|
|
In the previous examples (Quick-Start Guides), we read SPI images from hdf5 files. In real experiment, data will be in [Xtc2](https://confluence.slac.stanford.edu/display/PSDMInternal/XTC+for+LCLS-II) format and [PSANA2](https://github.com/slac-lcls/lcls2) is needed to read this type of data.
|
|
|
|
|
|
As of 6/2/22, Psana2 enabled versions of Spinifel live in separated branches. Please contact Mona [https://gitlab.osti.gov/mtip/spinifel/-/tree/psana2](psana2-mpi branch) or Seema (Legion) for more information.
|
|
|
As of 6/2/22, Psana2 enabled versions of Spinifel live in separated branches. Please contact Mona [psana2-mpi branch](https://gitlab.osti.gov/mtip/spinifel/-/tree/psana2) or Seema (Legion) for more information.
|
|
|
|
|
|
This guideline is for running on Cori using simulated data converted from hdf5 format (3iyf_sim_400k.h5). For running with larger data set or on Summit, please see the last section for detail on locations of the data.
|
|
|
|
... | ... | @@ -42,6 +42,10 @@ salloc -N 1 -t 4:00:00 -C gpu --gpus=1 -A m1759 -q special --tasks-per-node=10 |
|
|
#!/bin/bash
|
|
|
srun -n 3 -G 1 python -m spinifel --settings=./settings/cgpu_mpi.toml --mode=mpi
|
|
|
```
|
|
|
## Comparing output with hdf5
|
|
|
|
|
|
Two log files below were generated after running spinifel with psana2 using the instruction above. No. of images per rank is set to 1000 and No. of max generations is 3. Note that for hdf5 run, we use srun -n 1 (instead of -n 3) since there are no reserved cores for the hdf5 version.
|
|
|
- [On-screen output from Xtc2]
|
|
|
## Running with larger dataset
|
|
|
|
|
|
Data are currently being transferred to both Cori and Summit (06/02/22). |