Add LUT-based leaky relu for QASYMM8 on CPU

* Add LUT generation function for Leaky ReLU.
* Some additional changes in the existing LUT implementation:
  + Bring back the NEON implementation of hard swish for 32-bit
    build. Library size of 64-bit build is not affected.
  + Add some extra #ifdef to remove unnecessary code in 32-bit
    build.

Resolves: COMPMID-5386
Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Change-Id: I1ea49611cc922765ee741e31138c888401d33e9b
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7845
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 11 insertions, 0 deletions

diff --git a/arm_compute/core/QuantizationInfo.h b/arm_compute/core/QuantizationInfo.h
index 0bd0f21bc1..21d962d08b 100644
--- a/arm_compute/core/QuantizationInfo.h
+++ b/arm_compute/core/QuantizationInfo.h
@@ -409,6 +409,17 @@ inline qasymm8_t qasymm8_hard_swish(qasymm8_t                      in,
     return tmp;
 }
 
+inline qasymm8_t qasymm8_leaky_relu(qasymm8_t                      in,
+                                    const UniformQuantizationInfo &qi_in,
+                                    const UniformQuantizationInfo &qi_out,
+                                    float                          alpha)
+{
+    float tmp_f         = dequantize_qasymm8(in, qi_in);
+    tmp_f               = tmp_f > 0 ? tmp_f : tmp_f * alpha;
+    const qasymm8_t tmp = quantize_qasymm8(tmp_f, qi_out);
+    return tmp;
+}
+
 /** Dequantize a value given a 8-bit symmetric quantization scheme
  *
  * @param[in] value Value to dequantize