ae2f_docs
clspv_clkeys.h
Go to the documentation of this file.
1#ifndef ae2fVK_clspv_clkeys_h
2#define ae2fVK_clspv_clkeys_h
3
4#ifndef ae2fVK_clspv_IS_OPENCL
5#define ae2fVK_clspv_IS_OPENCL 0
6#endif
7
9
10#define ae2f_CL(...) __VA_ARGS__
11#define ae2f_NCL(...)
12
13#else
14
15#define ae2f_CL(...)
16#define ae2f_NCL(...) __VA_ARGS__
17
18#include <stddef.h>
19#include <stdint.h>
20
21#define __kernel
22#define __global
23#define __constant
24#define __local
25
26#endif
27
28#define int8_t ae2f_CL(char) ae2f_NCL(int8_t)
29#define uint8_t ae2f_CL(uchar) ae2f_NCL(uint8_t)
31
32#define int16_t ae2f_CL(short) ae2f_NCL(int16_t)
33#define uint16_t ae2f_CL(ushort) ae2f_NCL(uint16_t)
35
36#define int32_t ae2f_CL(int) ae2f_NCL(int32_t)
37#define uint32_t ae2f_CL(uint) ae2f_NCL(uint32_t)
39
40#define int64_t ae2f_CL(long) ae2f_NCL(int64_t)
41#define uint64_t ae2f_CL(ulong) ae2f_NCL(uint64_t)
43
44#define size_t uint
45
58
59typedef enum cl_mem_fence_flags {
62} cl_mem_fence_flags;
63
64void barrier(cl_mem_fence_flags flags);
65
66#endif
67
68/** @brief Contains both LOCAL and GLOBAL */
69#define CLK_ALL_MEM_FENCE (CLK_LOCAL_MEM_FENCE | CLK_GLOBAL_MEM_FENCE)
70
71#define work_group_reduce_add(x) 0
72
73#endif
#define ae2f_structdef(key, name)
Definition Cast.h:110
ae2f_float ae2f_float_t
Definition Float.h:38
#define _r_isz(lidx)
Definition Mlp.cl.c:54
#define m_weight
#define _r_osz(lidx)
Definition Mlp.cl.c:55
__kernel void kTrainAuto(__global void *glob, __local ae2f_float_t *loc, lrlsz_t lr)
loc ae2f_float_t[lsz - 1][Page]: OutStream ae2f_float_t[lsz - 1][Page]: DeltaStream
Definition Mlp.cl.c:290
#define p_weight
sizeof(ae2f_float_t) * pgsz_sqr * llsz
Definition Mlp.cl.c:37
#define _r_weight(lidx)
Definition Mlp.cl.c:51
#define ACT(layer_idx, r, x)
Definition Mlp.cl.c:10
#define pgsz
Definition Mlp.cl.c:21
#define ACT_DERIV(layer_idx, r, x)
Definition Mlp.cl.c:14
#define p_outstream
sizeof(ae2f_float_t) * lsz * pgsz
Definition Mlp.cl.c:33
#define r_out
Definition Mlp.cl.c:58
#define pgsz_sqr
Definition Mlp.cl.c:22
__kernel void kPredictStream(__global void *glob, __local ae2f_float_t *loc, const uint32_t lsz)
loc ae2f_float_t[Page]: inp ae2f_float_t[Page]: out
Definition Mlp.cl.c:118
#define llsz
Definition Mlp.cl.c:28
#define m_bias
#define _r_bias(lidx)
Definition Mlp.cl.c:52
#define l_inp(O_R)
Definition Mlp.cl.c:73
#define l_delta
Definition Mlp.cl.c:77
#define r_inp
Definition Mlp.cl.c:57
#define l_out(O_R)
Definition Mlp.cl.c:74
#define l_delta_then
Definition Mlp.cl.c:78
#define r_weight_then
Definition Mlp.cl.c:67
#define _r_out(lidx)
Definition Mlp.cl.c:50
#define p_deltastream
sizeof(ae2f_float_t) * pgsz * llsz
Definition Mlp.cl.c:43
#define p_goal
sizeof(ae2f_float_t) * pgsz
Definition Mlp.cl.c:46
#define p_bias
sizeof(ae2f_float_t) * pgsz * llsz
Definition Mlp.cl.c:40
__kernel void kPredict(__global void *glob, __local ae2f_float_t *loc, const uint32_t lsz)
loc ae2f_float_t[Page]: inp ae2f_float_t[Page]: out
Definition Mlp.cl.c:93
char STATIC_ASSERT_LRLSZEL_SZ[sizeof(lrlszel_t)==(sizeof(uint32_t) > sizeof(ae2f_float_t) ? sizeof(uint32_t) :sizeof(ae2f_float_t)) ? 1 :-1]
Definition Mlp.cl.c:155
#define p_layerszlist
lsz * sizeof(uint32_t)
Definition Mlp.cl.c:31
#define _r_inp(lidx)
Definition Mlp.cl.c:49
#define r_delta
Definition Mlp.cl.c:61
#define _r_delta(lidx)
Definition Mlp.cl.c:53
#define lp_deltastream
Definition Mlp.cl.c:76
#define r_weight
Definition Mlp.cl.c:59
#define r_isz
Definition Mlp.cl.c:62
#define lsz
Definition Mlp.cl.c:25
char STATIC_ASSERT_LRLSZ_SZ[sizeof(lrlsz_t)==sizeof(lrlszel_t) *3 ? 1 :-1]
Definition Mlp.cl.c:157
#define r_osz
Definition Mlp.cl.c:63
#define r_bias
Definition Mlp.cl.c:60
__kernel void kFollow(__global void *glob, __local ae2f_float_t *loc, lrlsz_t lr)
loc ae2f_float_t[lsz - 1][Page]: OutStream ae2f_float_t[lsz - 1][Page]: DeltaStream
Definition Mlp.cl.c:166
size_t get_local_linear_id()
#define __global
#define uint32_t
#define uint64_t
#define __local
#define int32_t
size_t get_global_linear_id()
size_t get_num_groups(uint dimindx)
#define uint16_t
#define size_t
#define CLK_ALL_MEM_FENCE
Contains both LOCAL and GLOBAL.
size_t get_global_offset(uint dimindx)
#define ushort
#define ulong
size_t get_group_id(uint dimindx)
#define __kernel
size_t get_global_id(uint dimindx)
uint get_work_dim()
#define work_group_reduce_add(x)
size_t get_local_id(uint dimindx)
#define int8_t
#define ae2fVK_clspv_IS_OPENCL
Definition clspv_clkeys.h:5
cl_mem_fence_flags
@ CLK_GLOBAL_MEM_FENCE
@ CLK_LOCAL_MEM_FENCE
#define int64_t
size_t get_local_size(uint dimindx)
#define uchar
size_t get_global_size(uint dimindx)
#define uint8_t
size_t get_enqueued_local_size(uint dimindx)
#define int16_t
#define uint
void barrier(cl_mem_fence_flags flags)
#define ae2f_CL(...)
#define ae2f_NCL(...)
#define __ae2f_AnnSlpFollowOneW_imp(inp, delta, weight, learningrate, inp_sz, inp_idx, out_sz, out_idx)
Weights.
Definition Slp.auto.h:366
#define __ae2f_AnnSlpFollowOneB_imp(r_bias, delta, learningrate_bias)
Definition Slp.auto.h:385
#define __ae2f_AnnSlpFetchDeltaOne_imp(v_fetchdelta_0, v_fetchdelta_1, out, out_desired, actderiv_opt, lossderiv, retdelta, oidx, osz)
Definition Slp.auto.h:513
#define CL_Q_CVRT(x)
Definition mac.h:12
#define CAST(t, x)
Definition mac.h:5
#define _clMlpGetHD1(r_delta_then, i_weight, i_delta, i_iidx, i_isz, i_oidx, i_osz)
Definition Mlp.auto.h:73
#define _clMlpRvrse(v_tmp, r_delta_then, i_oidx, i_iidx, i_isz, i_actderiv_then, i_inp, i_deltaseed)
delta to delta Propagate
Definition Mlp.auto.h:30
#define clMlpGetHD(...)
GetHidDelta Need no structure.
Definition Mlp.auto.h:120
void clMlpGetHD1_t(ae2f_float_t *const r_delta, const ae2f_float_t *const i_weight, const ae2f_float_t *const i_delta, const size_t i_iidx, const size_t i_isz, const size_t i_oidx, const size_t i_osz)
Definition Mlp.auto.h:10
#define _clMlpGetHD(ONE, r_delta_then, i_weight, i_delta, i_iidx, i_isz, i_oidx, i_osz)
Definition Mlp.auto.h:99
#define clSlpPredict
Definition Slp.auto.h:88
#define _clSlpPredict(v_predict, loc, p_inp, p_weight, p_bias, iidx, isz, oidx, osz, ACT)
Definition Slp.auto.h:23
ae2f_AnnSlpPredictOne_t _clSlpPredict_t
Definition Slp.auto.h:21
#define clSlpPredict_t
Definition Slp.auto.h:89