Skip to content

Commit

Permalink
C_FTYPE string adapted for single-precision and cosmetics
Browse files Browse the repository at this point in the history
  • Loading branch information
thehrh committed Aug 1, 2024
1 parent 24035d7 commit df6e11b
Showing 1 changed file with 4 additions and 4 deletions.
8 changes: 4 additions & 4 deletions pisa/stages/GLOBALS.md
Original file line number Diff line number Diff line change
Expand Up @@ -12,14 +12,14 @@ Here we keep track of which global constants are available, what their purpose i
| `NUMBA_CUDA_AVAIL` | Availability of Numba's CUDA interface | `False` (unless CUDA-capable GPU available) | |
| `TARGET` | Numba compilation target | `cpu` | `PISA_TARGET` (GPU target only possible if `NUMBA_CUDA_AVAIL`) |
| `OMP_NUM_THREADS` | Number of threads allocated to OpenMP | `1` | `OMP_NUM_THREADS` |
| `PISA_NUM_THREADS` | Global limit for number of threads (also upper limit for `OMP_NUM_THREADS`) | `1` (`numba.config.NUMBA_NUM_THREADS`) for `TARGET='cpu'`(`'parallel'`) | `PISA_NUM_THREADS` |
| `PISA_NUM_THREADS` | Global limit for number of threads (also upper limit for `OMP_NUM_THREADS`) | `1` (`numba.config.NUMBA_NUM_THREADS`) for `TARGET='cpu'` (`'parallel'`) | `PISA_NUM_THREADS` |
| `FTYPE` | Global floating-point data type | `np.float64` | `PISA_FTYPE` |
| `CTYPE` | Global complex-valued floating-point data type | `np.complex128`(`np.complex64`) for `FTYPE=np.float64(32)` | |
| `CTYPE` | Global complex-valued floating-point data type | `np.complex128` (`np.complex64`) for `FTYPE=np.float64(32)` | |
| `ITYPE` | Global integer data type | `np.int64` (`np.int32`) for `FTYPE=np.float64(32)` | |
| `HASH_SIGFIGS` | Number of significant digits used for hashing numbers, depends on `FTYPE` | `12(5)` for `FTYPE=np.float64(32)` | |
| `EPSILON` | Best numerical precision, derived from `HASH_SIGFIGS` | `10**(-HASH_SIGFIGS)` | |
| `C_FTYPE` | C floating-point type corresponding to `FTYPE` | `'double'('single')` for `FTYPE=np.float64(32)` | |
| `C_PRECISION_DEF` | C precision of floating-point calculations, derived from `FTYPE` | `'DOUBLE_PRECISION'('SINGLE_PRECISION')` for `FTYPE=np.float64(32)` | |
| `C_FTYPE` | C floating-point type corresponding to `FTYPE` | `'double'` (`'float'`) for `FTYPE=np.float64(32)` | |
| `C_PRECISION_DEF` | C precision of floating-point calculations, derived from `FTYPE` | `'DOUBLE_PRECISION'` (`'SINGLE_PRECISION'`) for `FTYPE=np.float64(32)` | |
| `CACHE_DIR` | Root directory for storing PISA cache files | `'~/.cache/pisa'` | 1.`PISA_CACHE_DIR`, 2.`XDG_CACHE_HOME/pisa` |

## Usage
Expand Down

0 comments on commit df6e11b

Please sign in to comment.