tree dac23e9d51fd03375e8ed756fdab96e8359228a9
parent 72b7471955f32d2fccb170ab1ab40c7eecfa7f44
author Gunes Bayir <gunes.bayir@arm.com> 1697014565 +0100
committer SiCong Li <sicong.li@arm.com> 1698667264 +0000

Use dynamic quantization in OpenCL™ Direct Convolution tests

This patch calculates the output quantization info based on the inputs'
quantization information. The previous approach was using the same
quantization information for input, weights and output.

This implementation does not cover the cases where we have fused
activation function.

Note that no Neon™ tests are changed since there were not any quantized
Neon Direct Convolution tests.

Resolves: COMPMID-6485
Change-Id: Id32241320acae0b58552546d6d6f665cd5c63044
Signed-off-by: Gunes Bayir <gunes.bayir@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10470
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>