vtune not capturing B580 GPU vulkan usage.

[Stolk]

I am using the latest vtune 2025-7 on Linux.

My machine has an Intel B580 GPU in it.

If I analyze my application doing Vulkan Compute, I get no GPU information back from vtune analysis GPU Compute / Media Hotspots (preview)
I also tried all other analysis modes, none of them see GPU activity. Why not?

It is as if nothing was run on the GPU.

However, using Vulkan, three different kernels were executed on the B580 by my application.

To investigate this, I decided to run my application in OpenCL mode (I can tell my app to use CUDA, OpenCL or Vulkan.)

When doing OpenCL compute kernels, vtune does manage to return information on the GPU usage, including the execution times for the kernels. (NOTE: those kernels run appr 10x slower than when executing the app without vtune. Is that expected?)

So my questions: Why does vtune 2025-7 not support Vulkan?

Also: to get performance data out of it, I decided to try Intel GPA, again. However, I could not find a Linux download for GPA. Whereas I definitely have used GPA in the past, on Linux. There is only a Windows binary available at the moment?

Thanks!

[Stolk]

When capturing OpenCL app:

Data collection is completed successfully

Dec 17 2025 15:28:47 The result file '/home/bram/intel/vtune/projects/gib/r000gh/r000gh.vtune' is successfully created and added to the project .

When capturing Vulkan app:

Data collection is completed with warnings

Dec 17 2025 18:36:24 The result file '/home/bram/intel/vtune/projects/gib/r001gh/r001gh.vtune' is created and added to the project . Please see warning messages for details.

[Instrumentation Engine]: GTPin: GTPin didn't find any kernels... Exiting without doing anything.

[Stolk]

From the self check log:

Getting available devices:
Command line:
sycl-ls
Stdout:
  [opencl:cpu][opencl:0] Intel(R) OpenCL, Intel(R) Core(TM) i7-14700 OpenCL 3.0 (Build 0) [2025.20.10.0.10_160000]
  [opencl:gpu][opencl:1] Intel(R) OpenCL Graphics, Intel(R) Arc(TM) B580 Graphics OpenCL 3.0 NEO  [25.31.034666]

Checking DPC++ application as prerequisite for GPU analyses...
Setting envirnoment variable:
ONEAPI_DEVICE_SELECTOR=opencl:gpu
Command line:
/home/intel/oneapi/vtune/2025.7/bin64/self_check_apps/matrix.dpcpp/matrix.dpcpp
Stdout:
  Address of buf1 = 0x7872bd3ff010
  Offset of buf1 = 0x7872bd3ff180
  Address of buf2 = 0x7872bcbfe010
  Offset of buf2 = 0x7872bcbfe1c0
  Address of buf3 = 0x7872bc3fd010
  Offset of buf3 = 0x7872bc3fd100
  Address of buf4 = 0x7872bbbfc010
  Offset of buf4 = 0x7872bbbfc140
  Using multiply kernel: multiply1
  <LOADER>[INFO]: loaded adapter 0x0x3245c7a0 (libur_adapter_opencl.so.0) from /opt/intel/oneapi/2025.3/lib/libur_adapter_opencl.so.0
  SYCL_UR_TRACE: Device allowed by SYCL_DEVICE_ALLOWLIST
  SYCL_UR_TRACE:   platform: Intel(R) OpenCL
  SYCL_UR_TRACE:   device: Intel(R) Core(TM) i7-14700
  SYCL_UR_TRACE: Device allowed by SYCL_DEVICE_ALLOWLIST
  SYCL_UR_TRACE:   platform: Intel(R) OpenCL Graphics
  SYCL_UR_TRACE:   device: Intel(R) Arc(TM) B580 Graphics
  SYCL_UR_TRACE: Device allowed by SYCL_DEVICE_ALLOWLIST
  SYCL_UR_TRACE:   platform: Intel(R) OpenCL Graphics
  SYCL_UR_TRACE:   device: Intel(R) Arc(TM) B580 Graphics
  
  Available devices:
  Intel(R) Arc(TM) B580 Graphics
  
  *****
  * Running on Intel(R) Arc(TM) B580 Graphics
  *****
  No subdevices available, using the whole device
  
  SYCL_UR_TRACE: Device allowed by SYCL_DEVICE_ALLOWLIST
  SYCL_UR_TRACE:   platform: Intel(R) OpenCL Graphics
  SYCL_UR_TRACE:   device: Intel(R) Arc(TM) B580 Graphics
  ************
  * Finished *
  ************
  Elapsed Time: 4.94363s

Setting envirnoment variable:
ONEAPI_DEVICE_SELECTOR=level_zero:gpu
Command line:
/home/intel/oneapi/vtune/2025.7/bin64/self_check_apps/matrix.dpcpp/matrix.dpcpp
Stdout:
  Address of buf1 = 0x7571891ff010
  Offset of buf1 = 0x7571891ff180
  Address of buf2 = 0x7571889fe010
  Offset of buf2 = 0x7571889fe1c0
  Address of buf3 = 0x7571881fd010
  Offset of buf3 = 0x7571881fd100
  Address of buf4 = 0x7571879fc010
  Offset of buf4 = 0x7571879fc140
  Using multiply kernel: multiply1
  <LOADER>[INFO]: loaded adapter 0x0x2cebdb40 (libur_adapter_level_zero.so.0) from /opt/intel/oneapi/2025.3/lib/libur_adapter_level_zero.so.0
  <LOADER>[INFO]: loaded adapter 0x0x2cec1990 (libur_adapter_level_zero_v2.so.0) from /opt/intel/oneapi/2025.3/lib/libur_adapter_level_zero_v2.so.0
  
  Available devices:
  ERROR: no appropriate devices found

Setting envirnoment variable:
ZES_ENABLE_SYSMAN=1
Command line:
/home/intel/oneapi/vtune/2025.7/bin64/self_check_apps/matrix.dpcpp/matrix.dpcpp
Stdout:
  Address of buf1 = 0x74b0423ff010
  Offset of buf1 = 0x74b0423ff180
  Address of buf2 = 0x74b041bfe010
  Offset of buf2 = 0x74b041bfe1c0
  Address of buf3 = 0x74b0413fd010
  Offset of buf3 = 0x74b0413fd100
  Address of buf4 = 0x74b040bfc010
  Offset of buf4 = 0x74b040bfc140
  Using multiply kernel: multiply1
  <LOADER>[INFO]: loaded adapter 0x0x1cc6b50 (libur_adapter_level_zero.so.0) from /opt/intel/oneapi/2025.3/lib/libur_adapter_level_zero.so.0
  <LOADER>[INFO]: loaded adapter 0x0x1cca9a0 (libur_adapter_level_zero_v2.so.0) from /opt/intel/oneapi/2025.3/lib/libur_adapter_level_zero_v2.so.0

  Available devices:
  ERROR: no appropriate devices found

Checking DPC++ application as prerequisite for GPU analyses: Fail
Unable to run DPC++ application on GPU connected to this system. If you are using an Intel GPU and want to verify profiling support for DPC++ applications, check these requirements:
* Install Intel(R) GPU driver.
* Install Intel(R) Level Zero GPU runtime.
* Install Intel(R) oneAPI DPC++ Runtime and set the environment.