Skip to content

Commit 50a8290

Browse files
riverlijunjiececiliapeng2011
authored andcommitted
update partition size
1 parent d9a7806 commit 50a8290

File tree

1 file changed

+5
-1
lines changed

1 file changed

+5
-1
lines changed

src/plugins/intel_gpu/src/graph/impls/cm/paged_attention_gen.cpp

Lines changed: 5 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -174,7 +174,11 @@ size_t get_partition_size() {
174174
// k_partition_blok_num = 1;
175175
// const size_t k_partition_blok_num = 16;
176176
// return k_partition_blok_num * PA_KV_CACHE_BLOCK_SIZE; // 128
177-
return 256;
177+
if (PA_KV_CACHE_BLOCK_SIZE < 128) {
178+
return 128;
179+
} else {
180+
return PA_KV_CACHE_BLOCK_SIZE;
181+
}
178182
}
179183

180184
size_t get_partition_num(const size_t kv_len) {

0 commit comments

Comments
 (0)