Wiki

User Tools

Site Tools

Trace: • two-pass
ubuntu:gpu:amd_gpu:install_tools

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision		Previous revision
ubuntu:gpu:amd_gpu:install_tools [2023/06/04 18:43] – created peterubuntu:gpu:amd_gpu:install_tools [2023/06/04 20:07] (current) – peter
Line 1: Line 1:
 ====== Ubuntu - GPU - AMD GPU - Install Tools ====== ====== Ubuntu - GPU - AMD GPU - Install Tools ======
  
 +  * clinfo
 +  * clpeak
 +  * glxinfo
   * radeontop   * radeontop
 +
 +----
 +
 +===== clinfo =====
 +
 +Check supported OpenCL extensions.
 +
 +<code bash>
 +sudo apt install clinfo
 +</code>
 +
 +
 +<code bash>
 +clinfo
 +</code>
 +
 +returns:
 +
 +<code bash>
 +clinfo
 +Number of platforms: 1
 +  Platform Profile: FULL_PROFILE
 +  Platform Version: OpenCL 2.1 AMD-APP (3513.0)
 +  Platform Name: AMD Accelerated Parallel Processing
 +  Platform Vendor: Advanced Micro Devices, Inc.
 +  Platform Extensions: cl_khr_icd cl_amd_event_callback 
 +
 +
 +  Platform Name: AMD Accelerated Parallel Processing
 +Number of devices: 1
 +  Device Type: CL_DEVICE_TYPE_GPU
 +  Vendor ID: 1002h
 +  Board name: Radeon RX 7900 XTX
 +  Device Topology: PCI[ B#12, D#0, F#0 ]
 +  Max compute units: 48
 +  Max work items dimensions: 3
 +    Max work items[0]: 1024
 +    Max work items[1]: 1024
 +    Max work items[2]: 1024
 +  Max work group size: 256
 +  Preferred vector width char: 4
 +  Preferred vector width short: 2
 +  Preferred vector width int: 1
 +  Preferred vector width long: 1
 +  Preferred vector width float: 1
 +  Preferred vector width double: 1
 +  Native vector width char: 4
 +  Native vector width short: 2
 +  Native vector width int: 1
 +  Native vector width long: 1
 +  Native vector width float: 1
 +  Native vector width double: 1
 +  Max clock frequency: 3220Mhz
 +  Address bits: 64
 +  Max memory allocation: 21890072576
 +  Image support: Yes
 +  Max number of images read arguments: 128
 +  Max number of images write arguments: 8
 +  Max image 2D width: 16384
 +  Max image 2D height: 16384
 +  Max image 3D width: 16384
 +  Max image 3D height: 16384
 +  Max image 3D depth: 8192
 +  Max samplers within kernel: 29772
 +  Max size of kernel argument: 1024
 +  Alignment (bits) of base address: 1024
 +  Minimum alignment (bytes) for any datatype: 128
 +  Single precision floating point capability
 +    Denorms: Yes
 +    Quiet NaNs: Yes
 +    Round to nearest even: Yes
 +    Round to zero: Yes
 +    Round to +ve and infinity: Yes
 +    IEEE754-2008 fused multiply-add: Yes
 +  Cache type: Read/Write
 +  Cache line size: 64
 +  Cache size: 32768
 +  Global memory size: 25753026560
 +  Constant buffer size: 21890072576
 +  Max number of constant args: 8
 +  Local memory type: Scratchpad
 +  Local memory size: 65536
 +  Max pipe arguments: 16
 +  Max pipe active reservations: 16
 +  Max pipe packet size: 415236096
 +  Max global variable size: 21890072576
 +  Max global variable preferred total size: 25753026560
 +  Max read/write image args: 64
 +  Max on device events: 1024
 +  Queue on device max size: 8388608
 +  Max on device queues: 1
 +  Queue on device preferred size: 262144
 +  SVM capabilities:  
 +    Coarse grain buffer: Yes
 +    Fine grain buffer: Yes
 +    Fine grain system: No
 +    Atomics: No
 +  Preferred platform atomic alignment: 0
 +  Preferred global atomic alignment: 0
 +  Preferred local atomic alignment: 0
 +  Kernel Preferred work group size multiple: 32
 +  Error correction support: 0
 +  Unified memory for Host and Device: 0
 +  Profiling timer resolution: 1
 +  Device endianess: Little
 +  Available: Yes
 +  Compiler available: Yes
 +  Execution capabilities:  
 +    Execute OpenCL kernels: Yes
 +    Execute native function: No
 +  Queue on Host properties:  
 +    Out-of-Order: No
 +    Profiling : Yes
 +  Queue on Device properties:  
 +    Out-of-Order: Yes
 +    Profiling : Yes
 +  Platform ID: 0x7f1ae3ff0eb0
 +  Name: gfx1100
 +  Vendor: Advanced Micro Devices, Inc.
 +  Device OpenCL C version: OpenCL C 2.0 
 +  Driver version: 3513.0 (HSA1.1,LC)
 +  Profile: FULL_PROFILE
 +  Version: OpenCL 2.0 
 +  Extensions: cl_khr_fp64 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_fp16 cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_media_ops cl_amd_media_ops2 cl_khr_image2d_from_buffer cl_khr_subgroups cl_khr_depth_images cl_amd_copy_buffer_p2p cl_amd_assembly_program 
 +</code>
 +
 +----
 +
 +===== clpeak =====
 +
 +Measure speed.
 +
 +<code bash>
 +sudo apt install clpeak
 +</code>
 +
 +
 +<code bash>
 +clpeak
 +</code>
 +
 +returns:
 +
 +<code bash>
 +Platform: AMD Accelerated Parallel Processing
 +  Device: gfx1100
 +    Driver version  : 3513.0 (HSA1.1,LC) (Linux x64)
 +    Compute units   : 48
 +    Clock frequency : 3220 MHz
 +
 +    Global memory bandwidth (GBPS)
 +      float   : 742.69
 +      float2  : 790.45
 +      float4  : 825.86
 +      float8  : 857.23
 +      float16 : 878.19
 +
 +    Single-precision compute (GFLOPS)
 +      float   : 34058.41
 +      float2  : 34472.16
 +      float4  : 34262.00
 +      float8  : 34510.05
 +      float16 : 32940.55
 +
 +    Half-precision compute (GFLOPS)
 +      half   : 34022.91
 +      half2  : 65896.89
 +      half4  : 66729.46
 +      half8  : 62740.96
 +      half16 : 64157.85
 +
 +    Double-precision compute (GFLOPS)
 +      double   : 1190.26
 +      double2  : 1188.94
 +      double4  : 1186.53
 +      double8  : 1180.77
 +      double16 : 1148.82
 +
 +    Integer compute (GIOPS)
 +      int   : 8555.43
 +      int2  : 8381.68
 +      int4  : 8347.36
 +      int8  : 8427.74
 +      int16 : 8431.46
 +
 +    Transfer bandwidth (GBPS)
 +      enqueueWriteBuffer         : 18.65
 +      enqueueReadBuffer          : 17.34
 +      enqueueMapBuffer(for read) : 233422.14
 +        memcpy from mapped ptr   : 19.16
 +      enqueueUnmap(after write)  : 360921.62
 +        memcpy to mapped ptr     : 18.79
 +
 +    Kernel launch latency : 13.96 us
 +</code>
 +
 +----
 +
 +==== glxinfo ====
 +
 +glxinfo shows information about the OpenGL and GLX implementations running on a given X display.
 +
 +<code bash>
 +sudo apt update
 +sudo apt install mesa-utils
 +glxinfo -B
 +</code>
 +
 +returns:
 +
 +<code bash>
 +name of display: :0
 +display: :0  screen: 0
 +direct rendering: Yes
 +Extended renderer info (GLX_MESA_query_renderer):
 +    Vendor: AMD (0x1002)
 +    Device: Radeon RX 7900 XTX (gfx1100, LLVM 15.0.3, DRM 3.48, 5.19.0-43-generic) (0x744c)
 +    Version: 22.3.0
 +    Accelerated: yes
 +    Video memory: 24576MB
 +    Unified memory: no
 +    Preferred profile: core (0x1)
 +    Max core profile version: 4.6
 +    Max compat profile version: 4.6
 +    Max GLES1 profile version: 1.1
 +    Max GLES[23] profile version: 3.2
 +Memory info (GL_ATI_meminfo):
 +    VBO free memory - total: 23811 MB, largest block: 23811 MB
 +    VBO free aux. memory - total: 32053 MB, largest block: 32053 MB
 +    Texture free memory - total: 23811 MB, largest block: 23811 MB
 +    Texture free aux. memory - total: 32053 MB, largest block: 32053 MB
 +    Renderbuffer free memory - total: 23811 MB, largest block: 23811 MB
 +    Renderbuffer free aux. memory - total: 32053 MB, largest block: 32053 MB
 +Memory info (GL_NVX_gpu_memory_info):
 +    Dedicated video memory: 24576 MB
 +    Total available memory: 56703 MB
 +    Currently available dedicated video memory: 23811 MB
 +OpenGL vendor string: AMD
 +OpenGL renderer string: Radeon RX 7900 XTX (gfx1100, LLVM 15.0.3, DRM 3.48, 5.19.0-43-generic)
 +OpenGL core profile version string: 4.6 (Core Profile) Mesa 22.3.0-devel
 +OpenGL core profile shading language version string: 4.60
 +OpenGL core profile context flags: (none)
 +OpenGL core profile profile mask: core profile
 +
 +OpenGL version string: 4.6 (Compatibility Profile) Mesa 22.3.0-devel
 +OpenGL shading language version string: 4.60
 +OpenGL context flags: (none)
 +OpenGL profile mask: compatibility profile
 +
 +OpenGL ES profile version string: OpenGL ES 3.2 Mesa 22.3.0-devel
 +OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20
 +</code>
 +
  
 ---- ----
Line 11: Line 267:
 <code bash> <code bash>
 sudo apt install radeontop sudo apt install radeontop
 +</code>
 +
 +<code bash>
 +radeontop
 </code> </code>
  
ubuntu/gpu/amd_gpu/install_tools.1685904195.txt.gz · Last modified: 2023/06/04 18:43 by peter
Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki