Skip to content

GitLab

Switch branch/tag
Find file Normal viewHistoryPermalink
flash_fwd_mla_metadata.cu 8.4 KB
Newer Older
zhanghj2's avatar
恢复支持旧接口  
zhanghj2 committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172 
#include "flash_fwd_mla_kernel.h"

static constexpr int MaxBatchSize = 4096;

__global__ void __launch_bounds__(64, 1)
get_mla_metadata_kernel(const Mla_metadata_params params) {
    int *seqlens_k_ptr = params.seqlens_k_ptr;
    int *tile_scheduler_metadata_ptr = params.tile_scheduler_metadata_ptr;
    int *num_splits_ptr = params.num_splits_ptr;
    int batch_size = params.batch_size;
    int block_size_n = params.block_size_n;
    int fixed_overhead_num_blocks = params.fixed_overhead_num_blocks;
    int num_sm_parts = params.num_sm_parts;

    __shared__ int num_blocks_shared[MaxBatchSize];
    __shared__ int num_splits_shared[MaxBatchSize];

    int total_num_blocks = 0;
    for (int i = threadIdx.x; i < batch_size; i += 64) {
        int num_blocks = cutlass::ceil_div(seqlens_k_ptr[i], block_size_n);
        total_num_blocks += num_blocks + fixed_overhead_num_blocks;
        num_blocks_shared[i] = num_blocks;
    }
    for (int offset = 32; offset >= 1; offset /= 2) {
        // total_num_blocks += __shfl_xor(uint32_t(-1), total_num_blocks, offset);
        total_num_blocks += __shfl_xor(total_num_blocks, offset, 64);
    }
    __syncthreads();

    if (threadIdx.x == 0) {
        int payload = cutlass::ceil_div(total_num_blocks, num_sm_parts) + fixed_overhead_num_blocks;

        int now_idx = 0, now_block = 0, now_n_split_idx = 0, cum_num_splits = 0;
        num_splits_shared[0] = 0;
        for (int i = 0; i < num_sm_parts; ++i) {
            int tile_scheduler_metadata0[4], tile_scheduler_metadata1;
            tile_scheduler_metadata0[0] = now_idx;
            tile_scheduler_metadata0[1] = now_block * block_size_n;
            tile_scheduler_metadata1 = now_n_split_idx;
            int remain_payload = payload;
            while (now_idx < batch_size) {
                int num_blocks = num_blocks_shared[now_idx];
                int now_remain_blocks = num_blocks - now_block;
                if (remain_payload >= now_remain_blocks + fixed_overhead_num_blocks) {
                    cum_num_splits += now_n_split_idx + 1;
                    num_splits_shared[now_idx + 1] = cum_num_splits;
                    remain_payload -= now_remain_blocks + fixed_overhead_num_blocks;
                    ++now_idx;
                    now_block = 0;
                    now_n_split_idx = 0;
                } else {
                    if (remain_payload - fixed_overhead_num_blocks > 0) {
                        now_block += remain_payload - fixed_overhead_num_blocks;
                        ++now_n_split_idx;
                        remain_payload = 0;
                    }
                    break;
                }
            }
            tile_scheduler_metadata0[2] = now_block > 0 ? now_idx : now_idx - 1;
            tile_scheduler_metadata0[3] = now_block > 0 ? now_block * block_size_n : seqlens_k_ptr[now_idx - 1];
            *reinterpret_cast<int4 *>(tile_scheduler_metadata_ptr + i * TileSchedulerMetaDataSize) = *reinterpret_cast<int4 *>(tile_scheduler_metadata0);
            tile_scheduler_metadata_ptr[i * TileSchedulerMetaDataSize + 4] = tile_scheduler_metadata1;
        }
        //FLASH_DEVICE_ASSERT(now_idx == batch_size && now_block == 0 && now_n_split_idx == 0);
    }
    __syncthreads();

    for (int i = threadIdx.x; i <= batch_size; i += 64) {
        num_splits_ptr[i] = num_splits_shared[i];
    }
}

void get_mla_metadata_func(Mla_metadata_params &params, cudaStream_t stream) {
    FLASH_ASSERT(params.batch_size < MaxBatchSize);
    get_mla_metadata_kernel<<<1, 64, 0, stream>>>(params);
    CHECK_CUDA_KERNEL_LAUNCH();
}


__global__ void __launch_bounds__(64, 1)
get_mla_decoding_metadata_kernel(const GetDecodingMetadataParams params) {
    int *seqlens_k_ptr = params.seqlens_k_ptr;
    int *tile_scheduler_metadata_ptr = params.tile_scheduler_metadata_ptr;
    int *num_splits_ptr = params.num_splits_ptr;
    int batch_size = params.batch_size;
    int block_size_n = params.block_size_n;
    int fixed_overhead_num_blocks = params.fixed_overhead_num_blocks;
    int num_sm_parts = params.num_sm_parts;

    extern __shared__ int shared_mem[];
    int* num_blocks_shared = shared_mem; // [batch_size]
    int* num_splits_shared = shared_mem + batch_size; // [batch_size+1]
    int* seqlens_k_shared = shared_mem + batch_size*2+1; // [batch_size]
    int* first_block_idx_shared = shared_mem + batch_size*3+1; // [batch_size]
    int* last_block_idx_shared = shared_mem + batch_size*4+1; // [batch_size]

    int total_num_blocks = 0;
    for (int i = threadIdx.x; i < batch_size; i += 64) {
        int cur_s_k = params.topk == -1 ? __ldg(seqlens_k_ptr + i) : params.topk;
        seqlens_k_shared[i] = cur_s_k;
        int first_token_idx = 0;
        int last_token_idx = max(cur_s_k-1, 0);
        int cur_first_block_idx = first_token_idx / block_size_n;
        int cur_last_block_idx = last_token_idx / block_size_n;
        // NOTE Should attend to tokens [first_token_idx, last_token_idx], i.e. blocks [cur_first_block_idx, cur_last_block_idx]
        // NOTE Before clamping, first_token_idx <= last_token_idx always holds, so after clamping, first_token_idx <= last_token_idx still holds.
        // NOTE if seqlens_k is 0, then first_token_idx == last_token_idx == cur_first_block_idx == cur_last_block_idx == 0. So the sequence will have 1 block. We will correct this later in this kernel.
        int num_blocks = cur_last_block_idx - cur_first_block_idx + 1;
        total_num_blocks += num_blocks + fixed_overhead_num_blocks;
        num_blocks_shared[i] = num_blocks;
        first_block_idx_shared[i] = cur_first_block_idx;
        last_block_idx_shared[i] = cur_last_block_idx;
    }
    for (int offset = 32; offset >= 1; offset /= 2) {
        // total_num_blocks += __shfl_xor_sync(uint32_t(-1), total_num_blocks, offset);
        total_num_blocks += __shfl_xor(total_num_blocks, offset, 64);

    }
    __syncthreads();

    if (threadIdx.x == 0) {
        int payload = cutlass::ceil_div(total_num_blocks, num_sm_parts) + fixed_overhead_num_blocks;

        int now_idx = 0, now_block = 0, now_n_split_idx = 0, cum_num_splits = 0;
        num_splits_shared[0] = 0;
        for (int i = 0; i < num_sm_parts; ++i) {
            int tile_scheduler_metadata0[4], tile_scheduler_metadata1;
            tile_scheduler_metadata0[0] = now_idx;
            tile_scheduler_metadata0[1] = now_block + first_block_idx_shared[now_idx];
            tile_scheduler_metadata1 = now_n_split_idx;
            int remain_payload = payload;
            while (now_idx < batch_size) {
                int num_blocks = num_blocks_shared[now_idx];
                int now_remain_blocks = num_blocks - now_block;
                if (remain_payload >= now_remain_blocks + fixed_overhead_num_blocks) {
                    cum_num_splits += now_n_split_idx + 1;
                    num_splits_shared[now_idx + 1] = cum_num_splits;
                    remain_payload -= now_remain_blocks + fixed_overhead_num_blocks;
                    ++now_idx;
                    now_block = 0;
                    now_n_split_idx = 0;
                } else {
                    if (remain_payload - fixed_overhead_num_blocks > 0) {
                        now_block += remain_payload - fixed_overhead_num_blocks;
                        ++now_n_split_idx;
                        remain_payload = 0;
                    }
                    break;
                }
            }
            tile_scheduler_metadata0[2] = now_block > 0 ? now_idx : now_idx - 1;
            tile_scheduler_metadata0[3] = now_block > 0 ? now_block + first_block_idx_shared[now_idx] : (seqlens_k_shared[now_idx-1] == 0 ? 0 : last_block_idx_shared[now_idx-1] + 1);
            *reinterpret_cast<int4 *>(tile_scheduler_metadata_ptr + i * TileSchedulerMetaDataSize) = *reinterpret_cast<int4 *>(tile_scheduler_metadata0);
            tile_scheduler_metadata_ptr[i * TileSchedulerMetaDataSize + 4] = tile_scheduler_metadata1;
        }
        // FLASH_DEVICE_ASSERT(now_idx == batch_size && now_block == 0 && now_n_split_idx == 0);
    }
    __syncthreads();

    for (int i = threadIdx.x; i <= batch_size; i += 64) {
        num_splits_ptr[i] = num_splits_shared[i];
    }
}


void run_get_mla_metadata_kernel(GetDecodingMetadataParams &params, cudaStream_t stream) {
    int smem_size = sizeof(int) * (params.batch_size*5+1);
    // CHECK_CUDA(cudaFuncSetAttribute(get_mla_metadata_kernel, cudaFuncAttributeMaxDynamicSharedMemorySize, smem_size));
    get_mla_decoding_metadata_kernel<<<1, 64, smem_size, stream>>>(params);
    CHECK_CUDA_KERNEL_LAUNCH();
}