/src/llama.cpp/src/llama-impl.h

Source
#pragma once

#include "ggml.h" // for ggml_log_level

#include <string>
#include <type_traits>
#include <vector>

#ifdef __GNUC__
#    if defined(__MINGW32__) && !defined(__clang__)
#        define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(gnu_printf, __VA_ARGS__)))
#    else
#        define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
#    endif
#else
#    define LLAMA_ATTRIBUTE_FORMAT(...)
#endif

//
// logging
//

LLAMA_ATTRIBUTE_FORMAT(2, 3)
void llama_log_internal        (ggml_log_level level, const char * format, ...);
void llama_log_callback_default(ggml_log_level level, const char * text, void * user_data);

#define LLAMA_LOG(...)       llama_log_internal(GGML_LOG_LEVEL_NONE , __VA_ARGS__)
#define LLAMA_LOG_INFO(...)  llama_log_internal(GGML_LOG_LEVEL_INFO , __VA_ARGS__)
#define LLAMA_LOG_WARN(...)  llama_log_internal(GGML_LOG_LEVEL_WARN , __VA_ARGS__)
#define LLAMA_LOG_ERROR(...) llama_log_internal(GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
#define LLAMA_LOG_DEBUG(...) llama_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
#define LLAMA_LOG_CONT(...)  llama_log_internal(GGML_LOG_LEVEL_CONT , __VA_ARGS__)

//
// helpers
//

template <typename T>
struct no_init {
    T value;
    no_init() = default;
};

template <typename dst_t, typename src_t>
static inline dst_t llama_cast(src_t v) {
    if constexpr (std::is_same_v<src_t, dst_t>) {
        return v;
    } else if constexpr (std::is_same_v<src_t, ggml_fp16_t> && std::is_same_v<dst_t, float>) {
        return ggml_fp16_to_fp32(v);
    } else if constexpr (std::is_same_v<src_t, float> && std::is_same_v<dst_t, ggml_fp16_t>) {
        return ggml_fp32_to_fp16(v);
    } else {
        static_assert(std::is_same_v<dst_t, void>, "unsupported type combination");
    }
}

static inline ggml_tensor * llama_mul_mat_hadamard(
        ggml_context * ctx,
        ggml_tensor * cur,
        ggml_tensor * rot) {
    const auto n = rot->ne[0];

    ggml_tensor * res;

    if (!ggml_is_contiguous(cur)) {
        res = ggml_cont_2d(ctx, cur, n, ggml_nelements(cur)/n);
    } else {
        res = ggml_reshape_2d(ctx, cur, n, ggml_nelements(cur)/n);
    }
    res = ggml_mul_mat(ctx, rot, res);
    ggml_mul_mat_set_hint(res, GGML_HINT_SRC0_IS_HADAMARD);
    res = ggml_reshape_4d(ctx, res, cur->ne[0], cur->ne[1], cur->ne[2], cur->ne[3]);

    return res;
}

struct time_meas {
    time_meas(int64_t & t_acc, bool disable = false);
    ~time_meas();

    const int64_t t_start_us;

    int64_t & t_acc;
};

template <typename T>
struct buffer_view {
    T * data;
    size_t size = 0;

    bool has_data() const {
        return data && size > 0;
    }
};

void replace_all(std::string & s, const std::string & search, const std::string & replace);

// TODO: rename to llama_format ?
LLAMA_ATTRIBUTE_FORMAT(1, 2)
std::string format(const char * fmt, ...);

std::string llama_format_tensor_shape(const std::vector<int64_t> & ne);
std::string llama_format_tensor_shape(const struct ggml_tensor * t);

std::string gguf_kv_to_str(const struct gguf_context * ctx_gguf, int i);

Line	Count	Source
1		#pragma once
2
3		#include "ggml.h" // for ggml_log_level
4
5		#include <string>
6		#include <type_traits>
7		#include <vector>
8
9		#ifdef __GNUC__
10		# if defined(__MINGW32__) && !defined(__clang__)
11		# define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(gnu_printf, __VA_ARGS__)))
12		# else
13		# define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
14		# endif
15		#else
16		# define LLAMA_ATTRIBUTE_FORMAT(...)
17		#endif
18
19		//
20		// logging
21		//
22
23		LLAMA_ATTRIBUTE_FORMAT(2, 3)
24		void llama_log_internal (ggml_log_level level, const char * format, ...);
25		void llama_log_callback_default(ggml_log_level level, const char * text, void * user_data);
26
27		#define LLAMA_LOG(...) llama_log_internal(GGML_LOG_LEVEL_NONE , __VA_ARGS__)
28	9.62k	#define LLAMA_LOG_INFO(...) llama_log_internal(GGML_LOG_LEVEL_INFO , __VA_ARGS__) Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_0::operator()(unsigned int) const Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_1::operator()(unsigned int) const Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_2::operator()(unsigned int) const Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_3::operator()(unsigned int) const Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_4::operator()(unsigned int) const Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_5::operator()(unsigned int) const Unexecuted instantiation: llama-model.cpp:llama_model::print_info() const::$_6::operator()(unsigned int) const
29	3	#define LLAMA_LOG_WARN(...) llama_log_internal(GGML_LOG_LEVEL_WARN , __VA_ARGS__)
30	8.68k	#define LLAMA_LOG_ERROR(...) llama_log_internal(GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
31	0	#define LLAMA_LOG_DEBUG(...) llama_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
32	0	#define LLAMA_LOG_CONT(...) llama_log_internal(GGML_LOG_LEVEL_CONT , __VA_ARGS__)
33
34		//
35		// helpers
36		//
37
38		template <typename T>
39		struct no_init {
40		T value;
41		no_init() = default;
42		};
43
44		template <typename dst_t, typename src_t>
45	0	static inline dst_t llama_cast(src_t v) {
46	0	if constexpr (std::is_same_v<src_t, dst_t>) {
47	0	return v;
48	0	} else if constexpr (std::is_same_v<src_t, ggml_fp16_t> && std::is_same_v<dst_t, float>) {
49	0	return ggml_fp16_to_fp32(v);
50	0	} else if constexpr (std::is_same_v<src_t, float> && std::is_same_v<dst_t, ggml_fp16_t>) {
51	0	return ggml_fp32_to_fp16(v);
52		} else {
53		static_assert(std::is_same_v<dst_t, void>, "unsupported type combination");
54		}
55	0	} Unexecuted instantiation: llama-graph.cpp:unsigned short llama_cast<unsigned short, float>(float) Unexecuted instantiation: llama-graph.cpp:float llama_cast<float, unsigned short>(unsigned short) Unexecuted instantiation: llama-graph.cpp:float llama_cast<float, float>(float) Unexecuted instantiation: llama-kv-cache.cpp:unsigned short llama_cast<unsigned short, float>(float) Unexecuted instantiation: llama-kv-cache.cpp:float llama_cast<float, float>(float)
56
57		static inline ggml_tensor * llama_mul_mat_hadamard(
58		ggml_context * ctx,
59		ggml_tensor * cur,
60	0	ggml_tensor * rot) {
61	0	const auto n = rot->ne[0];
62
63	0	ggml_tensor * res;
64
65	0	if (!ggml_is_contiguous(cur)) {
66	0	res = ggml_cont_2d(ctx, cur, n, ggml_nelements(cur)/n);
67	0	} else {
68	0	res = ggml_reshape_2d(ctx, cur, n, ggml_nelements(cur)/n);
69	0	}
70	0	res = ggml_mul_mat(ctx, rot, res);
71	0	ggml_mul_mat_set_hint(res, GGML_HINT_SRC0_IS_HADAMARD);
72	0	res = ggml_reshape_4d(ctx, res, cur->ne[0], cur->ne[1], cur->ne[2], cur->ne[3]);
73
74	0	return res;
75	0	} Unexecuted instantiation: llama.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-adapter.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-arch.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-batch.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-context.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-graph.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-impl.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-kv-cache.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-kv-cache-iswa.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-kv-cache-dsa.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-kv-cache-dsv4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-memory-hybrid.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-memory-hybrid-iswa.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-memory-recurrent.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-mmap.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-model-loader.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-model.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-sampler.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-vocab.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: afmoe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: apertus.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: arcee.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: arctic.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: arwkv7.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: baichuan.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: bailingmoe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: bailingmoe2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: bert.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: bitnet.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: bloom.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: chameleon.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: chatglm.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: codeshell.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: cogvlm.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: cohere2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: cohere2moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: command-r.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: dbrx.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: deci.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: deepseek.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: deepseek2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: deepseek2ocr.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: deepseek32.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: deepseek4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: dflash.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: dots1.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: dream.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: eagle3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: ernie4-5-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: ernie4-5.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: eurobert.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: exaone-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: exaone.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: exaone4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: falcon-h1.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: falcon.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma-embedding.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma3n.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma4-assistant.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gemma4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: glm-dsa.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: glm4-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: glm4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gpt2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: gptneox.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: granite-hybrid.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: granite-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: granite.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: grok.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: grovemoe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: hunyuan-dense.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: hunyuan-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: hunyuan-vl.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: hy-v3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: internlm2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: jais.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: jais2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: jamba.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: jina-bert-v2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: jina-bert-v3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: kimi-linear.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: lfm2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: lfm2moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llada-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llada.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-embed.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: maincoder.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mamba-base.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mamba.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mamba2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mellum.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mimo2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: minicpm.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: minicpm3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: minimax-m2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mistral3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mistral4.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: modern-bert.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: mpt.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: nemotron-h-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: nemotron-h.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: nemotron.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: neo-bert.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: nomic-bert-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: nomic-bert.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: olmo.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: olmo2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: olmoe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: openai-moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: openelm.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: orion.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: paddleocr.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: pangu-embed.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: phi2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: phi3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: phimoe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: plamo.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: plamo2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: plamo3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: plm.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen2moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen2vl.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen35.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen35moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen3moe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen3next.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen3vl.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: qwen3vlmoe.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: refact.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: rnd1.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: rwkv6.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: rwkv6qwen2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: rwkv7-base.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: rwkv7.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: seed-oss.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: smallthinker.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: smollm3.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: stablelm.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: starcoder.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: starcoder2.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: step35.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: t5.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: t5encoder.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: talkie.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: wavtokenizer-dec.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: xverse.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: llama-grammar.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: delta-net-base.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor) Unexecuted instantiation: rwkv6-base.cpp:llama_mul_mat_hadamard(ggml_context, ggml_tensor, ggml_tensor)
76
77		struct time_meas {
78		time_meas(int64_t & t_acc, bool disable = false);
79		~time_meas();
80
81		const int64_t t_start_us;
82
83		int64_t & t_acc;
84		};
85
86		template <typename T>
87		struct buffer_view {
88		T * data;
89		size_t size = 0;
90
91	0	bool has_data() const {
92	0	return data && size > 0;
93	0	} Unexecuted instantiation: buffer_view<float>::has_data() const Unexecuted instantiation: buffer_view<int>::has_data() const
94		};
95
96		void replace_all(std::string & s, const std::string & search, const std::string & replace);
97
98		// TODO: rename to llama_format ?
99		LLAMA_ATTRIBUTE_FORMAT(1, 2)
100		std::string format(const char * fmt, ...);
101
102		std::string llama_format_tensor_shape(const std::vector<int64_t> & ne);
103		std::string llama_format_tensor_shape(const struct ggml_tensor * t);
104
105		std::string gguf_kv_to_str(const struct gguf_context * ctx_gguf, int i);

Coverage Report

Created: 2026-07-16 06:35