You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When running stdstars super-fast on GPUs, @dmargala noted that re-calculating standard star template magnitudes every time is a significant contributor to the total stdstar runtime. These could be pre-calculated using standard filters needed by DESI (Gaia plus PHOTSYS=N/S for g,r,z) and saved in the $DESI_BASIS_TEMPLATES/stdstar_templates*.fits files themselves instead of calling desispec.scripts.stdstar.get_magnitude in desispec.scripts.stdstar.pyL521 . Note that there are other later calls to get_magnitude on the extincted flux of the selected model; this would still need to be custom.
I think this is independent of the recent work on dust, E(B-V), and Vega vs. AB but it touches similar code, so let's wait for PR #1246 and desihub/desiutil#169 before adding this optimization.
The text was updated successfully, but these errors were encountered:
When running stdstars super-fast on GPUs, @dmargala noted that re-calculating standard star template magnitudes every time is a significant contributor to the total stdstar runtime. These could be pre-calculated using standard filters needed by DESI (Gaia plus PHOTSYS=N/S for g,r,z) and saved in the $DESI_BASIS_TEMPLATES/stdstar_templates*.fits files themselves instead of calling
desispec.scripts.stdstar.get_magnitude
in desispec.scripts.stdstar.pyL521 . Note that there are other later calls toget_magnitude
on the extincted flux of the selected model; this would still need to be custom.I think this is independent of the recent work on dust, E(B-V), and Vega vs. AB but it touches similar code, so let's wait for PR #1246 and desihub/desiutil#169 before adding this optimization.
The text was updated successfully, but these errors were encountered: