llama.cpp k2-fast2: MBZUAI model/K2Horizon + FAST2 sliding-window override

c59fd93 verified about 1 month ago

1.85 kB

#pragma once

#include <cstdint>
#include <memory>
#include <utility>
#include <vector>
#include <cstdio>

struct llama_file;
struct llama_mmap;
struct llama_mlock;

using llama_files  = std::vector<std::unique_ptr<llama_file>>;
using llama_mmaps  = std::vector<std::unique_ptr<llama_mmap>>;
using llama_mlocks = std::vector<std::unique_ptr<llama_mlock>>;

struct llama_file {
    llama_file(const char * fname, const char * mode, bool use_direct_io = false);
    llama_file(FILE * file);
    ~llama_file();

    size_t tell() const;
    size_t size() const;

    int file_id() const; // fileno overload

    void seek(size_t offset, int whence) const;

    void read_raw(void * ptr, size_t len);
    void read_raw_unsafe(void * ptr, size_t len);
    void read_aligned_chunk(void * dest, size_t size);
    uint32_t read_u32();

    void write_raw(const void * ptr, size_t len) const;
    void write_u32(uint32_t val) const;

    size_t read_alignment() const;
    bool has_direct_io() const;
private:
    struct impl;
    std::unique_ptr<impl> pimpl;
};

struct llama_mmap {
    // list of [first, last) byte ranges within a file
    using ranges = std::vector<std::pair<size_t, size_t>>;

    llama_mmap(const llama_mmap &) = delete;
    llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1, bool numa = false,
               const ranges & lazy_ranges = {});
    ~llama_mmap();

    size_t size() const;
    void * addr() const;

    void unmap_fragment(size_t first, size_t last);

    static const bool SUPPORTED;

private:
    struct impl;
    std::unique_ptr<impl> pimpl;
};

struct llama_mlock {
    llama_mlock();
    ~llama_mlock();

    void init(void * ptr);
    void grow_to(size_t target_size);

    static const bool SUPPORTED;

private:
    struct impl;
    std::unique_ptr<impl> pimpl;
};

size_t llama_path_max();

	#pragma once

	#include <cstdint>
	#include <memory>
	#include <utility>
	#include <vector>
	#include <cstdio>

	struct llama_file;
	struct llama_mmap;
	struct llama_mlock;

	using llama_files = std::vector<std::unique_ptr<llama_file>>;
	using llama_mmaps = std::vector<std::unique_ptr<llama_mmap>>;
	using llama_mlocks = std::vector<std::unique_ptr<llama_mlock>>;

	struct llama_file {
	llama_file(const char * fname, const char * mode, bool use_direct_io = false);
	llama_file(FILE * file);
	~llama_file();

	size_t tell() const;
	size_t size() const;

	int file_id() const; // fileno overload

	void seek(size_t offset, int whence) const;

	void read_raw(void * ptr, size_t len);
	void read_raw_unsafe(void * ptr, size_t len);
	void read_aligned_chunk(void * dest, size_t size);
	uint32_t read_u32();

	void write_raw(const void * ptr, size_t len) const;
	void write_u32(uint32_t val) const;

	size_t read_alignment() const;
	bool has_direct_io() const;
	private:
	struct impl;
	std::unique_ptr<impl> pimpl;
	};

	struct llama_mmap {
	// list of [first, last) byte ranges within a file
	using ranges = std::vector<std::pair<size_t, size_t>>;

	llama_mmap(const llama_mmap &) = delete;
	llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1, bool numa = false,
	const ranges & lazy_ranges = {});
	~llama_mmap();

	size_t size() const;
	void * addr() const;

	void unmap_fragment(size_t first, size_t last);

	static const bool SUPPORTED;

	private:
	struct impl;
	std::unique_ptr<impl> pimpl;
	};

	struct llama_mlock {
	llama_mlock();
	~llama_mlock();

	void init(void * ptr);
	void grow_to(size_t target_size);

	static const bool SUPPORTED;

	private:
	struct impl;
	std::unique_ptr<impl> pimpl;
	};

	size_t llama_path_max();