Intel^® VTune^™ Profiler 2022.0.0

Recommendations:

GPU Time, % of Elapsed time: 20.0%

GPU utilization is low. Switch to the for in-depth analysis of host activity. Poor GPU utilization can prevent the application from offloading effectively.

EU Array Stalled/Idle: 39.8% of Elapsed time with GPU busy

GPU metrics detect some kernel issues. Use GPU Compute/Media Hotspots (preview) to understand how well your application runs on the specified hardware.

Elapsed Time: 17.819s

GPU Time, % of Elapsed time: 20.0%

GPU utilization is low. Consider offloading more work to the GPU to increase overall application performance.

Top Hotspots when GPU was idle:

Function	Module	CPU Time
[Outside any known module]	[Unknown]	5.801s
[Skipped stack frame(s)]	[Unknown]	1.011s
std::__malloc_alloc::allocate	libstlport-dynamic.so	0.558s
memcmp	libc-dynamic.so	0.444s
Intel::OpenCL::Utils::OclNonReentrantSpinMutex::Lock	libcpu_device.so.2021.13.11.0	0.334s
[Others]	N/A	2.901s

Hottest Host Tasks:

Host Task	Task Time	% of Elapsed Time(%)	Task Count
tbb_parallel_for	4.004s	22.5%	41
clCreateContext	3.617s	20.3%	3
clWaitForEvents	3.533s	19.8%	14
clEnqueueNDRangeKernel	0.327s	1.8%	6
tbb_custom	0.318s	1.8%	5
[Others]	0.114s	0.6%	27

Hottest GPU Computing Tasks:

Computing Task	Total Time	Execution	% of Total Time(%)	SIMD Width	Peak Occupancy(%)	EU Threads Occupancy(%)	SIMD Utilization(%)
dppyPy_dppy_py_devfn__5F__5F_main_5F__5F__2E_pairwise_5F_python_24_1_2E_USM_3A_ndarray_28_float64_2C__20_2d_2C__20_C_29__2E_USM_3A_ndarray_28_float64_2C__20_2d_2C__20_C_29__2E_USM_3A_ndarray_28_float64_2C__20_2d_2C__20_C_29_	3.526s	3.522s	99.9%	8	100.0%	94.3%	100.0%

Collection and Platform Info:
- Application Command Line: python "lab/pair_wise_kernel.py" "--steps" "1" "--size" "16384" "--repeat" "5" "--json" "result_gpu.json" "--usm" "-d" "3"
- Operating System: 5.4.0-80-generic DISTRIB_ID=Ubuntu DISTRIB_RELEASE=20.04 DISTRIB_CODENAME=focal DISTRIB_DESCRIPTION="Ubuntu 20.04.4 LTS"
- Computer Name: s001-n179
- Result Size: 249.9 MB
- Collection start time: 18:57:43 10/05/2022 UTC
- Collection stop time: 18:58:01 10/05/2022 UTC
- Collector Type: Event-based sampling driver,Driverless Perf system-wide sampling,User-mode sampling and tracing
- CPU:
  - Name: Intel(R) microarchitecture code named Coffeelake
  - Frequency: 3.696 GHz
  - Logical CPU Count: 12
- GPU:
  - Name: HD Graphics P630
  - Vendor: Intel Corporation
  - EU Count: 24
  - Max EU Thread Count: 7
  - Max Core Frequency: 1.200 GHz
  - GPU OpenCL Info:
    Version: OpenCL 3.0 NEO
    Max Compute Units: 24
    Max Work Group Size: 256
    Local Memory: 65.5 KB
    SVM Capabilities: