- Some performance on low-end GPU may drop because of that, but only for a few hash-modes
- Dropped scalar code (aka warp) since we do not have any vector datatypes anymore
- Renamed C++ overloading functions memcat32_9 -> memcat_c32_w4x4_a3x4
- Still need to fix kernels to new function names, needs to be done manually
- Temperature Management needs to be rewritten partially because of conflicting datatypes names
- Added code to create different codepaths for NV on AMD in runtime in host (see data.vendor_id)
- Added code to create different codepaths for NV on AMD in runtime in kernels (see IS_NV and IS_AMD)
- First tests working for -m 0, for example
- Great performance increases in general for NV so far
- Tested amp_* and markov_* kernel
- Migrated special NV optimizations for rule processor
TODOS:
- Let oclHashcat actually use the new paths
- Find a better way for native compilation
- Replace /bin/cp with /usr/bin/install where it has to copy files recursive
Problem here is; This process requires alot of memory. If host memory is too small it will segfault somewhere inside AMD's OpenCL runtime library. Therefore it's better to stick to NPROCS as part of a workaround.
This reverts commit e5adccbf38.
