Minimum compute capability
R2026bMinimum CUDA compute capability for code generation
Description
App Configuration Pane: GPU Code
Configuration Objects: coder.GpuCodeConfig
The Minimum compute capability parameter specifies the minimum compute capability for which GPU Coder™ generates CUDA® code.
CUDA GPU compute capability is a numeric representation of the hardware features and instruction sets available for a GPU architecture. Selecting a higher compute capability enables new CUDA features, but you cannot run the compiled code on a GPU whose compute capability is less than the value of the Minimum compute capability parameter. For code generation, different targets such as CUDA MEX functions or standalone executables have different compute capability requirements.
This table lists the CUDA compute capability requirements for code generation.
| Target | Compute Capability |
|---|---|
CUDA MEX | |
Source code, static or dynamic library, and executables | 3.2 or higher. |
Deep learning applications in 8-bit integer precision | 6.1, 7.0 or higher. |
Deep learning applications in half-precision (16-bit floating point) | 5.3, 6.0, 6.2 or higher. |
Note
If you specify the Custom compute capability parameter, GPU Coder ignores this setting and uses the custom compute capability.
Settings
AutoThis setting is the default setting.
Uses the compute capability of the GPU device that you select for GPU code generation. If no GPU device is available, the code generator uses a compute capability value of 7.5.
3.2|3.5|3.7|5.0|5.2|5.3|6.0|6.1|6.2|7.0|7.2|7.5|8.0|8.6|8.7|8.9|9.0|10.0|10.3|11.0|12.0|12.1Builds generated code for the specified GPU compute capability.
Programmatic Use
Property:
ComputeCapability |
Values: "Auto" |
"3.2" | "3.5" | "3.7" |
"5.0" | "5.2" | "5.3" |
"6.0" | "6.1" | "6.2" |
"7.0" | "7.2" | "7.5" |
"8.0" | "8.6" | "8.7" |
"8.9" | "9.0" | "10.0"
| "10.3" | "11.0" | "12.0"
| "12.1" |
Default: "Auto" |