GPU Accelerated Spectra

Example using GPU calculation with eq_spectrum_gpu()

This method requires a GPU - Currently, only Nvidia GPU’s are supported. For more information on how to setup your system to run GPU-accelerated methods using CUDA, check GPU Spectrum Calculation on RADIS

Note

in the example below, the code runs on the GPU by default. In case no Nvidia GPU is detected, the code will instead be ran on CPU. This can be toggled manually by setting the backend keyword either to 'gpu-cuda' or 'cpu-cuda'. The run time reported below is for CPU.

/home/docs/checkouts/readthedocs.org/user_builds/radis/checkouts/latest/radis/db/molparam.py:252: FutureWarning:

The 'delim_whitespace' keyword in pd.read_csv is deprecated and will be removed in a future version. Use ``sep='\s+'`` instead

In ``/home/docs/.radisdb/hitemp`` keep only relevant input files:
/home/docs/.radisdb/hitemp/CO2-02_02125-02250_HITEMP2010.hdf5
/home/docs/.radisdb/hitemp/CO2-02_02250-02500_HITEMP2010.hdf5
Calculating Equilibrium Spectrum
Physical Conditions
----------------------------------------
   Tgas                 1500.0 K
   Trot                 1500.0 K
   Tvib                 1500.0 K
   isotope              1,2,3
   mole_fraction        0.8
   molecule             CO2
   overpopulation       None
   path_length          0.2 cm
   pressure             1.0 bar
   rot_distribution     boltzmann
   self_absorption      True
   state                X
   vib_distribution     boltzmann
   wavenum_max          2450.0000 cm-1
   wavenum_min          2150.0000 cm-1
Computation Parameters
----------------------------------------
   Tref                 296 K
   add_at_used          numpy
   broadening_method    voigt
   cutoff               0 cm-1/(#.cm-2)
   dbformat             hitemp-radisdb
   dbpath               /home/docs/.radisdb/hitemp/CO2-02_02125-02250_HITEMP2010.hdf5,/home/docs/.radisdb/hitemp/CO2-02_0225...
   diluent              air
   folding_thresh       1e-06
   include_neighbouring_lines  True
   memory_mapping_engine  auto
   neighbour_lines      0 cm-1
   optimization         simple
   parfuncfmt           hapi
   parsum_mode          full summation
   pseudo_continuum_threshold  0
   sparse_ldm           False
   truncation           50 cm-1
   waveunit             cm-1
   wstep                0.002 cm-1
   zero_padding         150001
----------------------------------------
0.91s - Spectrum calculated
/home/docs/checkouts/readthedocs.org/user_builds/radis/checkouts/latest/radis/lbl/factory.py:1107: FutureWarning:

Calling float on a single element Series is deprecated and will raise a TypeError in the future. Use float(ser.iloc[0]) instead

Can't find libcuda.so.1...
/home/docs/checkouts/readthedocs.org/user_builds/radis/checkouts/latest/radis/gpu/gpu.py:112: NoGPUWarning:

Failed to load CUDA context, this happened either becauseCUDA is not installed properly, or you have no NVIDIA GPU. Continuing with emulated GPU on CPU...This means *NO* GPU acceleration!

Number of lines loaded: 1128265

mode: CPU
Finished calculating spectrum!
0.55s - Spectrum calculated

(<Figure size 640x480 with 2 Axes>, [<Axes: >, <Axes: xlabel='Wavelength (nm)'>])

from radis import SpectrumFactory, plot_diff

sf = SpectrumFactory(
    2150,
    2450,  # cm-1
    molecule="CO2",
    isotope="1,2,3",
    wstep=0.002,
)

sf.fetch_databank("hitemp")

T = 1500.0  # K
p = 1.0  # bar
x = 0.8
l = 0.2  # cm
w_slit = 0.5  # cm-1

s_cpu = sf.eq_spectrum(
    name="CPU",
    Tgas=T,
    pressure=p,
    mole_fraction=x,
    path_length=l,
)
s_cpu.apply_slit(w_slit, unit="cm-1")

s_gpu = sf.eq_spectrum_gpu(
    name="GPU",
    Tgas=T,
    pressure=p,
    mole_fraction=x,
    path_length=l,
    backend="gpu-cuda",
)
s_gpu.apply_slit(w_slit, unit="cm-1")

plot_diff(s_cpu, s_gpu, var="radiance", wunit="nm", method="diff")