summaryrefslogtreecommitdiff
path: root/gpu.py
AgeCommit message (Collapse)Author
2011-08-11mergeStan Seibert
2011-08-11Tell CUDA to enlarge L1 cache.Stan Seibert
2011-08-11merge headsAnthony LaTorre
2011-08-11add consume() to itertoolset. checksum the number of bits shifted in each ↵Anthony LaTorre
iteration of the bounding volume hierarchy construction. only cache the bounding volume hierarchy information.
2011-08-11Show number of registers used in CUDA kernelsStan Seibert
2011-08-11Switch from texture to float3 array for upper and lower bounds. 10% speed ↵Stan Seibert
boost!
2011-08-09Automatically detect kernprof and profile main() in sim.py. Also ↵Stan Seibert
synchronize CUDA calls.
2011-08-09The --use-fast-math gives the simulation another 25% speed boost.Stan Seibert
2011-08-09fix variable name in GPU.color_solids()Anthony LaTorre
2011-08-09merge headsAnthony LaTorre
2011-08-09switch to indexing child nodes by start and stop indices instead of start ↵Anthony LaTorre
and length; this reduces a bit of arithmetic when traversing the bounding volume hierarchy and makes the code in Geometry.build() more concise. add an untested cuda kernel to interleave the bits in three uint64 arrays.
2011-08-09Put number of detected photons into charge value for channel.Stan Seibert
2011-08-09Store a photon history for each hit channel. If multiple photons hit theStan Seibert
same channel, their history bits are OR'ed together.
2011-08-08propagate() takes an array of photon offsets and a range ofStan Seibert
offsets to load. Now events with more photons than RNG states can be propagated through multiple kernel calls. Also lays the groundwork for consolidating photons between steps to reduce the amount of propagation work required.
2011-08-08toggle fullscreen mode with f11 key. use matplotlib colors maps defined in ↵Anthony LaTorre
matplotlib.cm.
2011-08-08merge headsAnthony LaTorre
2011-08-08add a simple event viewer. view events by running ./camera.py ↵Anthony LaTorre
<detector_name> -i <name_of_io_file>.
2011-08-08Bug fixes when retrieving and rerunning photon lists through propagate.Stan Seibert
2011-08-05Warn user if for some reason the default BVH quantization is used becauseStan Seibert
they didn't build the geometry before loading it on the GPU.
2011-08-04Implement propagate and daq functions in gpu module.Stan Seibert
2011-08-03add a GPU class to handle both the gpu context and module; since the ↵Anthony LaTorre
geometry requires global device pointers, there should be a one to one correspondence between modules and contexts. the current plan is to perform all gpu operations within this class. also add a simple color map to display hit pmt charge and timing information.
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