself) | UKCA_ COND_ COFF_V (111s, itself) | Itself (69s) | UKCA_ BINAPARA (74s, itself) | Itself (22s) | ASAD_SPIMPMJP (243s) | 1,189s | ||||
| SPLINSLV2 (151s, itself) | SPFULJAC (76s, itself) | 1,173s | |||||||||
UKCA_CHEMISTRY_CTL is significant here, as oppose the Backward-Euler run at N96 (where UKCA_CHEMISTRY_CTL_BE was 1.23s). This was meant to be Backward-Euler run but clearly it isn't. And we could probably lose about 266s by running a Backward-Euler run.
| ATMOS_PHYSICS1* (1,047s) | EG_CORRECT _TRACERS (45s) | |||||||||||||
| Itself (275s) | RAD_CTL (332s) | MICROPHYS_CTL (122s) | NI_GWD _CTL (300s) | ⇓ | ⇓ | EG_MASS_ CONSERVATION (41s) | ||||||||
| LW_RAD (237s) | SW_RAD (77s) | LS_PPN (113s) | G_ WAVE _5A (242s) | GW_ USSP (58s) | GLOBAL _2D_ SUMS (87s, itself) | Itself (29s) | ||||||||
| RADIANCE_CALC (309s) | LS_PPNC (106s) | SWAP_ BOUNDS (see table below) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UKCA_ RADAER _BAND_ AVERAGE (162s, itself) | SOLVE_BAND_K_EQV (116s) | UKCA_ RADAER_ COMPUTE _AOD (13s, itself) | LSP_ICE (78s) | Itself (28s) | ||||||||||
| MCICA_ SAMPLE (90s) | LSP_ INIT (18s) | LSP_ FALL (15s) | ||||||||||||
| MONOCHR- OMATIC_ RADIANCE (78s) | Itself (10s) | Itself (13s) | ||||||||||||
| MONOCHR- OMATIC_ RADIANCE _TSEQ (67s) | ||||||||||||||
| MCICA_ COLUMN (67s) | ||||||||||||||
| TWO_COEFF (46s) | ||||||||||||||
| TRANS_ SOURCE_ COEFF (27s) | Itself (7s) | |||||||||||||
| Itself (17s) | ||||||||||||||
| ATMOS_PHYSICS2 (470s) | EG_SL_HELMHOLTZ (382s) | ||||
| NI_CONV_CTL (222s) | NI_IMP_CTL (90s) | SWAP_BOUNDS, SWAP_BOUNDS_2D_MV & SWAP_BOUNDS_MV (see table below) | EG_BICGSTAB (253s) | EG_HELM_RHS_STAR (92s) | |
|---|---|---|---|---|---|
| GLUE_CONV_6A (165s) | IMP_SOLVER (47s) | EG_PRECON (173s) | EG_SISL_INIT (69s) | ||
| Itself (79s) | MID_CONV_6A (50s) | TRI_SOR_DP_DP (173s) | EG_SISL_INIT_UVW (66s) | ||
| Itself (20s) | Itself (126s) | Itself (39s) | |||
| Routines | Total mean time | |||
|---|---|---|---|---|
| EG_SWAP_BOUNDS_DP (s) | ATMOS_PHYSICS1, ATMOS_PHYSICS2, G_WAVE_5A, ... | 219 + ... | ||
| SWAP_BOUNDS & SWAP_BOUNDS_DP (437 + 252 = 689s) | SWAP_BOUNDS_MV (114s, itself) | 803s | ||
| SWAP_BOUNDS_EW_DP (360s) | SWAP_BOUNDS_NS_DP (328s, itself) | 802s | ||
| SWAP_BOUNDS_EW_H1_DP (263s, itself) | Itself (97s) | 802s | ||
