From 7cdca79ade365adc2a7a37afef96acd6f51a1485 Mon Sep 17 00:00:00 2001 From: Kimish Patel Date: Fri, 21 Mar 2025 07:27:13 -0700 Subject: [PATCH] Fix * 4 debug fix that got accidentally landed in op_sdpa.cpp As the title. Test: CI --- extension/llm/custom_ops/op_sdpa.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extension/llm/custom_ops/op_sdpa.cpp b/extension/llm/custom_ops/op_sdpa.cpp index 89634574027..14b2bc694a6 100644 --- a/extension/llm/custom_ops/op_sdpa.cpp +++ b/extension/llm/custom_ops/op_sdpa.cpp @@ -382,7 +382,7 @@ void cpu_flash_attention( /* qk_sum */ qSplitSize + /* dst */ qSplitSize * headSize; - int64_t size_bytes = size_per_thread * num_thread * query.element_size() * 4; + int64_t size_bytes = size_per_thread * num_thread * query.element_size(); std::vector buf_vec(size_bytes); void* buf = reinterpret_cast(buf_vec.data()); // Need to double check the following