cosmopolitan/third_party/ggml/fp16.internal.h

#ifndef COSMOPOLITAN_THIRD_PARTY_GGML_FP16_INTERNAL_H_
#define COSMOPOLITAN_THIRD_PARTY_GGML_FP16_INTERNAL_H_
#include "libc/literal.h"
#include "libc/str/str.h"
#include "third_party/ggml/fp16.h"
#include "third_party/intel/immintrin.internal.h"
#include "third_party/libcxx/math.h"
#if !(__ASSEMBLER__ + __LINKER__ + 0)
COSMOPOLITAN_C_START_

#define GGML_GELU_FP16
#define GGML_SILU_FP16

extern ggml_fp16_t table_gelu_f16[1 << 16];
extern ggml_fp16_t table_silu_f16[1 << 16];
extern ggml_fp16_t table_exp_f16[1 << 16];
extern float table_f32_f16[1 << 16];
#if defined(__ARM_NEON) || defined(__wasm_simd128__)
extern const uint64_t table_b2b_0[1 << 8];
extern const uint64_t table_b2b_1[1 << 8];
#endif

inline static void ggml_vec_gelu_f16(const int n, ggml_fp16_t* y,
                                     const ggml_fp16_t* x) {
  const uint16_t* i16 = (const uint16_t*)x;
  for (int i = 0; i < n; ++i) {
    y[i] = table_gelu_f16[i16[i]];
  }
}

// clang-format off

// 16-bit float
// on Arm, we use __fp16
// on x86, we use uint16_t
#ifdef __ARM_NEON

// 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/
//

#define GGML_COMPUTE_FP16_TO_FP32(x) ((float) (x))
#define GGML_COMPUTE_FP32_TO_FP16(x) (x)

#define GGML_FP16_TO_FP32(x) ((float) (x))
#define GGML_FP32_TO_FP16(x) (x)

#else

#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

// 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.
#if !defined(GGML_FP16_TO_FP32) || !defined(GGML_FP32_TO_FP16)

inline static float ggml_lookup_fp16_to_fp32(ggml_fp16_t f) {
    uint16_t s;
    memcpy(&s, &f, sizeof(uint16_t));
    return table_f32_f16[s];
}

#define GGML_FP16_TO_FP32(x) ggml_lookup_fp16_to_fp32(x)
#define GGML_FP32_TO_FP16(x) GGML_COMPUTE_FP32_TO_FP16(x)

#endif


COSMOPOLITAN_C_END_
#endif /* !(__ASSEMBLER__ + __LINKER__ + 0) */
#endif /* COSMOPOLITAN_THIRD_PARTY_GGML_FP16_INTERNAL_H_ */