hw-smi

hw-smi

A minimal, cross-compatible CPU/GPU telemetry monitor with accurate data directly from vendor APIs and beautiful ASCII visualization.

moritz@opencl-pc:~/hw-smi$ sudo bin/hw-smi --bars

moritz@opencl-pc:~/hw-smi$ sudo bin/hw-smi --graphs

Hardware/OS Support

Metric(Windows/Linux)	CPUsWin32//proc	Nvidia GPUs NVML	AMD GPUs   ADLX/AMDSMI	Intel GPUs  SYSMAN
device name	✅/✅	✅/✅	✅/✅	✅/✅
per-core/avg usage [%]	✅✅/✅✅	✅✅/✅✅	✅✅/✅✅	✅✅/❎✅
memory bandwidth [MB/s]	❌❌/❌❌	✅✅/✅✅	🟨🟨/✅✅	❎❎/❎❎
memory occupation [MB]	✅✅/✅✅	✅✅/✅✅	✅✅/✅✅	✅✅/✅✅
temperature [°C]	🫠✅/✅✅	✅✅/✅✅	✅✅/✅✅	❎✅/❎✅
power [W]	❌❌/❌❌	✅✅/✅✅	✅✅/✅✅	❎✅/✅✅
fan [RPM]	❌❌/❌❌	✅🟨/✅🟨	✅✅/✅🟨	✅🟨/🫠🟨
core clock [MHz]	🟨🟨/✅🟨	✅✅/✅✅	✅✅/✅✅	✅✅/✅✅
memory clock [MHz]	❌❌/❌❌	🫠🫠/🫠🫠	✅✅/✅✅	✅✅/🫠✅
PCIe bandwidth [MB/s]	🟨🟨/🟨🟨	✅✅/✅✅	❌❌/🫠✅	❎✅/❎❎

Legend	Description
AB/CD	A current value (Windows), B max value (Windows)C current value (   Linux   ), D max value (   Linux   )
✅	supported and working
❎	supported and working, but administrator/sudo permissions required
🟨	vendor API does not directly provide metric, but workaround/estimate/default possible
🫠	available but broken in vendor API
❌	unavailable in vendor API and no suitable default value possible

• https://github.com/GPUOpen-LibrariesAndSDKs/ADLX/issues/27
• https://github.com/ROCm/rocm-systems/issues/5036 / 5035 / 5034 / 14222
• https://github.com/intel/drivers.gpu.control-library/issues/120 / 146 / 138 / 149
• https://github.com/intel/compute-runtime/issues/932 / 926 / 925 / 924

GPU Driver Installation

• Windows

  • Download and install the AMD/Intel/Nvidia GPU Drivers, which contain the OpenCL Runtime.
  • Reboot.

• Linux

  • Download and install Nvidia GPU Drivers, which contain the OpenCL Runtime, with:

    sudo apt update && sudo apt upgrade -y
    sudo apt install -y g++ git ocl-icd-libopencl1 ocl-icd-opencl-dev nvidia-driver-580
    sudo shutdown -r now
    

  • Download and install AMD GPU Drivers, which contain the OpenCL Runtime, with:

    sudo apt update && sudo apt upgrade -y
    sudo apt install -y g++ git ocl-icd-libopencl1 ocl-icd-opencl-dev rocm-opencl-icd rocm-opencl-runtime
    mkdir -p ~/amdgpu
    wget -P ~/amdgpu https://repo.radeon.com/amdgpu-install/31.60/ubuntu/resolute/amdgpu-install_31.60.316000-1_all.deb
    sudo apt install -y ~/amdgpu/amdgpu-install*.deb
    sudo amdgpu-install -y --usecase=graphics,rocm,opencl --opencl=rocr
    sudo usermod -a -G render,video $(whoami)
    rm -r ~/amdgpu
    sudo shutdown -r now
    

  • Download and install the OpenCL Runtime and oneAPI Level-Zero-Loader, with:

    sudo apt update && sudo apt upgrade -y
    sudo apt install -y g++ git ocl-icd-libopencl1 ocl-icd-opencl-dev intel-opencl-icd libze1
    sudo usermod -a -G render $(whoami)
    sudo shutdown -r now
    

Compiling the Source Code

Windows

• Download and install Visual Studio Community. In Visual Studio Installer, add:
  • Desktop development with C++
  • MSVC v142
  • Windows 10 SDK
• Download and unzip hw-smi.
• Open hw-smi.vcxproj in Visual Studio Community.
• Compile by clicking the ► Local Windows Debugger button.
• Go to hw-smi/bin/ folder and double-click hw-smi.exe.
• Alternatively, run from CMD:

  hw-smi.exe
  hw-smi.exe --bars
  hw-smi.exe --graphs
  hw-smi.exe --log
  hw-smi.exe --help
  

• Note that it will also work without administrator permissions. However, some telemetry counters on Intel GPUs are not available without administrator permissions.

Linux

• Clone from GitHub:

  git clone https://github.com/ProjectPhysX/hw-smi.git && cd hw-smi
  

• Compile:

  chmod +x make.sh
  ./make.sh
  

• Run:

  sudo bin/hw-smi
  sudo bin/hw-smi --bars
  sudo bin/hw-smi --graphs
  sudo bin/hw-smi --log
  sudo bin/hw-smi --help
  

• Note that it will also work without sudo. However, some telemetry counters on Intel GPUs are not available without sudo.

Logging Telemetry Data

• Run hw-smi with the --log command-line option. This will write all telemetry data into a file hw-smi-YYYY-DD-MM-hh-mm-ss.log located next to the hw-smi executable.
• The telemetry data in the file is formatted as follows:

                        , Intel Core i7-8700K                                                                         , NVIDIA TITAN Xp                                         
  Date       Time       ,   0,   1,   2,   3,   4,   5,   6,   7,   8,   9,  10,  11, CPU, RAM    , Tmp, Clk , PCIe BW, GPU, BW   , VRAM   , Tmp, Pwr , Fan , CClk, MClk, PCIe  
  YYYY-MM-DD hh:mm:ss.ss,   %,   %,   %,   %,   %,   %,   %,   %,   %,   %,   %,   %,   %,      MB,  'C,  MHz,    MB/s,   %,  GB/s,      MB,  'C,    W,  RPM,  MHz,  MHz,   MB/s
  Hardware Specs        , 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100,   16250,   ?, 5000,    3936, 100,   548,   12288, 100,  250, 5000, 1911, 1426,   3936
  2026-08-19 17:01:46.29,   0,   5,   0,   0,  11,   0,   5,   0,   0,   0,   0,   0,   4,    7953,   ?, 3696,       0,   0,     0,     163,  35,   16, 1150,  139,  101,      0
  2026-08-19 17:01:46.69,   8,   2,   2,  14,  14,   2,   8,   2,  26,   2,   8,   2,   3,    7952,   ?, 3696,       0,   0,     0,     163,  35,   16, 1150,  139,  101,      0
  2026-08-19 17:01:47.05,  10,   0,  10,   4,  22,   0,  10,   0,  10,   0,  16,   4,   7,    7946,   ?, 3696,       0,   0,     0,     163,  35,   16, 1150,  139,  101,      0
  2026-08-19 17:01:47.41,  12,  12,   0,  17,   0,   0,  12,   0,   6,   0,   0,  12,   8,    7940,   ?, 3696,       0,   0,     0,     163,  35,   16, 1150,  139,  101,      0
  2026-08-19 17:01:47.79,   8,   8,   8,  14,   0,   0,   8,   0,   0,   2,   2,   0,   8,    7930,   ?, 3696,       0,   0,     0,     163,  35,   16, 1150,  139,  101,      0