LshProjection

局部敏感哈希（LSH）投影。对每个哈希组，用 bits_per_hash 个种子对输入特征做加权哈希评分，按评分符号拼出一组 int32 哈希签名。

设哈希组数为 \(G\)，即 hash_group_num；每组位数为 \(B\)，即 bits_per_hash；特征维数为 \(F\)，即 feature_num。hash_seed 长度为 \(G \cdot B\)， feature / weight 长度为 \(F\)，output 长度为 \(G\)。

对组下标 \(i = 0,\ldots,G-1\)：

\[\begin{split}\begin{aligned} \mathrm{hash\_sign} &\leftarrow 0 \\ \text{对 } j = 0,\ldots,B-1\text{：}\quad s &\leftarrow \mathrm{hash\_seed}[i\cdot B + j] \\ \mathrm{score} &\leftarrow \sum_{k=0}^{F-1} w_k \cdot \mathrm{float}\!\big(\mathrm{FNV1a}_{32}(s \Vert \mathrm{feature}[k])\big) \\ b &\leftarrow \begin{cases} 1, & \mathrm{score} > 0 \\ 0, & \text{otherwise} \end{cases} \\ \mathrm{hash\_sign} &\leftarrow (\mathrm{hash\_sign}\ll 1) \lor b \\ \mathrm{output}[i] &\leftarrow \mathrm{hash\_sign} \end{aligned}\end{split}\]

其中 \(s \Vert \mathrm{feature}[k]\) 表示将 float 种子与 int32 特征按字节拼接共 8 字节后做 FNV1a-32。若 weight == NULL，则 \(w_k = 1\)；否则 \(w_k = \mathrm{weight}[k]\)。

输入：

hash_seed - 哈希种子数组，类型 float，长度 hash_group_num * bits_per_hash
feature - 特征向量，类型 int32，长度 feature_num
weight - 权重向量，长度 feature_num；类型随接口前缀变化；可为 NULL
hash_group_num - 哈希组数 \(G\)，即输出长度
bits_per_hash - 每组有效位数 \(B\)，通常不超过 32
feature_num - 特征维数 \(F\)
core_mask - 核掩码（仅共享存储版本使用）

输出：

output - 哈希签名数组，类型 int32，长度 hash_group_num

支持平台：

FT78NE MT7004

备注

特征 feature 与输出 output 固定为 int32
FT78NE 权重类型：int8、int16、int32、fp32、fp64
MT7004 权重类型：int16、int32、fp16、fp32
内部哈希为 FNV1a-32；weight 为 NULL 时不加权重

共享存储版本:

void i8_lsh_projection_s(const float *hash_seed, const int32_t *feature, const int8_t *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num, int core_mask)

void i16_lsh_projection_s(const float *hash_seed, const int32_t *feature, const int16_t *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num, int core_mask)

void i32_lsh_projection_s(const float *hash_seed, const int32_t *feature, const int32_t *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num, int core_mask)

void hp_lsh_projection_s(const float *hash_seed, const int32_t *feature, const float16 *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num, int core_mask)

void fp_lsh_projection_s(const float *hash_seed, const int32_t *feature, const float *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num, int core_mask)

void dp_lsh_projection_s(const float *hash_seed, const int32_t *feature, const double *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num, int core_mask)

C调用示例：

// MT7004 示例（共享存储多核，DDR 地址）
void TestLshProjectionSMCFp32(int hash_group_num, int core_mask) {
    int core_id = get_core_id();
    int logic_core_id = GetLogicCoreId(core_mask, core_id);
    int core_num = GetCoreNum(core_mask);
    float *hash_seed = (float *)0x88000000;
    int32_t *feature = (int32_t *)0x8A000000;
    float *weight = (float *)0x8A001000;
    int32_t *output = (int32_t *)0x90000000;
    int bits_per_hash = 4;
    int feature_num = 5;
    sys_bar(0, core_num);
    fp_lsh_projection_s(hash_seed, feature, weight, output, hash_group_num, bits_per_hash, feature_num, core_mask);
}

void main() {
    int core_mask = 0b1111;
    TestLshProjectionSMCFp32(256, core_mask);
}

私有存储版本:

void i8_lsh_projection_p(const float *hash_seed, const int32_t *feature, const int8_t *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num)

void i16_lsh_projection_p(const float *hash_seed, const int32_t *feature, const int16_t *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num)

void i32_lsh_projection_p(const float *hash_seed, const int32_t *feature, const int32_t *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num)

void hp_lsh_projection_p(const float *hash_seed, const int32_t *feature, const float16 *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num)

void fp_lsh_projection_p(const float *hash_seed, const int32_t *feature, const float *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num)

void dp_lsh_projection_p(const float *hash_seed, const int32_t *feature, const double *weight, int32_t *output, int hash_group_num, int bits_per_hash, int feature_num)

C调用示例：

// MT7004 示例（私有存储单核，AM 地址）
void TestLshProjectionAMFp32(int hash_group_num) {
    float *hash_seed = (float *)0x10000000;
    int32_t *feature = (int32_t *)0x10020000;
    float *weight = (float *)0x10030000;
    int32_t *output = (int32_t *)0x10040000;
    int bits_per_hash = 4;
    int feature_num = 5;
    fp_lsh_projection_p(hash_seed, feature, weight, output, hash_group_num, bits_per_hash, feature_num);
}

void main() {
    TestLshProjectionAMFp32(256);
}