1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
|
vreinterpret_s32_s8() casts int8x8_t to int32x2_t. However, third argument
of vdotq_lane_s32() is of type int8x8_t.
--- a/third_party/tflite/src/tensorflow/lite/kernels/internal/optimized/depthwiseconv_3x3_filter_common.h
+++ b/third_party/tflite/src/tensorflow/lite/kernels/internal/optimized/depthwiseconv_3x3_filter_common.h
@@ -129,16 +129,14 @@ inline int32x4_t vdotq_four_lane_s32(int32x4_t acc, int8x16_t lhs,
int8x16_t rhs, const int lane) {
switch (lane) {
case 0:
- return vdotq_lane_s32(acc, lhs, vreinterpret_s32_s8(vget_low_s8(rhs)), 0);
+ return vdotq_lane_s32(acc, lhs, vget_low_s8(rhs), 0);
case 1:
- return vdotq_lane_s32(acc, lhs, vreinterpret_s32_s8(vget_low_s8(rhs)), 1);
+ return vdotq_lane_s32(acc, lhs, vget_low_s8(rhs), 1);
case 2:
- return vdotq_lane_s32(acc, lhs, vreinterpret_s32_s8(vget_high_s8(rhs)),
- 0);
+ return vdotq_lane_s32(acc, lhs, vget_high_s8(rhs), 0);
case 3:
default:
- return vdotq_lane_s32(acc, lhs, vreinterpret_s32_s8(vget_high_s8(rhs)),
- 1);
+ return vdotq_lane_s32(acc, lhs, vget_high_s8(rhs), 1);
}
}
|
