search

ARM-software / ComputeLibrary

The Compute Library is a set of computer vision and machine learning functions optimised for both Arm CPUs and GPUs using SIMD technologies.
MIT License
2.75k stars 767 forks source link

Segfault when running DWC tests on WoA #1099

Closed morgolock closed 3 months ago

morgolock commented 3 months ago

Output of 'strings libarm_compute.so | grep arm_compute_version':

Platform: Snapdragon (TM) 8cx Gen 2 @ 3.15 GHz 3.15 GHz

Operating System: Windows 11

Problem description:

The following test segfaults on WoA

build\tests\arm_compute_validation.exe --filter-id=24749
Version = arm_compute_version=v0.0-unreleased Build options: {'opencl': '0', 'neon': '1', 'os': 'windows', 'examples': '0', 'validation_tests': '1', 'benchmark_examples': '0', 'build': 'native', 'arch': 'armv8a', 'Werror': '0', 'debug': '1', 'asserts': '1', 'standalone': '1'} Git hash=b'e7d918afeb419ce2464c13c7c067e62a29f1b4c8'
CommandLine = build\tests\arm_compute_validation.exe --filter-id=24749
Seed = 392366724
cpu_has_sve = false
cpu_has_sve2 = false
cpu_has_svef32mm = false
cpu_has_svei8mm = false
cpu_has_svebf16 = false
cpu_has_sme = false
cpu_has_sme2 = false
cpu_has_fp16 = false
cpu_has_bf16 = false
cpu_has_dotprod = false
cpu_has_i8mm = false
CPU0 = GENERIC
Iterations = 1
Threads = 1
Dataset mode = PRECOMMIT
Running [24749] 'NEON/DepthwiseConvolutionLayer/Quantized/QASYMM8/Generic/Dilation/RunSmall@In=33,33,3:Weights=3x3:Info=1,1;2,2,2,2:Dilation=2x2:DepthMultiplier=2:DataType=QASYMM8:IgnoredQuantizationInfo=Scale:0~Offset:0:IgnoredQuantizationInfo=Scale:0~Offset:0:DataLayout=NHWC:ActivationInfo='
morgolock commented 3 months ago

Fixed the the patch https://review.mlplatform.org/c/ml/ComputeLibrary/+/11182