This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new 64ca1cd Include required CMSIS headers in Cortex-M micro kernel.
(#6988)
64ca1cd is described below
commit 64ca1cd88be091359b93948af8f51616cbbbe541
Author: Andrew Reusch <[email protected]>
AuthorDate: Sun Nov 29 07:06:25 2020 -0800
Include required CMSIS headers in Cortex-M micro kernel. (#6988)
* The existing kernels referenced CMSIS functions presuming that
those functions were defined by user code. This was the case with
the old blog post build flow. Add #include, since it's impossible
to compile the kernels without it.
* TODO: port those functions to the micro kernels and remove external
dependency
---
python/tvm/topi/arm_cpu/cortex_m7/micro_kernel/gemm.py | 3 +++
1 file changed, 3 insertions(+)
diff --git a/python/tvm/topi/arm_cpu/cortex_m7/micro_kernel/gemm.py
b/python/tvm/topi/arm_cpu/cortex_m7/micro_kernel/gemm.py
index 943aee0..fb6f7a5 100644
--- a/python/tvm/topi/arm_cpu/cortex_m7/micro_kernel/gemm.py
+++ b/python/tvm/topi/arm_cpu/cortex_m7/micro_kernel/gemm.py
@@ -129,6 +129,9 @@ def gemm_MxKxN_impl(M, K, N, uniq_id):
#ifdef __cplusplus
extern "C"
#endif
+#include <arm_math.h>
+#include <arm_nnsupportfunctions.h>
+
__STATIC_FORCEINLINE int32_t gemm_{M}x{K}x{N}_body_{uniq_id}(
int8_t *aa, int8_t *bb, int32_t *cc,
int A_stride, int B_stride, int C_stride) {{
