summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)Author
2011-07-10added a hybrid monte carlo ray tracing rendering algorithmAnthony LaTorre
2011-06-24argument '-j' to threadtest.py now specifies a list of device ids to be ↵Anthony LaTorre
used. GPUThread objects now shallow copy the geometry so that threads are not writing to the same memory when the geometry is loaded onto the gpu. the model number for the 12" Hamamatsu PMT is R11708, not r7081 (which is the model for the 10" PMT); all references to the 12" PMT were changed accordingly. only allocate space for 20 materials and 20 surfaces on the gpu instead of 100 to save some space. started to modify track.py to build its own photons and module since the GPUThread object only copies photon hit times back from the gpu (not track information), but I am waiting to find out if pycuda GPUArrays can be used with vector types.
2011-06-24speedup mesh intersection by skipping directly to the child node of nodes ↵Anthony LaTorre
with only one child.
2011-06-23move photon initialization to the gpu. it's unclear if this is a speed ↵Anthony LaTorre
improvement.
2011-06-22Start jittering the PMT times in the DAQ simulation.Stan Seibert
2011-06-22Switch from float4 to float3 for vertices.Stan Seibert
2011-06-22Fix accidental reuse of absorption_length when computing scattering length.Stan Seibert
2011-06-21Switch vertex list on GPU from a texture to a standard device array.Stan Seibert
This avoids the texture size limitation.
2011-06-21A GPU-side "DAQ" implementation that identifies the first photon on each ↵Stan Seibert
channel, which is presumed to trigger that channel. Major speed up in conversion of detection times to time PDFs.
2011-06-20pack material and surface indices into the fourth byte of the triangle array ↵Anthony LaTorre
on the GPU. you can now take a screenshot of an image rendered with view.py() by pressing the f12 key.
2011-06-19reverse the face orientation on all triangles in the stl files exported from ↵Anthony LaTorre
sketchup.
2011-06-18change curandstate array length to 100,000. gputhread.run() now correctly ↵Anthony LaTorre
initializes ALL of the states.
2011-06-17visually tested optics code. added models of the inner and outer meshes for ↵Anthony LaTorre
the 12" hamamatsu and sno pmts. ratdb.py is able to parse ratdb files. chromaticity.py provides a function to map wavelength -> rgb color. lbne detector model now includes an outer black cylinder and pmts with a glass layer and photocathode/reflective surfaces.
2011-06-12added some fun models; added some untested code to implement absorption, ↵Anthony LaTorre
scattering, reflection, and refraction
2011-06-07Switch triangle texture to device array, use int32 and float32Stan Seibert
datatypes everywhere, and build final mesh without concatenation of lists. This allows for very large detectors, like full size LBNE.
2011-06-03load material/surface index lookup arrays to the gpu and bind them to ↵Anthony LaTorre
textures. also, forgot to include the gpu code for material/surface structures
2011-06-03load materials/surfaces onto the gpuAnthony LaTorre
2011-06-02material/surface codes or color are packed into the fourth byte of the ↵Anthony LaTorre
triangle data texture depending on the color keyword passed to Geometry.build()
2011-06-02triangle mesh is now stored everywhere as a split list of vertices and trianglesAnthony LaTorre
2011-06-01first step towards moving to a new mesh/solid/geometry structureAnthony LaTorre
2011-05-27kernel ran off the end of the ray arrayAnthony LaTorre
2011-05-23lbne model now imports python modules directlyAnthony LaTorre
2011-05-20faster bounding volume hierarchy constructionAnthony LaTorre
2011-05-18added test likelihoodAnthony LaTorre
2011-05-18added some more documentation and a more accurate miniature version of lbneAnthony LaTorre
2011-05-17added documentationAnthony LaTorre
2011-05-16added stl mesh viewerAnthony LaTorre
2011-05-15cleanup. fixed testsAnthony LaTorre
2011-05-15new geometry class. beginning to implement physics by defining a material ↵Anthony LaTorre
class; each triangle will have a material linked to both of its sides
2011-05-13use floatsAnthony LaTorre
2011-05-13added a bounding volume hierarchyAnthony LaTorre
2011-05-10move triangles to meshAnthony LaTorre
2011-05-09pull triangle vertices in kernel functionAnthony LaTorre
2011-05-09improve triangle intersection algorithm by allowing it to terminate earlyAnthony LaTorre
2011-05-09use pixel pointer instead of pixel arrayAnthony LaTorre
2011-05-09slight speed increaseAnthony LaTorre
2011-05-07tie fighterAnthony LaTorre