2023-10-30 17:19:15 +00:00
|
|
|
#pragma once
|
|
|
|
|
|
|
|
#include "ggml.h"
|
|
|
|
|
|
|
|
// GGML internal header
|
|
|
|
|
|
|
|
#include <assert.h>
|
2024-01-04 08:12:26 +00:00
|
|
|
#include <stdlib.h> // load `stdlib.h` before other headers to work around MinGW bug: https://sourceforge.net/p/mingw-w64/bugs/192/
|
2023-10-30 17:19:15 +00:00
|
|
|
#include <stddef.h>
|
|
|
|
#include <stdbool.h>
|
|
|
|
#include <string.h> // memcpy
|
|
|
|
#include <math.h> // fabsf
|
|
|
|
|
|
|
|
#ifdef __cplusplus
|
|
|
|
extern "C" {
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// static_assert should be a #define, but if it's not,
|
|
|
|
// fall back to the _Static_assert C11 keyword.
|
|
|
|
// if C99 - static_assert is noop
|
|
|
|
// ref: https://stackoverflow.com/a/53923785/4039976
|
2024-02-05 12:09:47 +00:00
|
|
|
#ifndef __cplusplus
|
2023-10-30 17:19:15 +00:00
|
|
|
#ifndef static_assert
|
|
|
|
#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201100L)
|
|
|
|
#define static_assert(cond, msg) _Static_assert(cond, msg)
|
|
|
|
#else
|
|
|
|
#define static_assert(cond, msg) struct global_scope_noop_trick
|
|
|
|
#endif
|
|
|
|
#endif
|
2024-02-05 12:09:47 +00:00
|
|
|
#endif
|
2023-10-30 17:19:15 +00:00
|
|
|
|
|
|
|
// __FMA__ and __F16C__ are not defined in MSVC, however they are implied with AVX2/AVX512
|
|
|
|
#if defined(_MSC_VER) && (defined(__AVX2__) || defined(__AVX512F__))
|
|
|
|
#ifndef __FMA__
|
|
|
|
#define __FMA__
|
|
|
|
#endif
|
|
|
|
#ifndef __F16C__
|
|
|
|
#define __F16C__
|
|
|
|
#endif
|
|
|
|
#ifndef __SSE3__
|
|
|
|
#define __SSE3__
|
|
|
|
#endif
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// 16-bit float
|
|
|
|
// on Arm, we use __fp16
|
|
|
|
// on x86, we use uint16_t
|
|
|
|
#if defined(__ARM_NEON) && !defined(_MSC_VER)
|
|
|
|
|
|
|
|
// if YCM cannot find <arm_neon.h>, make a symbolic link to it, for example:
|
|
|
|
//
|
|
|
|
// $ ln -sfn /Library/Developer/CommandLineTools/usr/lib/clang/13.1.6/include/arm_neon.h ./src/
|
|
|
|
//
|
|
|
|
#include <arm_neon.h>
|
|
|
|
|
2024-03-11 09:28:51 +00:00
|
|
|
typedef __fp16 ggml_fp16_internal_t;
|
|
|
|
|
2024-02-22 21:21:39 +00:00
|
|
|
#define GGML_COMPUTE_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x)
|
|
|
|
#define GGML_COMPUTE_FP32_TO_FP16(x) ggml_compute_fp32_to_fp16(x)
|
|
|
|
|
|
|
|
#define GGML_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x)
|
|
|
|
|
|
|
|
static inline float ggml_compute_fp16_to_fp32(ggml_fp16_t h) {
|
2024-03-11 09:28:51 +00:00
|
|
|
ggml_fp16_internal_t tmp;
|
2024-02-22 21:21:39 +00:00
|
|
|
memcpy(&tmp, &h, sizeof(ggml_fp16_t));
|
|
|
|
return (float)tmp;
|
|
|
|
}
|
2023-10-30 17:19:15 +00:00
|
|
|
|
2024-02-22 21:21:39 +00:00
|
|
|
static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) {
|
|
|
|
ggml_fp16_t res;
|
2024-03-11 09:28:51 +00:00
|
|
|
ggml_fp16_internal_t tmp = f;
|
2024-02-22 21:21:39 +00:00
|
|
|
memcpy(&res, &tmp, sizeof(ggml_fp16_t));
|
|
|
|
return res;
|
|
|
|
}
|
2023-10-30 17:19:15 +00:00
|
|
|
|
|
|
|
#else
|
|
|
|
|
2024-03-11 09:28:51 +00:00
|
|
|
typedef uint16_t ggml_fp16_internal_t;
|
|
|
|
|
2023-10-30 17:19:15 +00:00
|
|
|
#ifdef __wasm_simd128__
|
|
|
|
#include <wasm_simd128.h>
|
|
|
|
#else
|
|
|
|
#ifdef __POWER9_VECTOR__
|
|
|
|
#include <altivec.h>
|
|
|
|
#undef bool
|
|
|
|
#define bool _Bool
|
|
|
|
#else
|
|
|
|
#if defined(_MSC_VER) || defined(__MINGW32__)
|
|
|
|
#include <intrin.h>
|
|
|
|
#else
|
|
|
|
#if defined(__AVX__) || defined(__AVX2__) || defined(__AVX512F__) || defined(__SSSE3__) || defined(__SSE3__)
|
|
|
|
#if !defined(__riscv)
|
|
|
|
#include <immintrin.h>
|
|
|
|
#endif
|
|
|
|
#endif
|
|
|
|
#endif
|
|
|
|
#endif
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef __riscv_v_intrinsic
|
|
|
|
#include <riscv_vector.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef __F16C__
|
|
|
|
|
|
|
|
#ifdef _MSC_VER
|
|
|
|
#define GGML_COMPUTE_FP16_TO_FP32(x) _mm_cvtss_f32(_mm_cvtph_ps(_mm_cvtsi32_si128(x)))
|
|
|
|
#define GGML_COMPUTE_FP32_TO_FP16(x) _mm_extract_epi16(_mm_cvtps_ph(_mm_set_ss(x), 0), 0)
|
|
|
|
#else
|
|
|
|
#define GGML_COMPUTE_FP16_TO_FP32(x) _cvtsh_ss(x)
|
|
|
|
#define GGML_COMPUTE_FP32_TO_FP16(x) _cvtss_sh(x, 0)
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#elif defined(__POWER9_VECTOR__)
|
|
|
|
|
|
|
|
#define GGML_COMPUTE_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x)
|
|
|
|
#define GGML_COMPUTE_FP32_TO_FP16(x) ggml_compute_fp32_to_fp16(x)
|
|
|
|
/* the inline asm below is about 12% faster than the lookup method */
|
|
|
|
#define GGML_FP16_TO_FP32(x) GGML_COMPUTE_FP16_TO_FP32(x)
|
|
|
|
#define GGML_FP32_TO_FP16(x) GGML_COMPUTE_FP32_TO_FP16(x)
|
|
|
|
|
|
|
|
static inline float ggml_compute_fp16_to_fp32(ggml_fp16_t h) {
|
|
|
|
register float f;
|
|
|
|
register double d;
|
|
|
|
__asm__(
|
|
|
|
"mtfprd %0,%2\n"
|
|
|
|
"xscvhpdp %0,%0\n"
|
|
|
|
"frsp %1,%0\n" :
|
|
|
|
/* temp */ "=d"(d),
|
|
|
|
/* out */ "=f"(f):
|
|
|
|
/* in */ "r"(h));
|
|
|
|
return f;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) {
|
|
|
|
register double d;
|
|
|
|
register ggml_fp16_t r;
|
|
|
|
__asm__( /* xscvdphp can work on double or single precision */
|
|
|
|
"xscvdphp %0,%2\n"
|
|
|
|
"mffprd %1,%0\n" :
|
|
|
|
/* temp */ "=d"(d),
|
|
|
|
/* out */ "=r"(r):
|
|
|
|
/* in */ "f"(f));
|
|
|
|
return r;
|
|
|
|
}
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
|
|
|
// FP16 <-> FP32
|
|
|
|
// ref: https://github.com/Maratyszcza/FP16
|
|
|
|
|
|
|
|
static inline float fp32_from_bits(uint32_t w) {
|
|
|
|
union {
|
|
|
|
uint32_t as_bits;
|
|
|
|
float as_value;
|
|
|
|
} fp32;
|
|
|
|
fp32.as_bits = w;
|
|
|
|
return fp32.as_value;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline uint32_t fp32_to_bits(float f) {
|
|
|
|
union {
|
|
|
|
float as_value;
|
|
|
|
uint32_t as_bits;
|
|
|
|
} fp32;
|
|
|
|
fp32.as_value = f;
|
|
|
|
return fp32.as_bits;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline float ggml_compute_fp16_to_fp32(ggml_fp16_t h) {
|
|
|
|
const uint32_t w = (uint32_t) h << 16;
|
|
|
|
const uint32_t sign = w & UINT32_C(0x80000000);
|
|
|
|
const uint32_t two_w = w + w;
|
|
|
|
|
|
|
|
const uint32_t exp_offset = UINT32_C(0xE0) << 23;
|
|
|
|
#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) || defined(__GNUC__) && !defined(__STRICT_ANSI__)
|
|
|
|
const float exp_scale = 0x1.0p-112f;
|
|
|
|
#else
|
|
|
|
const float exp_scale = fp32_from_bits(UINT32_C(0x7800000));
|
|
|
|
#endif
|
|
|
|
const float normalized_value = fp32_from_bits((two_w >> 4) + exp_offset) * exp_scale;
|
|
|
|
|
|
|
|
const uint32_t magic_mask = UINT32_C(126) << 23;
|
|
|
|
const float magic_bias = 0.5f;
|
|
|
|
const float denormalized_value = fp32_from_bits((two_w >> 17) | magic_mask) - magic_bias;
|
|
|
|
|
|
|
|
const uint32_t denormalized_cutoff = UINT32_C(1) << 27;
|
|
|
|
const uint32_t result = sign |
|
|
|
|
(two_w < denormalized_cutoff ? fp32_to_bits(denormalized_value) : fp32_to_bits(normalized_value));
|
|
|
|
return fp32_from_bits(result);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) {
|
|
|
|
#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) || defined(__GNUC__) && !defined(__STRICT_ANSI__)
|
|
|
|
const float scale_to_inf = 0x1.0p+112f;
|
|
|
|
const float scale_to_zero = 0x1.0p-110f;
|
|
|
|
#else
|
|
|
|
const float scale_to_inf = fp32_from_bits(UINT32_C(0x77800000));
|
|
|
|
const float scale_to_zero = fp32_from_bits(UINT32_C(0x08800000));
|
|
|
|
#endif
|
|
|
|
float base = (fabsf(f) * scale_to_inf) * scale_to_zero;
|
|
|
|
|
|
|
|
const uint32_t w = fp32_to_bits(f);
|
|
|
|
const uint32_t shl1_w = w + w;
|
|
|
|
const uint32_t sign = w & UINT32_C(0x80000000);
|
|
|
|
uint32_t bias = shl1_w & UINT32_C(0xFF000000);
|
|
|
|
if (bias < UINT32_C(0x71000000)) {
|
|
|
|
bias = UINT32_C(0x71000000);
|
|
|
|
}
|
|
|
|
|
|
|
|
base = fp32_from_bits((bias >> 1) + UINT32_C(0x07800000)) + base;
|
|
|
|
const uint32_t bits = fp32_to_bits(base);
|
|
|
|
const uint32_t exp_bits = (bits >> 13) & UINT32_C(0x00007C00);
|
|
|
|
const uint32_t mantissa_bits = bits & UINT32_C(0x00000FFF);
|
|
|
|
const uint32_t nonsign = exp_bits + mantissa_bits;
|
|
|
|
return (sign >> 16) | (shl1_w > UINT32_C(0xFF000000) ? UINT16_C(0x7E00) : nonsign);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define GGML_COMPUTE_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x)
|
|
|
|
#define GGML_COMPUTE_FP32_TO_FP16(x) ggml_compute_fp32_to_fp16(x)
|
|
|
|
|
|
|
|
#endif // __F16C__
|
|
|
|
|
|
|
|
#endif // __ARM_NEON
|
|
|
|
|
|
|
|
// precomputed f32 table for f16 (256 KB)
|
|
|
|
// defined in ggml.c, initialized in ggml_init()
|
|
|
|
extern float ggml_table_f32_f16[1 << 16];
|
|
|
|
|
|
|
|
// On ARM NEON, it's quicker to directly convert x -> x instead of calling into ggml_lookup_fp16_to_fp32,
|
|
|
|
// so we define GGML_FP16_TO_FP32 and GGML_FP32_TO_FP16 elsewhere for NEON.
|
|
|
|
// This is also true for POWER9.
|
2024-02-22 21:21:39 +00:00
|
|
|
#if !defined(GGML_FP16_TO_FP32)
|
2023-10-30 17:19:15 +00:00
|
|
|
inline static float ggml_lookup_fp16_to_fp32(ggml_fp16_t f) {
|
|
|
|
uint16_t s;
|
|
|
|
memcpy(&s, &f, sizeof(uint16_t));
|
|
|
|
return ggml_table_f32_f16[s];
|
|
|
|
}
|
|
|
|
|
|
|
|
#define GGML_FP16_TO_FP32(x) ggml_lookup_fp16_to_fp32(x)
|
2024-02-22 21:21:39 +00:00
|
|
|
#endif
|
2023-10-30 17:19:15 +00:00
|
|
|
|
2024-02-22 21:21:39 +00:00
|
|
|
#if !defined(GGML_FP32_TO_FP16)
|
|
|
|
#define GGML_FP32_TO_FP16(x) GGML_COMPUTE_FP32_TO_FP16(x)
|
2023-10-30 17:19:15 +00:00
|
|
|
#endif
|
|
|
|
|
2023-11-13 12:16:23 +00:00
|
|
|
#define GGML_HASHTABLE_FULL ((size_t)-1)
|
|
|
|
#define GGML_HASHTABLE_ALREADY_EXISTS ((size_t)-2)
|
|
|
|
|
2024-01-12 19:07:38 +00:00
|
|
|
struct ggml_hash_set ggml_hash_set_new(size_t size);
|
|
|
|
|
2023-11-13 12:16:23 +00:00
|
|
|
bool ggml_hash_contains (const struct ggml_hash_set hash_set, struct ggml_tensor * key);
|
|
|
|
|
|
|
|
// returns GGML_HASHTABLE_FULL if table is full, otherwise the current index of the key or where it should be inserted
|
|
|
|
size_t ggml_hash_find (const struct ggml_hash_set hash_set, struct ggml_tensor * key);
|
|
|
|
|
2023-12-07 20:26:54 +00:00
|
|
|
// returns GGML_HASHTABLE_ALREADY_EXISTS if key already exists, index otherwise, asserts if table is full
|
2023-11-13 12:16:23 +00:00
|
|
|
size_t ggml_hash_insert ( struct ggml_hash_set hash_set, struct ggml_tensor * key);
|
|
|
|
|
|
|
|
// return index, asserts if table is full
|
|
|
|
size_t ggml_hash_find_or_insert( struct ggml_hash_set hash_set, struct ggml_tensor * key);
|
2023-10-30 17:19:15 +00:00
|
|
|
|
|
|
|
#ifdef __cplusplus
|
|
|
|
}
|
|
|
|
#endif
|