projects / lattice-boltzmann-epiphany.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 033d3e6) | patch
Makefile: run everything internally with float
authorSebastian <git@sraa.de>
Fri, 25 Jul 2014 14:07:09 +0000 (14:07 +0000)
committerSebastian <git@sraa.de>
Fri, 25 Jul 2014 14:07:09 +0000 (14:07 +0000)
commit57534c6c26fafacd5344bfd09d5bf486327e659d
tree805d1e74113c08a54b737a3c166e3216158fce76tree | snapshot
parent033d3e6771ae79e5f817e0d3c0d456e7e35b4656commit | diff
Makefile: run everything internally with float

- treat floating point constants as single precision
  (this removes all dependencies for double)
- use internal.ldf linker script
  (puts libc functions in local memory)

speedup is immense. running code from external memory
is extremely slow, especially when all cores fight for it.
lb/Makefile diff | blob | history
Lattice Boltzmann implementation for Adapteva Epiphany