Contents

• 1. Introduction
• 2. Using the cuFFT API
  • 2.1. Accessing cuFFT
  • 2.2. Fourier Transform Setup
    • 2.2.1. Free Memory Requirement
    • 2.2.2. Plan Initialization Time
  • 2.3. Fourier Transform Types
    • 2.3.1. Half-precision cuFFT Transforms
    • 2.3.2. Bfloat16-precision cuFFT Transforms
  • 2.4. Data Layout
  • 2.5. Multidimensional Transforms
  • 2.6. Advanced Data Layout
  • 2.7. Streamed cuFFT Transforms
  • 2.8. Multiple GPU cuFFT Transforms
    • 2.8.1. Plan Specification and Work Areas
    • 2.8.2. Helper Functions
    • 2.8.3. Multiple GPU 2D and 3D Transforms on Permuted Input
    • 2.8.4. Supported Functionality
  • 2.9. cuFFT Callback Routines
    • 2.9.1. Overview of the cuFFT Callback Routine Feature
    • 2.9.2. LTO Load and Store Callback Routines
      • 2.9.2.1. Specifying LTO Load and Store Callback Routines
      • 2.9.2.2. LTO Callback Routine Function Details
    • 2.9.3. Legacy Load and Store Callback Routines
      • 2.9.3.1. Specifying Legacy Load and Store Callback Routines
      • 2.9.3.2. Legacy Callback Routine Function Details
    • 2.9.4. Coding Considerations for the cuFFT Callback Routine Feature
      • 2.9.4.1. Coding Considerations for LTO Callback Routines
      • 2.9.4.2. Coding Considerations for Legacy Callback Routines
  • 2.10. Thread Safety
  • 2.11. CUDA Graphs Support
  • 2.12. Static Library and Callback Support
    • 2.12.1. Static library without legacy callback support
  • 2.13. Accuracy and Performance
  • 2.14. Caller Allocated Work Area Support
  • 2.15. cuFFT Link-Time Optimized Kernels
• 3. cuFFT API Reference
  • 3.1. Return value cufftResult
  • 3.2. cuFFT Basic Plans
    • 3.2.1. cufftPlan1d()
    • 3.2.2. cufftPlan2d()
    • 3.2.3. cufftPlan3d()
    • 3.2.4. cufftPlanMany()
  • 3.3. cuFFT Extensible Plans
    • 3.3.1. cufftCreate()
    • 3.3.2. cufftDestroy()
    • 3.3.3. cufftMakePlan1d()
    • 3.3.4. cufftMakePlan2d()
    • 3.3.5. cufftMakePlan3d()
    • 3.3.6. cufftMakePlanMany()
    • 3.3.7. cufftMakePlanMany64()
    • 3.3.8. cufftXtMakePlanMany()
  • 3.4. cuFFT Plan Properties
    • 3.4.1. cufftSetPlanPropertyInt64()
    • 3.4.2. cufftGetPlanPropertyInt64()
    • 3.4.3. cufftResetPlanProperty()
  • 3.5. cuFFT Estimated Size of Work Area
    • 3.5.1. cufftEstimate1d()
    • 3.5.2. cufftEstimate2d()
    • 3.5.3. cufftEstimate3d()
    • 3.5.4. cufftEstimateMany()
  • 3.6. cuFFT Refined Estimated Size of Work Area
    • 3.6.1. cufftGetSize1d()
    • 3.6.2. cufftGetSize2d()
    • 3.6.3. cufftGetSize3d()
    • 3.6.4. cufftGetSizeMany()
    • 3.6.5. cufftGetSizeMany64()
    • 3.6.6. cufftXtGetSizeMany()
  • 3.7. cufftGetSize()
  • 3.8. cuFFT Caller Allocated Work Area Support
    • 3.8.1. cufftSetAutoAllocation()
    • 3.8.2. cufftSetWorkArea()
    • 3.8.3. cufftXtSetWorkAreaPolicy()
  • 3.9. cuFFT Execution
    • 3.9.1. cufftExecC2C() and cufftExecZ2Z()
    • 3.9.2. cufftExecR2C() and cufftExecD2Z()
    • 3.9.3. cufftExecC2R() and cufftExecZ2D()
    • 3.9.4. cufftXtExec()
    • 3.9.5. cufftXtExecDescriptor()
  • 3.10. cuFFT and Multiple GPUs
    • 3.10.1. cufftXtSetGPUs()
    • 3.10.2. cufftXtSetWorkArea()
    • 3.10.3. cuFFT Multiple GPU Execution
      • 3.10.3.1. cufftXtExecDescriptorC2C() and cufftXtExecDescriptorZ2Z()
      • 3.10.3.2. cufftXtExecDescriptorR2C() and cufftXtExecDescriptorD2Z()
      • 3.10.3.3. cufftXtExecDescriptorC2R() and cufftXtExecDescriptorZ2D()
    • 3.10.4. Memory Allocation and Data Movement Functions
      • 3.10.4.1. cufftXtMalloc()
        • 3.10.4.1.1. Parameter cufftXtSubFormat
      • 3.10.4.2. cufftXtFree()
      • 3.10.4.3. cufftXtMemcpy()
        • 3.10.4.3.1. Parameter cufftXtCopyType
    • 3.10.5. General Multiple GPU Descriptor Types
      • 3.10.5.1. cudaXtDesc
      • 3.10.5.2. cudaLibXtDesc
  • 3.11. cuFFT Callbacks
    • 3.11.1. cufftXtSetJITCallback()
    • 3.11.2. cufftXtSetCallback()
    • 3.11.3. cufftXtClearCallback()
    • 3.11.4. cufftXtSetCallbackSharedSize()
  • 3.12. cufftSetStream()
  • 3.13. cufftGetVersion()
  • 3.14. cufftGetProperty()
  • 3.15. cuFFT Types
    • 3.15.1. Parameter cufftType
    • 3.15.2. Parameters for Transform Direction
    • 3.15.3. Type definitions for callbacks
      • 3.15.3.1. Type definitions for LTO callbacks
      • 3.15.3.2. Type definitions for legacy callbacks
    • 3.15.4. Other cuFFT Types
      • 3.15.4.1. cufftHandle
      • 3.15.4.2. cufftReal
      • 3.15.4.3. cufftDoubleReal
      • 3.15.4.4. cufftComplex
      • 3.15.4.5. cufftDoubleComplex
  • 3.16. Common types
    • 3.16.1. cudaDataType
    • 3.16.2. libraryPropertyType
• 4. Multiple GPU Data Organization
  • 4.1. Multiple GPU Data Organization for Batched Transforms
  • 4.2. Multiple GPU Data Organization for Single 2D and 3D Transforms
  • 4.3. Multiple-GPU Data Organization for Single 1D Transforms
• 5. FFTW Conversion Guide
• 6. FFTW Interface to cuFFT
• 7. Deprecated Functionality
• 8. Notices
  • 8.1. Notice
  • 8.2. OpenCL
  • 8.3. Trademarks