Lets simulate the water case of the SPHERIC validation test number 9:
CUSL VIII finish is coming:
Unfortunately the selected projects and the schedule are not published yet, but I guess that I’m assisting anyway!
One of the main target of AQUAgpusph is the possibility to compute confident enough pressure values.
In this line the standard deviation of the interpolation process has been implemented in AQUAgpusph. Hence now you have a mean value, and a measurement of the dispersion of the data near to the sensor.
It could be extremely useful to get high quality pressure records:
In the previous figure the pressure computed in a simulation of the validation test number 10 of the SPHERIC benchmark is compared with the experimental results.
This new feature can be found in the unstable 1.5 branch of the code.
As I reported some weeks ago, I’m working in the code optimization. To do it I implemented the following changes:
- Now all the data (including the output one) is sorted before compute the interactions
- Hence the permutations vectors are not needed anymore in the interactions stage
- Some useless output data have been removed
- The registers pressure have been optimized
Let’s say goodbye to more than 10% of the simulation time!
For the moment the optimized version can be found in the optimization branch of the git repository.
P.S. I performed this work with CodeXL by AMD, basically due to NVidia suddenly decided to remove the OpenCL support in the profiler in CUDA 5.
NVidia is expending a lot of resources trying to destroy OpenCL, while AMD is beating them in the hardware side (the real world).
I just hope NVidia people change it’s direction before they reach the final which are deserving.
Also CodeXL is free!
With the last changes implemented in the code the registries pressure during the interactions computations has been dramatically increased, which reduces the computation device occupancy.
This performance lost is my next target to fix (partially at least)…