updated working version

updated working version

now includes .txt files

update

adding new precision-templated structure

adding new precision-templated structure

update

NR approximation of the inverse of the square root

updated intel compiler

GPU version

update with pseudo-working version

update

added GPU makefile

update

update

sanatization

sanatization

update

update with intel compilers

removing ldg accessess

correct values

sanatization

sanatization

NR implementation of the inv sqrt

update

updated version of the scripts

Merge branch 'master' of ssh://c4science.ch/diffusion/1448/lenstool-hpc

adding fast sqrt reciprocal for SIS

adding message when exiting

Mixed precision Grid Gradient Benchmark

adding Ofast

update

mpi check functions

update

update

update

update

improved makefiles

casting literals explicitely to prevent compilers to do it for us (badly)

bug correction

bug correction

removing debugging comments

update to match the new interface names

update for merge

renaming of the executable to match the benchmark

adding a counter to prevent comms when no images are found

adding MPI and GPU flags

reworked cleaner interface for gradient and grid gradient computations

removing unecessary code + cleanup

removing unecessary output

interface update update

adding MPI support

small updates

1 rcp instead of 2 divisions

update for intel version 17

update with a working MPI implementation

inserting MPI

update

removing absolute paths to inludes and libs

changing type 5 to 1 for lenstool

further separation in the chi computation

version 5 inserted

deleted default makefile

intel mpi loaded

separation between image localisation and chi computation

sepaating souce location and chi computation

inserting more tests

update

removed default makefile

GPU makefile to separate CPU and GPU versions

adding absolute path of the code for global variables

update and bug corrections

GPU-enable code makefile

update after the rebase

update after the rebase

update after the rebase

update after the rebase

update after the rebase

update after the rebase

update after the rebase

local chi implementation

debugged makefile

update after the rebase

deleted benchmark

update

new makefiles

updated to reflect the fact that the GPU version of the gradient moved to src

update

updated to reflect the fact that the GPU version of the gradient moved to src

adding GPU library

cleanup

adding vector off flag (which could be unecessary)

adding already computed rotations into the gradient computation

updated benchmark

a little bit of tabulation to make things clearer

added vectorization off toggle flag

updated for precomputed rotations

adding tabulation for readability

adding already computed rotations into the gradient computation

de-vectorized loop added to the gradient computation