hexagon: support for OP_CPY, host buffers now optional, hvx-utils refactoring and optimizations (#18822)
* hexagon: disable repack buffers if host buffers are disabled, improved handling of env vars * hexagon: add support for OP_CPY fp16/fp32 -> fp16/fp32 Factore out all hvx_copy functions into hvx-copy.h header and reduced code duplication. Update HTP ops infra to support OP_CPY * hexagon: cleanup and refactor hex/hvx/htp headers and helper libs hex is basically all scalar/core platform stuff (L2, DMA, basic utils) hvx is all hvx related utils, helpers, etc htp is higher level stuff like Ops, etc hvx-utils library got a nice round of cleanup and refactoring to reduce duplication use hvx_vec_store_a where possible * hexagon: refactor HVX sigmoid functions to hvx-sigmoid.h Moved sigmoid and tanh vector functions from hvx-utils.h to a new header hvx-sigmoid.h. Implemented aligned and unaligned variants for sigmoid array processing using a macro pattern similar to hvx-copy.h. Updated act-ops.c to use the new aligned variant hvx_sigmoid_f32_aa. Removed unused hvx-sigmoid.c. * hexagon: factor out hvx-sqrt.h * hexagon: mintor update to hvx-utils.h * hexagon: remove spurios log * hexagon: factor out and optimize hvx_add/sub/mul * hexagon: remove _opt variants of add/sub/mul as they simply fully aligned versions * hexagon: refactor reduction functions to hvx-reduce.h Moved `hvx_self_max_f32` and `hvx_self_sum_f32` from `hvx-utils.h`/`.c` to `hvx-reduce.h`. Renamed them to `hvx_reduce_max_f32` and `hvx_reduce_sum_f32`. Added aligned (`_a`) and unaligned (`_u`) variants and used macros to unify logic. Updated `softmax-ops.c` to use the new functions. * hexagon: refactor the rest of arithmetic functions to hvx-arith.h Moved `hvx_sum_of_squares_f32`, `hvx_min_scalar_f32`, and `hvx_clamp_scalar_f32` from `hvx-utils.c/h` to `hvx-arith.h`. Implemented aligned/unaligned variants (`_aa`, `_au`, etc.) and used macros to reduce code duplication. Updated `hvx_min_scalar_f32` and `hvx_clamp_scalar_f32` to use `dst, src, ..., n` argument order. Updated call sites in `act-ops.c`. Refactor Hexagon HVX arithmetic functions (min, clamp) to hvx-arith.h Moved `hvx_min_scalar_f32` and `hvx_clamp_scalar_f32` from `hvx-utils.c/h` to `hvx-arith.h`. Implemented aligned/unaligned variants (`_aa`, `_au`, etc.) and used macros to reduce code duplication. Updated these functions to use `dst, src, ..., n` argument order and updated call sites in `act-ops.c`. `hvx_sum_of_squares_f32` remains in `hvx-utils.c` as requested. * hexagon: refactor hvx_sum_of_squares_f32 - Modify `hvx_sum_of_squares_f32` in `ggml/src/ggml-hexagon/htp/hvx-reduce.h` to use `dst, src` signature. - Implement `_a` (aligned) and `_u` (unaligned) variants for `hvx_sum_of_squares_f32`. - Update `hvx_reduce_loop_body` macro to support both returning and storing results via `finalize_op`. - Update existing reduction functions in `hvx-reduce.h` to use the updated macro. - Update `rms_norm_htp_f32` in `ggml/src/ggml-hexagon/htp/unary-ops.c` to match the new signature. * hexagon: use hvx_splat instead of memset * hexagon: consistent use of f32/f16 in all function names to match the rest of GGML * hexagon: fix hvx_copy_f16_f32 on v75 and older * hexagon: update readme to include GGML_HEXAGON_EXPERIMENTAL * scripts: update snapdragon/adb scripts to enable host param
This commit is contained in:
Max Krasnyansky
@@ -0,0 +1,132 @@
|
||||
#ifndef HVX_DUMP_H
|
||||
#define HVX_DUMP_H
|
||||
|
||||
#include <HAP_farf.h>
|
||||
|
||||
#include <stdbool.h>
|
||||
#include <stdint.h>
|
||||
|
||||
#include "hex-utils.h"
|
||||
#include "hvx-types.h"
|
||||
|
||||
static void hvx_vec_dump_f16_n(char * pref, HVX_Vector v, uint32_t n) {
|
||||
HVX_VectorAlias u = { .v = v };
|
||||
|
||||
const uint32_t n0 = n / 16;
|
||||
const uint32_t n1 = n % 16;
|
||||
int i = 0;
|
||||
for (; i < n0; i++) {
|
||||
hex_dump_f16_line(pref, u.fp16 + (16 * i), 16);
|
||||
}
|
||||
if (n1) {
|
||||
hex_dump_f16_line(pref, u.fp16 + (16 * i), n1);
|
||||
}
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_f16(char * pref, HVX_Vector v) {
|
||||
hvx_vec_dump_f16_n(pref, v, 64);
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_f32_n(char * pref, HVX_Vector v, uint32_t n) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
float d[32];
|
||||
} u = { .v = v };
|
||||
|
||||
const uint32_t n0 = n / 16;
|
||||
const uint32_t n1 = n % 16;
|
||||
int i = 0;
|
||||
for (; i < n0; i++) {
|
||||
hex_dump_f32_line(pref, u.d + (16 * i), 16);
|
||||
}
|
||||
if (n1) {
|
||||
hex_dump_f32_line(pref, u.d + (16 * i), n1);
|
||||
}
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_f32_hmt(char * pref, HVX_Vector v) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
float d[32];
|
||||
} u = { .v = v };
|
||||
|
||||
FARF(HIGH, "%s: %.6f %.6f %.6f %.6f ... %.6f %.6f %.6f %.6f ... %.6f %.6f %.6f %.6f\n", pref, u.d[0], u.d[1],
|
||||
u.d[2], u.d[3], u.d[12], u.d[13], u.d[14], u.d[15], u.d[28], u.d[29], u.d[30], u.d[31]);
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_f32(char * pref, HVX_Vector v) {
|
||||
hvx_vec_dump_f32_n(pref, v, 32);
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_int32(char * pref, HVX_Vector v) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
int32_t d[32];
|
||||
} u = { .v = v };
|
||||
|
||||
for (int i = 0; i < 32 / 16; i++) {
|
||||
hex_dump_int32_line(pref, u.d + (16 * i), 16);
|
||||
}
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_int32_hmt(char * pref, HVX_Vector v) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
int32_t d[32];
|
||||
} u = { .v = v };
|
||||
|
||||
FARF(HIGH, "%s: %d %d %d %d ... %d %d %d %d ... %d %d %d %d\n", pref, u.d[0], u.d[1], u.d[2], u.d[3], u.d[12],
|
||||
u.d[13], u.d[14], u.d[15], u.d[28], u.d[29], u.d[30], u.d[31]);
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_int8_hmt(char * pref, HVX_Vector v) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
int8_t d[128];
|
||||
} u = { .v = v };
|
||||
|
||||
FARF(HIGH, "%s: %d %d %d %d ... %d %d %d %d ... %d %d %d %d\n", pref, u.d[0], u.d[1], u.d[2], u.d[3], u.d[60],
|
||||
u.d[61], u.d[62], u.d[63], u.d[124], u.d[125], u.d[126], u.d[127]);
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_int8(char * pref, HVX_Vector v) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
int8_t d[128];
|
||||
} u = { .v = v };
|
||||
|
||||
for (int i = 0; i < 128 / 16; i++) {
|
||||
hex_dump_int8_line(pref, u.d + (16 * i), 16);
|
||||
}
|
||||
}
|
||||
|
||||
static void hvx_vec_dump_uint8(char * pref, HVX_Vector v) {
|
||||
union {
|
||||
HVX_Vector v;
|
||||
uint8_t d[128];
|
||||
} u = { .v = v };
|
||||
|
||||
for (int i = 0; i < 128 / 16; i++) {
|
||||
hex_dump_uint8_line(pref, u.d + (16 * i), 16);
|
||||
}
|
||||
}
|
||||
|
||||
static bool hvx_vec_eq(HVX_Vector v0, HVX_Vector v1, size_t n) {
|
||||
typedef union {
|
||||
HVX_Vector v;
|
||||
int8_t d[128];
|
||||
} U;
|
||||
|
||||
U u0 = { .v = v0 };
|
||||
U u1 = { .v = v1 };
|
||||
|
||||
for (int i = 0; i < n; i++) {
|
||||
if (u0.d[i] != u1.d[i]) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
#endif /* HVX_DUMP_H */
|
||||
Reference in New Issue
Block a user