Progress of CUDA kernelize (File Point of View)
Opened this issue · 2 comments
sonots commented
ls ext/cumo/narray/gen/tmpl/
- accum_binary.c (mulsum)
- accum.c (sum, prod, min, max, ptp?) TODO: Implement NaN version
- accum_index.c (max_index, min_index)
- allocate.c
- alloc_func.c
- aref.c (view) <= maybe not needed
- aset.c (indexing set) <= maybe not needed
- binary2.c
- binary.c
- binary_s.c
- bincount.c
- cast_array.c (store)
- cast.c (new_dim0, cast_array, store)
- class.c
- clip.c
- coerce_cast.c
- cond_binary.c
- cond_unary.c
- cum.c (cumsum, sumprod)
- each.c (m_data_to_num) <= cudaSynchronize()
- each_with_index.c <= cudaSynchronize()
- extract.c Extract element value as Ruby Object if self is a dimensionless NArray,
- extract_data.c Convert a data value of obj (with a single element) to dtype <= no need
- eye.c
- fill.c
- format.c
- format_to_a.c
- frexp.c
- init_class.c <= no need
- init_module.c <= no need
- inspect.c
- lib.c
- logseq.c
- map_with_index.c <= cudaSynchronize()
- median.c
- minmax.c (minmax) => fix accum's ptp, too
- module.c (no need)
- new_dim0.c
- poly.c
- pow.c
- powint.c
- qsort.c
- rand.c
- rand_norm.c
- seq.c
- set2.c
- sort.c
- sort_index.c
- store_array.c (store from Ruby array)
- store_bit.c (store from Numo::Bit)
- store.c (store_numeric, store_from, store_bit, store_array)
- store_from.c (store from other dtype arrays)
- store_numeric.c (store from a numeric value) new_dim0, store
- to_a.c
- unary2.c
- unary.c
- unary_ret2.c
- unary_s.c (sqrt, etc)
ls ext/cumo/narray/gen/tmpl_bit/
- allocate.c
- aref.c
- aset.c
- binary.c
- bit_count.c
- bit_reduce.c
- each.c
- each_with_index.c
- extract.c
- fill.c
- format.c
- format_to_a.c
- inspect.c
- mask.c
- none_p.c
- store_array.c
- store_bit.c
- store_from.c
- to_a.c
- unary.c
- where2.c
- where.c
sonots commented
narray.c
- initialize
- initialize_copy
- zeros
- ones
- linspace
- logspace
- eye
- view
- expand_dims
- reverse
- upcast
- byte_size
- from_binary
- from_string
- store_binary
- to_binary
- to_string
- marshal_dump
- marshal_load
- byte_size
- cast_to
- coerce
- column_major?
- row_major?
- byte_swapped?
- host_order?
- inplace
- inplace?
- inplace!
- out_of_place!
- not_inplace!
- ==
sonots commented
data.c
- flatten
- swapaxes
- transpose
- reshape
- diagonal
- swap_byte
- hton
- network_order?
- little_endian?
- vacs_order?