[Bug]: benchmark doesn't work with GPU of Apollo Lake

[js333031]

OpenVINO Version

2025.1.0.18503

Operating System

Other (Please specify in description)

Device used for inference

GPU

Framework

None

Model used

person-detection-retail-0013, ResNet50

Issue description

Hardware: E3950 (Apollo Lake)
OS: Ubuntu 22.04.1 LTS
Kernel: 6.8.0-57-generic
Used APT method to install OpenVINO on freshly installed OS. More information in attached files, including system information, packages, sample runs...

When executing the benchmark app on integrated GPU of E3950 (Apollo Lake), there are errors.

(openvino_env) apl@apl-Broxton-P:~/openvino_cpp_samples_build/intel64/Release$ ./benchmark_app -m /home/apl/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.xm
l -d GPU
[Step 1/11] Parsing and validating input arguments
[ INFO ] Parsing input parameters
[Step 2/11] Loading OpenVINO Runtime
[ INFO ] OpenVINO:
[ INFO ] Build ................................. 2025.1.0-18503-6fec06580ab-releases/2025/1
[ INFO ]
[ INFO ] Device info:
[ INFO ] GPU
[ INFO ] Build ................................. 2025.1.0-18503-6fec06580ab-releases/2025/1
[ INFO ]
[ INFO ]
[Step 3/11] Setting device configuration
[ WARNING ] Performance hint was not explicitly specified in command line. Device(GPU) performance hint will be set to THROUGHPUT.
[Step 4/11] Reading model files
[ INFO ] Loading model files
[ INFO ] Read model took 89.95 ms
[ INFO ] Original model I/O parameters:
[ INFO ] Network inputs:
[ INFO ]     data (node: data) : f32 / [N,C,H,W] / [1,3,320,544]
[ INFO ] Network outputs:
[ INFO ]     detection_out (node: detection_out) : f32 / [...] / [1,1,200,7]
[Step 5/11] Resizing model to match image sizes and given batch
[Step 6/11] Configuring input of the model
[ INFO ] Model batch size: 1
[ INFO ] Network inputs:
[ INFO ]     data (node: data) : u8 / [N,C,H,W] / [1,3,320,544]
[ INFO ] Network outputs:
[ INFO ]     detection_out (node: detection_out) : f32 / [...] / [1,1,200,7]
[Step 7/11] Loading the model to the device
[ ERROR ] Exception from src/inference/src/cpp/core.cpp:109:
Exception from src/inference/src/dev/plugin.cpp:53:
Check 'false' failed at src/plugins/intel_gpu/src/plugin/program_builder.cpp:165:
[GPU] ProgramBuilder build failed!
Exception from src/plugins/intel_gpu/src/runtime/ocl/ocl_memory.cpp:112:
[GPU] clWaitForEvents, error code: -14




free(): invalid next size (fast)
Aborted (core dumped)
(openvino_env) apl@apl-Broxton-P:~/openvino_cpp_samples_build/intel64/Release$

sysinfo.txt
GPU Error.txt
hello_query.txt
hello_query_gdb.txt

Step-by-step reproduction

No response

Relevant log output

Issue submission checklist

• I'm reporting an issue. It's not a question.
• I checked the problem with the documentation, FAQ, open issues, Stack Overflow, etc., and have not found a solution.
• There is reproducer code and related data files such as images, videos, models, etc.

[js333031]

Attaching clinfo.

compute-runtime were installed from https://github.com/intel/compute-runtime/releases/tag/24.35.30872.22

clinfo.txt

[js333031]

intel/compute-runtime#679 & intel/compute-runtime#697 could have hints or related...

dmesg shows following:

[31921.633607] perf: interrupt took too long (3154 > 3135), lowering kernel.perf_event_max_sample_rate to 63000
[32129.609760] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[32129.609893] i915 0000:00:02.0: [drm] benchmark_app[12713] context reset due to GPU hang
[32129.621373] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:e757fefe, in benchmark_app [12713]
[32265.609821] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[32265.609928] i915 0000:00:02.0: [drm] benchmark_app[12752] context reset due to GPU hang
[32265.621486] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:e757fefe, in benchmark_app [12752]
[32315.593830] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[32315.593938] i915 0000:00:02.0: [drm] benchmark_app[12765] context reset due to GPU hang
[32315.605393] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:e757fefe, in benchmark_app [12765]
[32333.577793] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[32333.577900] i915 0000:00:02.0: [drm] benchmark_app[12777] context reset due to GPU hang
[32333.589394] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:e757fefe, in benchmark_app [12777]
[32771.593786] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[32771.593911] i915 0000:00:02.0: [drm] benchmark_app[13168] context reset due to GPU hang
[32771.605365] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:e757fefe, in benchmark_app [13168]
[33359.143839] perf: interrupt took too long (3964 > 3942), lowering kernel.perf_event_max_sample_rate to 50000
[34200.586928] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[34200.587034] i915 0000:00:02.0: [drm] benchmark_app[15942] context reset due to GPU hang
[34200.598481] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:8ed1fff2, in benchmark_app [15942]
[34616.585875] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[34616.586011] i915 0000:00:02.0: [drm] benchmark_app[15964] context reset due to GPU hang
[34616.597574] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:e75f7efe, in benchmark_app [15964]
(openvino_env) apl@apl-Broxton-P:~/openvino_cpp_samples_build/intel64/Release$

[js333031]

Changed kernel to 5.15 but benchmark_app fails to use the iGPU. Attaching logs.

test2.txt

[sshlyapn]

Hi @js333031, from the recent clinfo logs (for 5.15 kernel + 24.35.30872.22 OCL Runtime version) I can't see any issues in the configuration
Can you please try to run benchmark_app using CPU device?
It also makes sense to try some other OpenCL workload, say something like clpeak util. Could you please try to run it?

[js333031]

Using pip install openvino==2024.0.0 instead of 2025.1.0, benchmark app works with GPU.