cui-llama.rn 1.2.4 → 1.3.0

package/cpp/ggml-impl.h

@@ -1,209 +1,550 @@
-#pragma once
-
-// GGML internal header
-
-#include "ggml.h"
-
-#include <assert.h>
-#include <stdlib.h> // load `stdlib.h` before other headers to work around MinGW bug: https://sourceforge.net/p/mingw-w64/bugs/192/
-#include <stdbool.h>
-#include <stdint.h>
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#undef MIN
-#undef MAX
-
-#define MIN(a, b) ((a) < (b) ? (a) : (b))
-#define MAX(a, b) ((a) > (b) ? (a) : (b))
-
-// required for mmap as gguf only guarantees 32-byte alignment
-#define TENSOR_ALIGNMENT 32
-
-// 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
-#ifndef __cplusplus
-#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
-#endif
-
-//
-// logging
-//
-
-LM_GGML_ATTRIBUTE_FORMAT(2, 3)
-void lm_ggml_log_internal        (enum lm_ggml_log_level level, const char * format, ...);
-void lm_ggml_log_callback_default(enum lm_ggml_log_level level, const char * text, void * user_data);
-
-#define LM_GGML_LOG(...)       lm_ggml_log_internal(LM_GGML_LOG_LEVEL_NONE , __VA_ARGS__)
-#define LM_GGML_LOG_INFO(...)  lm_ggml_log_internal(LM_GGML_LOG_LEVEL_INFO , __VA_ARGS__)
-#define LM_GGML_LOG_WARN(...)  lm_ggml_log_internal(LM_GGML_LOG_LEVEL_WARN , __VA_ARGS__)
-#define LM_GGML_LOG_ERROR(...) lm_ggml_log_internal(LM_GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
-#define LM_GGML_LOG_DEBUG(...) lm_ggml_log_internal(LM_GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
-#define LM_GGML_LOG_CONT(...)  lm_ggml_log_internal(LM_GGML_LOG_LEVEL_CONT , __VA_ARGS__)
-
-// bitset
-
-typedef uint32_t lm_ggml_bitset_t;
-
-static_assert(sizeof(lm_ggml_bitset_t) == 4, "bitset_t constants must be updated");
-#define BITSET_SHR 5 // log2(sizeof(lm_ggml_bitset_t)*8)
-#define BITSET_MASK (sizeof(lm_ggml_bitset_t)*8 - 1)
-
-static size_t lm_ggml_bitset_size(size_t n) {
-    return (n + BITSET_MASK) >> BITSET_SHR;
-}
-
-static inline bool lm_ggml_bitset_get(const lm_ggml_bitset_t * bitset, size_t i) {
-    return !!(bitset[i >> BITSET_SHR] & (1u << (i & BITSET_MASK)));
-}
-
-static inline void lm_ggml_bitset_set(lm_ggml_bitset_t * bitset, size_t i) {
-    bitset[i >> BITSET_SHR] |= (1u << (i & BITSET_MASK));
-}
-
-static inline void lm_ggml_bitset_clear(lm_ggml_bitset_t * bitset, size_t i) {
-    bitset[i >> BITSET_SHR] &= ~(1u << (i & BITSET_MASK));
-}
-
-// hash set
-
-#define LM_GGML_HASHSET_FULL ((size_t)-1)
-#define LM_GGML_HASHSET_ALREADY_EXISTS ((size_t)-2)
-
-struct lm_ggml_hash_set {
-    size_t size;
-    lm_ggml_bitset_t * used;       // whether or not the keys are in use i.e. set
-    struct lm_ggml_tensor ** keys; // actual tensors in the set, keys[i] is only defined if lm_ggml_bitset_get(used, i)
-};
-
-struct lm_ggml_hash_set lm_ggml_hash_set_new(size_t size);
-void                 lm_ggml_hash_set_free(struct lm_ggml_hash_set * hash_set);
-
-// returns the minimum size for a hash set that can hold min_sz elements
-size_t lm_ggml_hash_size(size_t min_sz);
-
-// remove all elements from the hash set
-void lm_ggml_hash_set_reset(struct lm_ggml_hash_set * hash_set);
-
-// returns true if key is in the hash set
-static bool lm_ggml_hash_contains(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
-
-// returns LM_GGML_HASHSET_FULL if table is full, otherwise the current index of the key or where it should be inserted
-static size_t lm_ggml_hash_find(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
-
-// returns LM_GGML_HASHSET_ALREADY_EXISTS if key already exists, index otherwise, asserts if table is full
-static size_t lm_ggml_hash_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
-
-// return index, asserts if table is full
-static size_t lm_ggml_hash_find_or_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
-
-// hash function for lm_ggml_tensor
-static inline size_t lm_ggml_hash(const struct lm_ggml_tensor * p) {
-    // the last 4 bits are always zero due to alignment
-    return (size_t)(uintptr_t)p >> 4;
-}
-
-static size_t lm_ggml_hash_find(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
-    size_t h = lm_ggml_hash(key) % hash_set->size;
-
-    // linear probing
-    size_t i = h;
-    while (lm_ggml_bitset_get(hash_set->used, i) && hash_set->keys[i] != key) {
-        i = (i + 1) % hash_set->size;
-        if (i == h) {
-            // visited all hash table entries -> not found
-            return LM_GGML_HASHSET_FULL;
-        }
-    }
-    return i;
-}
-
-static bool lm_ggml_hash_contains(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
-    size_t i = lm_ggml_hash_find(hash_set, key);
-    return i != LM_GGML_HASHSET_FULL && lm_ggml_bitset_get(hash_set->used, i);
-}
-
-static size_t lm_ggml_hash_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
-    size_t h = lm_ggml_hash(key) % hash_set->size;
-
-    // linear probing
-    size_t i = h;
-    do {
-        if (!lm_ggml_bitset_get(hash_set->used, i)) {
-            lm_ggml_bitset_set(hash_set->used, i);
-            hash_set->keys[i] = key;
-            return i;
-        }
-        if (hash_set->keys[i] == key) {
-            return LM_GGML_HASHSET_ALREADY_EXISTS;
-        }
-        i = (i + 1) % hash_set->size;
-    } while (i != h);
-
-    // visited all hash table entries -> not found
-    LM_GGML_ABORT("fatal error");
-}
-
-static size_t lm_ggml_hash_find_or_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
-    size_t h = lm_ggml_hash(key) % hash_set->size;
-
-    // linear probing
-    size_t i = h;
-    do {
-        if (!lm_ggml_bitset_get(hash_set->used, i)) {
-            lm_ggml_bitset_set(hash_set->used, i);
-            hash_set->keys[i] = key;
-            return i;
-        }
-        if (hash_set->keys[i] == key) {
-            return i;
-        }
-        i = (i + 1) % hash_set->size;
-    } while (i != h);
-
-    // visited all hash table entries -> not found
-    LM_GGML_ABORT("fatal error");
-}
-
-// computation graph
-
-enum lm_ggml_cgraph_eval_order {
-    LM_GGML_CGRAPH_EVAL_ORDER_LEFT_TO_RIGHT = 0,
-    LM_GGML_CGRAPH_EVAL_ORDER_RIGHT_TO_LEFT,
-    LM_GGML_CGRAPH_EVAL_ORDER_COUNT
-};
-
-struct lm_ggml_cgraph {
-    int size;
-    int n_nodes;
-    int n_leafs;
-
-    struct lm_ggml_tensor ** nodes;
-    struct lm_ggml_tensor ** grads;
-    struct lm_ggml_tensor ** leafs;
-
-    struct lm_ggml_hash_set visited_hash_set;
-
-    enum lm_ggml_cgraph_eval_order order;
-};
-
-struct lm_ggml_cgraph lm_ggml_graph_view(struct lm_ggml_cgraph * cgraph, int i0, int i1);
-
-// Memory allocation
-
-void * lm_ggml_aligned_malloc(size_t size);
-void lm_ggml_aligned_free(void * ptr, size_t size);
-
-#ifdef __cplusplus
-}
-#endif
+#pragma once
+
+// GGML internal header
+
+#include "ggml.h"
+#include <assert.h>
+#include <math.h>
+#include <stdlib.h> // load `stdlib.h` before other headers to work around MinGW bug: https://sourceforge.net/p/mingw-w64/bugs/192/
+#include <stdbool.h>
+#include <stdint.h>
+#include <string.h>
+
+#ifdef __ARM_FEATURE_SVE
+#include <arm_sve.h>
+#endif // __ARM_FEATURE_SVE
+
+#if defined(__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/
+//
+#include <arm_neon.h>
+#endif
+
+#if defined(__F16C__)
+#include <immintrin.h>
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#undef MIN
+#undef MAX
+
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+
+// required for mmap as gguf only guarantees 32-byte alignment
+#define TENSOR_ALIGNMENT 32
+
+// 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
+#ifndef __cplusplus
+    #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
+#endif
+
+static inline int lm_ggml_up32(int n) {
+    return (n + 31) & ~31;
+}
+
+//static inline int lm_ggml_up64(int n) {
+//    return (n + 63) & ~63;
+//}
+
+static inline int lm_ggml_up(int n, int m) {
+    // assert m is a power of 2
+    LM_GGML_ASSERT((m & (m - 1)) == 0);
+    return (n + m - 1) & ~(m - 1);
+}
+
+//
+// logging
+//
+
+LM_GGML_ATTRIBUTE_FORMAT(2, 3)
+void lm_ggml_log_internal        (enum lm_ggml_log_level level, const char * format, ...);
+void lm_ggml_log_callback_default(enum lm_ggml_log_level level, const char * text, void * user_data);
+
+#define LM_GGML_LOG(...)       lm_ggml_log_internal(LM_GGML_LOG_LEVEL_NONE , __VA_ARGS__)
+#define LM_GGML_LOG_INFO(...)  lm_ggml_log_internal(LM_GGML_LOG_LEVEL_INFO , __VA_ARGS__)
+#define LM_GGML_LOG_WARN(...)  lm_ggml_log_internal(LM_GGML_LOG_LEVEL_WARN , __VA_ARGS__)
+#define LM_GGML_LOG_ERROR(...) lm_ggml_log_internal(LM_GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
+#define LM_GGML_LOG_DEBUG(...) lm_ggml_log_internal(LM_GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
+#define LM_GGML_LOG_CONT(...)  lm_ggml_log_internal(LM_GGML_LOG_LEVEL_CONT , __VA_ARGS__)
+
+#define LM_GGML_DEBUG 0
+
+#if (LM_GGML_DEBUG >= 1)
+#define LM_GGML_PRINT_DEBUG(...) LM_GGML_LOG_DEBUG(__VA_ARGS__)
+#else
+#define LM_GGML_PRINT_DEBUG(...)
+#endif
+
+#if (LM_GGML_DEBUG >= 5)
+#define LM_GGML_PRINT_DEBUG_5(...) LM_GGML_LOG_DEBUG(__VA_ARGS__)
+#else
+#define LM_GGML_PRINT_DEBUG_5(...)
+#endif
+
+#if (LM_GGML_DEBUG >= 10)
+#define LM_GGML_PRINT_DEBUG_10(...) LM_GGML_LOG_DEBUG(__VA_ARGS__)
+#else
+#define LM_GGML_PRINT_DEBUG_10(...)
+#endif
+
+// tensor params
+
+static void lm_ggml_set_op_params(struct lm_ggml_tensor * tensor, const void * params, size_t params_size) {
+    LM_GGML_ASSERT(tensor != NULL); // silence -Warray-bounds warnings
+    assert(params_size <= LM_GGML_MAX_OP_PARAMS);
+    memcpy(tensor->op_params, params, params_size);
+}
+
+static int32_t lm_ggml_get_op_params_i32(const struct lm_ggml_tensor * tensor, uint32_t i) {
+    assert(i < LM_GGML_MAX_OP_PARAMS / sizeof(int32_t));
+    return ((const int32_t *)(tensor->op_params))[i];
+}
+
+static float lm_ggml_get_op_params_f32(const struct lm_ggml_tensor * tensor, uint32_t i) {
+    assert(i < LM_GGML_MAX_OP_PARAMS / sizeof(float));
+    return ((const float *)(tensor->op_params))[i];
+}
+
+static void lm_ggml_set_op_params_i32(struct lm_ggml_tensor * tensor, uint32_t i, int32_t value) {
+    assert(i < LM_GGML_MAX_OP_PARAMS / sizeof(int32_t));
+    ((int32_t *)(tensor->op_params))[i] = value;
+}
+
+static void lm_ggml_set_op_params_f32(struct lm_ggml_tensor * tensor, uint32_t i, float value) {
+    assert(i < LM_GGML_MAX_OP_PARAMS / sizeof(float));
+    ((float *)(tensor->op_params))[i] = value;
+}
+
+struct lm_ggml_map_custom1_op_params {
+    lm_ggml_custom1_op_t  fun;
+    int                n_tasks;
+    void             * userdata;
+};
+
+struct lm_ggml_map_custom2_op_params {
+    lm_ggml_custom2_op_t   fun;
+    int                 n_tasks;
+    void              * userdata;
+};
+
+struct lm_ggml_map_custom3_op_params {
+    lm_ggml_custom3_op_t fun;
+    int n_tasks;
+    void * userdata;
+};
+
+// bitset
+
+typedef uint32_t lm_ggml_bitset_t;
+
+static_assert(sizeof(lm_ggml_bitset_t) == 4, "bitset_t constants must be updated");
+#define BITSET_SHR 5 // log2(sizeof(lm_ggml_bitset_t)*8)
+#define BITSET_MASK (sizeof(lm_ggml_bitset_t)*8 - 1)
+
+static size_t lm_ggml_bitset_size(size_t n) {
+    return (n + BITSET_MASK) >> BITSET_SHR;
+}
+
+static inline bool lm_ggml_bitset_get(const lm_ggml_bitset_t * bitset, size_t i) {
+    return !!(bitset[i >> BITSET_SHR] & (1u << (i & BITSET_MASK)));
+}
+
+static inline void lm_ggml_bitset_set(lm_ggml_bitset_t * bitset, size_t i) {
+    bitset[i >> BITSET_SHR] |= (1u << (i & BITSET_MASK));
+}
+
+static inline void lm_ggml_bitset_clear(lm_ggml_bitset_t * bitset, size_t i) {
+    bitset[i >> BITSET_SHR] &= ~(1u << (i & BITSET_MASK));
+}
+
+// hash set
+
+#define LM_GGML_HASHSET_FULL ((size_t)-1)
+#define LM_GGML_HASHSET_ALREADY_EXISTS ((size_t)-2)
+
+struct lm_ggml_hash_set {
+    size_t size;
+    lm_ggml_bitset_t * used;       // whether or not the keys are in use i.e. set
+    struct lm_ggml_tensor ** keys; // actual tensors in the set, keys[i] is only defined if lm_ggml_bitset_get(used, i)
+};
+
+struct lm_ggml_hash_set lm_ggml_hash_set_new(size_t size);
+void                 lm_ggml_hash_set_free(struct lm_ggml_hash_set * hash_set);
+
+// returns the minimum size for a hash set that can hold min_sz elements
+size_t lm_ggml_hash_size(size_t min_sz);
+
+// remove all elements from the hash set
+void lm_ggml_hash_set_reset(struct lm_ggml_hash_set * hash_set);
+
+// returns true if key is in the hash set
+static bool lm_ggml_hash_contains(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
+
+// returns LM_GGML_HASHSET_FULL if table is full, otherwise the current index of the key or where it should be inserted
+static size_t lm_ggml_hash_find(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
+
+// returns LM_GGML_HASHSET_ALREADY_EXISTS if key already exists, index otherwise, asserts if table is full
+static size_t lm_ggml_hash_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
+
+// return index, asserts if table is full
+static size_t lm_ggml_hash_find_or_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key);
+
+// hash function for lm_ggml_tensor
+static inline size_t lm_ggml_hash(const struct lm_ggml_tensor * p) {
+    // the last 4 bits are always zero due to alignment
+    return (size_t)(uintptr_t)p >> 4;
+}
+
+static size_t lm_ggml_hash_find(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
+    size_t h = lm_ggml_hash(key) % hash_set->size;
+
+    // linear probing
+    size_t i = h;
+    while (lm_ggml_bitset_get(hash_set->used, i) && hash_set->keys[i] != key) {
+        i = (i + 1) % hash_set->size;
+        if (i == h) {
+            // visited all hash table entries -> not found
+            return LM_GGML_HASHSET_FULL;
+        }
+    }
+    return i;
+}
+
+static bool lm_ggml_hash_contains(const struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
+    size_t i = lm_ggml_hash_find(hash_set, key);
+    return i != LM_GGML_HASHSET_FULL && lm_ggml_bitset_get(hash_set->used, i);
+}
+
+static size_t lm_ggml_hash_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
+    size_t h = lm_ggml_hash(key) % hash_set->size;
+
+    // linear probing
+    size_t i = h;
+    do {
+        if (!lm_ggml_bitset_get(hash_set->used, i)) {
+            lm_ggml_bitset_set(hash_set->used, i);
+            hash_set->keys[i] = key;
+            return i;
+        }
+        if (hash_set->keys[i] == key) {
+            return LM_GGML_HASHSET_ALREADY_EXISTS;
+        }
+        i = (i + 1) % hash_set->size;
+    } while (i != h);
+
+    // visited all hash table entries -> not found
+    LM_GGML_ABORT("fatal error");
+}
+
+static size_t lm_ggml_hash_find_or_insert(struct lm_ggml_hash_set * hash_set, struct lm_ggml_tensor * key) {
+    size_t h = lm_ggml_hash(key) % hash_set->size;
+
+    // linear probing
+    size_t i = h;
+    do {
+        if (!lm_ggml_bitset_get(hash_set->used, i)) {
+            lm_ggml_bitset_set(hash_set->used, i);
+            hash_set->keys[i] = key;
+            return i;
+        }
+        if (hash_set->keys[i] == key) {
+            return i;
+        }
+        i = (i + 1) % hash_set->size;
+    } while (i != h);
+
+    // visited all hash table entries -> not found
+    LM_GGML_ABORT("fatal error");
+}
+
+// computation graph
+
+enum lm_ggml_cgraph_eval_order {
+    LM_GGML_CGRAPH_EVAL_ORDER_LEFT_TO_RIGHT = 0,
+    LM_GGML_CGRAPH_EVAL_ORDER_RIGHT_TO_LEFT,
+    LM_GGML_CGRAPH_EVAL_ORDER_COUNT
+};
+
+struct lm_ggml_cgraph {
+    int size;
+    int n_nodes;
+    int n_leafs;
+
+    struct lm_ggml_tensor ** nodes;
+    struct lm_ggml_tensor ** grads;
+    struct lm_ggml_tensor ** leafs;
+
+    struct lm_ggml_hash_set visited_hash_set;
+
+    enum lm_ggml_cgraph_eval_order order;
+};
+
+struct lm_ggml_cgraph lm_ggml_graph_view(struct lm_ggml_cgraph * cgraph, int i0, int i1);
+
+// Memory allocation
+
+void * lm_ggml_aligned_malloc(size_t size);
+void lm_ggml_aligned_free(void * ptr, size_t size);
+
+// FP16 to FP32 conversion
+
+#if defined(__ARM_NEON)
+    #ifdef _MSC_VER
+        typedef uint16_t lm_ggml_fp16_internal_t;
+    #else
+        typedef __fp16 lm_ggml_fp16_internal_t;
+    #endif
+#endif
+
+#if defined(__ARM_NEON) && !defined(_MSC_VER)
+    #define LM_GGML_COMPUTE_FP16_TO_FP32(x) lm_ggml_compute_fp16_to_fp32(x)
+    #define LM_GGML_COMPUTE_FP32_TO_FP16(x) lm_ggml_compute_fp32_to_fp16(x)
+
+    #define LM_GGML_FP16_TO_FP32(x) lm_ggml_compute_fp16_to_fp32(x)
+
+    static inline float lm_ggml_compute_fp16_to_fp32(lm_ggml_fp16_t h) {
+        lm_ggml_fp16_internal_t tmp;
+        memcpy(&tmp, &h, sizeof(lm_ggml_fp16_t));
+        return (float)tmp;
+    }
+
+    static inline lm_ggml_fp16_t lm_ggml_compute_fp32_to_fp16(float f) {
+        lm_ggml_fp16_t res;
+        lm_ggml_fp16_internal_t tmp = f;
+        memcpy(&res, &tmp, sizeof(lm_ggml_fp16_t));
+        return res;
+    }
+
+#elif defined(__F16C__)
+
+    #ifdef _MSC_VER
+        #define LM_GGML_COMPUTE_FP16_TO_FP32(x) _mm_cvtss_f32(_mm_cvtph_ps(_mm_cvtsi32_si128(x)))
+        #define LM_GGML_COMPUTE_FP32_TO_FP16(x) _mm_extract_epi16(_mm_cvtps_ph(_mm_set_ss(x), 0), 0)
+    #else
+        #define LM_GGML_COMPUTE_FP16_TO_FP32(x) _cvtsh_ss(x)
+        #define LM_GGML_COMPUTE_FP32_TO_FP16(x) _cvtss_sh(x, 0)
+    #endif
+
+#elif defined(__POWER9_VECTOR__)
+
+    #define LM_GGML_COMPUTE_FP16_TO_FP32(x) lm_ggml_compute_fp16_to_fp32(x)
+    #define LM_GGML_COMPUTE_FP32_TO_FP16(x) lm_ggml_compute_fp32_to_fp16(x)
+    /* the inline asm below is about 12% faster than the lookup method */
+    #define LM_GGML_FP16_TO_FP32(x) LM_GGML_COMPUTE_FP16_TO_FP32(x)
+    #define LM_GGML_FP32_TO_FP16(x) LM_GGML_COMPUTE_FP32_TO_FP16(x)
+
+    static inline float lm_ggml_compute_fp16_to_fp32(lm_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 lm_ggml_fp16_t lm_ggml_compute_fp32_to_fp16(float f) {
+        register double d;
+        register lm_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 lm_ggml_compute_fp16_to_fp32(lm_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__)) && (!defined(__cplusplus) || __cplusplus >= 201703L)
+        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 lm_ggml_fp16_t lm_ggml_compute_fp32_to_fp16(float f) {
+    #if (defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) || defined(__GNUC__) && !defined(__STRICT_ANSI__)) && (!defined(__cplusplus) || __cplusplus >= 201703L)
+        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 LM_GGML_COMPUTE_FP16_TO_FP32(x) lm_ggml_compute_fp16_to_fp32(x)
+    #define LM_GGML_COMPUTE_FP32_TO_FP16(x) lm_ggml_compute_fp32_to_fp16(x)
+
+#endif // defined(__ARM_NEON) && (!defined(__MSC_VER)
+
+// precomputed f32 table for f16 (256 KB)
+// defined in ggml.c, initialized in lm_ggml_init()
+LM_GGML_API float lm_ggml_table_f32_f16[1 << 16];
+
+// On ARM NEON, it's quicker to directly convert x -> x instead of calling into lm_ggml_lookup_fp16_to_fp32,
+// so we define LM_GGML_FP16_TO_FP32 and LM_GGML_FP32_TO_FP16 elsewhere for NEON.
+// This is also true for POWER9.
+#if !defined(LM_GGML_FP16_TO_FP32)
+inline static float lm_ggml_lookup_fp16_to_fp32(lm_ggml_fp16_t f) {
+    uint16_t s;
+    memcpy(&s, &f, sizeof(uint16_t));
+    return lm_ggml_table_f32_f16[s];
+}
+
+#define LM_GGML_FP16_TO_FP32(x) lm_ggml_lookup_fp16_to_fp32(x)
+#endif
+
+#if !defined(LM_GGML_FP32_TO_FP16)
+#define LM_GGML_FP32_TO_FP16(x) LM_GGML_COMPUTE_FP32_TO_FP16(x)
+#endif
+
+/**
+ * Converts brain16 to float32.
+ *
+ * The bfloat16 floating point format has the following structure:
+ *
+ *       ┌sign
+ *       │
+ *       │   ┌exponent
+ *       │   │
+ *       │   │      ┌mantissa
+ *       │   │      │
+ *       │┌──┴───┐┌─┴───┐
+ *     0b0000000000000000 brain16
+ *
+ * Since bf16 has the same number of exponent bits as a 32bit float,
+ * encoding and decoding numbers becomes relatively straightforward.
+ *
+ *       ┌sign
+ *       │
+ *       │   ┌exponent
+ *       │   │
+ *       │   │      ┌mantissa
+ *       │   │      │
+ *       │┌──┴───┐┌─┴───────────────────┐
+ *     0b00000000000000000000000000000000 IEEE binary32
+ *
+ * For comparison, the standard fp16 format has fewer exponent bits.
+ *
+ *       ┌sign
+ *       │
+ *       │  ┌exponent
+ *       │  │
+ *       │  │    ┌mantissa
+ *       │  │    │
+ *       │┌─┴─┐┌─┴──────┐
+ *     0b0000000000000000 IEEE binary16
+ *
+ * @see IEEE 754-2008
+ */
+static inline float lm_ggml_compute_bf16_to_fp32(lm_ggml_bf16_t h) {
+    union {
+        float f;
+        uint32_t i;
+    } u;
+    u.i = (uint32_t)h.bits << 16;
+    return u.f;
+}
+
+/**
+ * Converts float32 to brain16.
+ *
+ * This is binary identical with Google Brain float conversion.
+ * Floats shall round to nearest even, and NANs shall be quiet.
+ * Subnormals aren't flushed to zero, except perhaps when used.
+ * This code should vectorize nicely if using modern compilers.
+ */
+static inline lm_ggml_bf16_t lm_ggml_compute_fp32_to_bf16(float s) {
+    lm_ggml_bf16_t h;
+    union {
+        float f;
+        uint32_t i;
+    } u;
+    u.f = s;
+    if ((u.i & 0x7fffffff) > 0x7f800000) { /* nan */
+        h.bits = (u.i >> 16) | 64; /* force to quiet */
+        return h;
+    }
+    h.bits = (u.i + (0x7fff + ((u.i >> 16) & 1))) >> 16;
+    return h;
+}
+
+#define LM_GGML_FP32_TO_BF16(x) lm_ggml_compute_fp32_to_bf16(x)
+#define LM_GGML_BF16_TO_FP32(x) lm_ggml_compute_bf16_to_fp32(x)
+
+#ifdef __cplusplus
+}
+#endif