#0: multi core fold Os

tenstorrent · Mar 1, 2025 · 4c8f233 · 4c8f233
1 parent baa821f
commit 4c8f233
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 2 deletions.
diff --git a/ttnn/cpp/ttnn/operations/data_movement/fold/device/fold_multi_core_program_factory.cpp b/ttnn/cpp/ttnn/operations/data_movement/fold/device/fold_multi_core_program_factory.cpp
@@ -56,11 +56,12 @@ Fold::MultiCore::cached_program_t fold_multi_core(
     auto cb_dst0 = CreateCircularBuffer(program, all_cores, dst_cb_config);
 
     // Setup kernel
+    // Set build optimization level to Os. O2 was slower.
     tt::tt_metal::KernelHandle writer_kernel_id = tt::tt_metal::CreateKernel(
         program,
         "ttnn/cpp/ttnn/operations/data_movement/fold/device/kernels/dataflow/writer_cb2s_row_major.cpp",
         all_cores,
-        WriterDataMovementConfig({cb_src0_index, cb_dst0_index}));
+        WriterDataMovementConfig({cb_src0_index, cb_dst0_index}, {}, tt::tt_metal::KernelBuildOptLevel::Os));
 
     // Writer run-time args
     SetRuntimeArgs(

diff --git a/ttnn/cpp/ttnn/operations/data_movement/fold/device/fold_op_multi_core.cpp b/ttnn/cpp/ttnn/operations/data_movement/fold/device/fold_op_multi_core.cpp
@@ -56,7 +56,7 @@ cached_program_t fold_multi_core(const Tensor& input, const Tensor& output, uint
         program,
         "ttnn/cpp/ttnn/operations/data_movement/fold/device/kernels/dataflow/writer_cb2s_row_major.cpp",
         all_cores,
-        WriterDataMovementConfig({cb_src0_index, cb_dst0_index}));
+        WriterDataMovementConfig({cb_src0_index, cb_dst0_index}, {}, tt::tt_metal::KernelBuildOptLevel::Os));
 
     // Writer run-time args
     SetRuntimeArgs(