GpuMagick is fundamentally a command line utility. The command line is the real interface between you and GpuMagick.

You can open a terminal in GpuMagick folder by clicking on the Open terminal link in the bottom of the main user interface:



How to get all command line parameters:

> GpuMagick_0.2.0>GpuMagick_cli --help

--help : print all options
--version : show version
--width <number> : set the width of the 3D window
--height <number> : set the height of the 3D window
--demolist : print the list of available demos
--demo <name> : run a demo from its name.
  Use --demolist for all available names
--demofile <filename> : load and run a GeeXLab demo from a filename
--fullscreen : run the demo or graphics test in fullscreen
--glinfo : print a quick OpenGL report
--glinfo-all : print a detailed OpenGL report
--vkinfo : print a quick Vulkan report
--vkinfo-all : print a detailed Vulkan report
--clinfo : print OpenCL information
--cuinfo : print CUDA information
--physx4info : print PhysX 4 information
--gpuinfo : print GPU info
--glnvarchinfo : print information about NVIDIA GPU architecture
--vkinfo-extension <name> : info about a Vulkan extension.
  Should be used with --gpu-index
--glvendor : print the GL_VENDOR value
--glrenderer : print the GL_RENDERER value
--glversion : print the GL_VERSION value
--gpu-index <0 ; numgpus-1> : set the GPU index (0-based) for Vulkan/OpenCL/CUDA demos or info
--print-render-speed : print speed rendering info: frames, framerate, time step
--hw-polling-interval <milliseconds> : set the GPU monitoring polling interval
  in milliseconds (default: 1000)
--gpu-log-data : log GPU data into a CSV file
--benchmark : benchmark the demo
--duration-ms <milliseconds> : set the demo duration in milliseconds. Default is 60000.
--p720 : benchmark preset:720 (1280x720)
--p1080 : benchmark preset:1080 (1920x1080)
--p1440 : benchmark preset:1440 (2560x1440)
--p2160 : benchmark preset:2160 (3840x2160)
--no-score-box : disable the score box at the end of the benchmark
--gpu-monitor-print : print GPU monitoring info
--disable-traces : disable traces (no writing in the log file)
--logfile-suffix <suffix> : add a suffix to the log file name
--tessmark-tesslevel <number> : set the tessellation level for TessMark
  (default: 32)
--gimark-instances <number> : set the number of instances for GiMark
  (default: 500'000)
--glmemory-print : print OpenGL memory stats (OpenGL demos only)
--thread-affinity-mask : set the thread affinity mask. Default is 0 (= affinity mask not used)
--thread-core : set the core (from 1 to n) on which the demo should run. Default is 0 (= core number not used)
--dpi-awareness : enable DPI awareness on Windows platform
--gpuz-print : print some GPU-Z sensors in the terminal
--gpuz-log-data : log GPU-Z data into a CSV file

How to quickly get basic information about your GPUs:

> GpuMagick_cli --gpuinfo
GPU 0: GeForce RTX 2070 [10DE-1F02] --------------
- codename: TU106-400
- bios: 90.06.18.40.ee
- config (cores/TMUs/ROPs): 2304 / 144 / 64
- raytracing cores: 36
- tensor cores: 288
- TDP: 175 W
- memory: 8192MB
- bus width: 256-bit
- memory vendor: Samsung
- core base clock: 1411 MHz
- core boost clock: 1621 MHz
- memory boost clock: 7002 MHz
- power default: 100.0 % TDP
- power limit: 100.0 % TDP
- power max: 105.7 % TDP
- power min: 71.4 % TDP
- driver: 452.06
GPU 1: GeForce GT 1030 [10DE-1D01] --------------
- codename: GP108
- bios: 86.08.0c.00.1a
- config (cores/TMUs/ROPs): 384 / 24 / 16
- TDP: 35 W
- memory: 2048MB
- bus width: 64-bit
- memory vendor: MICRON
- core base clock: 1228 MHz
- core boost clock: 1469 MHz
- memory boost clock: 3005 MHz
- power default: 100.0 % TDP
- power limit: 100.0 % TDP
- power max: 100.0 % TDP
- power min: 85.0 % TDP
- driver: 452.06

How to quickly get the OpenGL basic information:

> GpuMagick_cli --glinfo
[ OpenGL info ]
GL_VENDOR: NVIDIA Corporation
GL_RENDERER: GeForce RTX 2070/PCIe/SSE2
GL_VERSION: 4.6.0 NVIDIA 452.06
Total memory size: 8192MB
Available memory size: 6986MB
(NVIDIA) WARP size: 32 - WARP/SM: 32 - # of SMs: 36

How to quickly get the Vulkan basic information:

> GpuMagick_cli --vkinfo
[ Vulkan info ]
- # of devices: 2
- [device 0]
 - name: GeForce RTX 2070
 - deviceID: 10DE-1F02 (4318-7938)
 - API version: 1.2.133
 - NVIDIA driver: 452.6.0.0
- [device 1]
 - name: GeForce GT 1030
 - deviceID: 10DE-1D01 (4318-7425)
 - API version: 1.2.133
 - NVIDIA driver: 452.6.0.0

How to quickly get the CUDA basic information:

> GpuMagick_cli --cuinfo
[ CUDA info ]
# of CUDA devices: 2
CUDA device 0
- name: GeForce RTX 2070
- compute capability: 7.5
- cuda cores: 2304
- cuda cores per SM: 64
- multiprocessors (SMs): 36
- core clock: 1620 MHz
- memory clock: 7001 MHz
- memory bus width: 256-bit
- memory bandwidth: 448 GB/s
- pciBusId: 0000:01:00.0
- total Memory: 4095 MB
- maxThreadsDim: [1024 ; 1024 ; 64]
- SIMDWidth: 32
- memPitch: 2147483647
- regsPerBlock: 65536
- textureAlign: 512
- pci bus id: 1
- L2 cache size: 4096 KB
- total constant memory: 65536 bytes
- shared_mem_per_block: 49152
- max_shared_mem_per_sm: 65536
- max_threads_per_block: 1024
- max_threads_per_sm: 1024
- warp_size: 32
- max_grid_size: [2147483647 ; 65535 ; 65535]
- max_block_dim: [1024 ; 1024 ; 64]
CUDA device 1
- name: GeForce GT 1030
- compute capability: 6.1
- cuda cores: 384
- cuda cores per SM: 128
- multiprocessors (SMs): 3
- core clock: 1468 MHz
- memory clock: 3004 MHz
- memory bus width: 64-bit
- memory bandwidth: 48 GB/s
- pciBusId: 0000:02:00.0
- total Memory: 2048 MB
- maxThreadsDim: [1024 ; 1024 ; 64]
- SIMDWidth: 32
- memPitch: 2147483647
- regsPerBlock: 65536
- textureAlign: 512
- pci bus id: 2
- L2 cache size: 512 KB
- total constant memory: 65536 bytes
- shared_mem_per_block: 49152
- max_shared_mem_per_sm: 98304
- max_threads_per_block: 1024
- max_threads_per_sm: 2048
- warp_size: 32
- max_grid_size: [2147483647 ; 65535 ; 65535]
- max_block_dim: [1024 ; 1024 ; 64]

How to quickly get the OpenCL basic information:

> GpuMagick_cli --clinfo
[ OpenCL info ]
# of OpenCL platforms: 1
OpenCL platform 0
- vendor: NVIDIA Corporation
- name: NVIDIA CUDA
- version: OpenCL 1.2 CUDA 11.0.228
- number of devices: 2
- device 0 ------
  - vendor: NVIDIA Corporation
  - name: GeForce RTX 2070
  - version: OpenCL 1.2 CUDA
  - driver version: 452.06
  - type: GPU
  - compute units: 36
  - clock frequency: 1620 MHz
  - flops: 58320
- device 1 ------
  - vendor: NVIDIA Corporation
  - name: GeForce GT 1030
  - version: OpenCL 1.2 CUDA
  - driver version: 452.06
  - type: GPU
  - compute units: 3
  - clock frequency: 1468 MHz
  - flops: 4404

How to list all demos:

> GpuMagick_cli --demolist
Demos list:
"furmark-gl" : GPU burner - OpenGL 3.2
"furmark-vk" : GPU burner - Vulkan
"tessmark-gl" : Tessellation - OpenGL 4.0
"tessmark-vk" : Tessellation - Vulkan
"gimark-gl" : Geometry Instancing - OpenGL 3.3
"julia32-gl" : Julia fractal in FP32 - OpenGL 2.1/3.2
"julia64-gl" : Julia fractal in FP64 - OpenGL 4.0
"gsmark-gl" : Geometry Shader - OpenGL 3.2
"plasmark-gl" : PlasMark - OpenGL 4.3
"rhodiumlc-gl" : Pixel shader test - OpenGL 2.1
"grassmark-gl" : Grass Rendering with Geometry Instancing- OpenGL 3.2+
"raymarching-gl" : Pixel shader test - OpenGL 3.2
"triangle-gl-gxl" : Simple RGB triangle in OpenGL 2.1
"triangle-vk-gxl" : Simple RGB triangle in Vulkan
"webcam" : Webcam (utility demo - Windows only) - OpenGL 3.2

How to launch FurMark (OpenGL) during 20 seconds, write the result in the CSV file and quit:

> GpuMagick_cli --demo furmark-gl --benchmark --duration-ms 20000 --gpu-log-data --no-score-box

How to launch TessMark (Vulkan) with the P1080 preset and quit:

> GpuMagick_cli --demo tessmark-vk --p1080 --no-score-box