| 348 | The SETTE tests have been repeated with the [http://forge.ipsl.jussieu.fr/nemo/browser/NEMO/trunk?rev=14922 trunk@14922] in order to include bug fixes that allow all SETTE tests to be run with `nn_hls = 2` and tiling. |
| 349 | |
| 350 | The tests are the same as detailed above except: |
| 351 | |
| 352 | * The [http://forge.ipsl.jussieu.fr/nemo/browser/NEMO/trunk?rev=14922 trunk@14922] is used (but still compared with results from the [http://forge.ipsl.jussieu.fr/nemo/browser/NEMO/trunk?rev=14820 trunk@14820]) |
| 353 | * [http://forge.ipsl.jussieu.fr/nemo/browser/utils/CI/sette?rev=14844 SETTE@14844] is used |
| 354 | * Additional tests with `key_loop_fusion` have been performed |
| 355 | * `nn_hls = 2` is set directly in namelist_ref, instead of via `USING_EXTRA_HALO`, in order to run all SETTE tests with the extended haloes (and tiling) |
| 356 | * The default tile size in namelist_ref is 99999i x 10j (to ensure there is always only 1 tile in i) |
| 357 | * Icebergs are not activated |
| 358 | |
| 359 | All SETTE tests pass and give the same results as the [http://forge.ipsl.jussieu.fr/nemo/browser/NEMO/trunk?rev=14820 trunk@14820], except AGRIF_DEMO which differs after 17 timesteps for all `nn_hls = 2` tests. |
| 360 | This is thought to be because one of the AGRIF domains in this configuration is not large enough for `nn_hls = 2`. |
| 361 | |
| 362 | ==== Regular checks |
| 363 | |
| 364 | All checks are the same as before, but the run time/memory changes are significant in some cases. |
| 365 | These are reported here for increases in time/memory larger than 10% that are present in both REPRO experiments of a configuration: |
| 366 | |
| 367 | * QCO, `nn_hls == 1` |
| 368 | * No significant changes |
| 369 | * QCO, `nn_hls == 2` |
| 370 | * GYRE_PISCES: time + 13-18%, memory + 13-18% |
| 371 | * QCO, `nn_hls == 2` and `ln_tile = .true.` |
| 372 | * AMM12: memory + 18% |
| 373 | * WED025: memory + 17% |
| 374 | * QCO, loop fusion and `nn_hls == 2` |
| 375 | * AMM12: time + 20% |
| 376 | * QCO, loop fusion, `nn_hls == 2` and `ln_tile = .true.` |
| 377 | * AGRIF_DEMO: time + 11-15% |
| 378 | * AMM12: memory + 17-20% |
| 379 | * WED025: memory + 19% |
| 380 | |
| 381 | * non-QCO, `nn_hls == 1` |
| 382 | * No significant changes |
| 383 | * non-QCO, `nn_hls == 2` |
| 384 | * No significant changes |
| 385 | * non-QCO, `nn_hls == 2` and `ln_tile = .true.` |
| 386 | * AGRIF_DEMO: memory + 13% |
| 387 | * AMM12: memory + 18-20% |
| 388 | * GYRE_PISCES: time + 11-24% |
| 389 | * ORCA2_ICE_OBS: memory + 12-16% |
| 390 | * WED025: memory + 15-16% |
| 391 | * non-QCO, loop fusion and `nn_hls == 2` |
| 392 | * ORCA2_ICE_OBS: time + 11-17% |
| 393 | * non-QCO, loop fusion, `nn_hls == 2` and `ln_tile = .true.` |
| 394 | * AGRIF_DEMO: memory + 11-12% |
| 395 | * AMM12: memory + 21-23% |
| 396 | * WED025: memory + 17-19% |
| 397 | |
| 398 | The time increases do not seem consistent enough to indicate a systematic issue. |
| 399 | However, there is evidence to suggest that tiling increases the memory cost of AGRIF_DEMO (11-13%), AMM12 (17-23%) and WED025 (15-19%). |
| 400 | This is partly due to the use of `nn_hls = 2`, which increases the domain size, but in AMM12 & WED025 this is only responsible for up to 7% of the increased memory cost. |