优化bf16精度

csrc/attention/attention_kernels_opt_tc.cu

@@ -124,11 +124,11 @@ inline __device__ void v_mmac_f32_16x16x16_f16(const half4_t& reg_a, const half4
 {
     
     if constexpr (is_half){
-     asm volatile("v_mmac_f32_16x16x16_f16 %0, %1, %2, %0" : 
+     asm volatile("\n s_nop 1 \n v_mmac_f32_16x16x16_f16 %0, %1, %2, %0" : 
              "=v"(reg_c) : "v"(reg_a), "v"(reg_b), "0"(reg_c));
     }
     else{
-     asm volatile("v_mmac_f32_16x16x16_bf16 %0, %1, %2, %0" : 
+     asm volatile("\n s_nop 1 \n v_mmac_f32_16x16x16_bf16 %0, %1, %2, %0" : 
        "=v"(reg_c) : "v"(reg_a), "v"(reg_b), "0"(reg_c));
     }
 }
@@ -147,7 +147,7 @@ inline __device__ void builtin_amdgcn_mmac(const half4_t& reg_a, const half4_t&
 template <typename scalar_t, typename cache_t, int HEAD_SIZE, int BLOCK_SIZE,
           int NUM_THREADS, vllm::Fp8KVCacheDataType KV_DTYPE,
           bool IS_BLOCK_SPARSE,int REUSE_KV_TIMES>  // Zero means no partitioning.
-__global__ void paged_attention_kernel_TC_with_mask(
+__global__ void paged_attention_kernel_TC(
     float* __restrict__ exp_sums,  // [num_seqs, num_heads, max_num_partitions]
     float* __restrict__ max_logits,  // [num_seqs, num_heads, max_num_partitions]
     scalar_t* __restrict__ out,  // [num_seqs, num_heads,head_size]
@@ -225,10 +225,12 @@ __global__ void paged_attention_kernel_TC_with_mask(
     for(int i=0;i<q_boundary;i++){
     if(thread_idx<16){
       half4x2 temp = *reinterpret_cast<const half4x2*>(q_ptr+i*HEAD_SIZE+thread_idx*8);
-      #pragma unroll
-      for(int k=0;k<4;k++){
-        temp.data[0][k]=((float)temp.data[0][k])*scale;
-        temp.data[1][k]=((float)temp.data[1][k])*scale;
+      if constexpr(is_half){
+        scalar_t *t=reinterpret_cast<scalar_t*>(&temp);
+        #pragma unroll
+        for(int k=0;k<8;k++){
+          from_float(t[k],to_float(t[k])*scale);
+        }
       }
       q_vecs[i][thread_idx]=temp;
     }
@@ -265,6 +267,9 @@ __global__ void paged_attention_kernel_TC_with_mask(
       int reuse_kv_idx=rows+i*4;
       if(reuse_kv_idx<REUSE_KV_TIMES){
         if(reuse_kv_idx>=q_boundary)qk_vec[i]=0;
+        else {
+          if constexpr(!is_half) qk_vec[i]*=scale;
+        }
         const int token_idx = block_idx * BLOCK_SIZE+rowid;
         if(alibi_slope[i] != 0){
           float alibi=alibi_slope[i]* (token_idx - seq_len + 1);
@@ -764,16 +769,16 @@ __global__ __launch_bounds__(NUM_THREADS, 1) void paged_attention_v2_reduce_kern
 
 #define LAUNCH_PAGED_ATTENTION_V2_TC(HEAD_SIZE)                                      \
   hipLaunchKernelGGL(                                                                \
-      (vllm::paged_attention_kernel_TC_with_mask<                                    \
+      (vllm::paged_attention_kernel_TC<                                              \
           T, CACHE_T, HEAD_SIZE, BLOCK_SIZE, NUM_THREADS, KV_DTYPE,                  \
           IS_BLOCK_SPARSE, REUSE_KV_TIMES>),                                         \
       dim3(grid), dim3(block), shared_mem_size, stream, exp_sums_ptr,                \
       max_logits_ptr,out_ptr, tmp_out_ptr, query_ptr, key_cache_ptr, value_cache_ptr,\
       num_heads, num_kv_heads, scale, block_tables_ptr, seq_lens_ptr,                \
       max_num_blocks_per_seq, alibi_slopes_ptr, q_stride, kv_block_stride,           \
-      kv_head_stride, k_scale_ptr, v_scale_ptr, tp_rank, blocksparse_local_blocks,           \
+      kv_head_stride, k_scale_ptr, v_scale_ptr, tp_rank, blocksparse_local_blocks,   \
       blocksparse_vert_stride, blocksparse_block_size,                               \
-      blocksparse_head_sliding_step,PARTITION_SIZE);\
+      blocksparse_head_sliding_step,PARTITION_SIZE);                                 \
   if (max_num_partitions<=64&&max_num_partitions>1){                                 \
       hipLaunchKernelGGL(                                                            \
       (vllm::paged_attention_v2_reduce_kernel_opt_tc<T, HEAD_SIZE, 64>),             \