SwiftCUBLAS

SwiftCUBLAS is a wrapper for the cuBLAS library, providing GPU-accelerated linear algebra operations in Swift. It includes utilities for matrix operations and a robust suite of tests. The package is tested on the latest CUDA runtime API (v12.5) on both Linux and Windows.

Operating System	Swift Version	CUDA Version	Supported
Linux	6.0	12.*	✅
Windows 11	6.0	12.*	✅

Installation

To include SwiftCUBLAS in your Swift project, add the following line to your Package.swift file:

.package(url:"https://github.com/machineko/SwiftCUBLAS", branch:"main")

Documentation

Docc generated for Swift wrapped API SwiftCUBLAS

CUDA runtime cuBLAS API

Example

@Testfunc testSimpleSGEMMRowMajor()asyncthrows{letcuStatus=CUDevice(index:0).setDevice()
#expect(cuStatus)letm=2letn=2letk=4varA:[Float32]=[1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,]varB:[Float32]=[8.0,7.0,6.0,5.0,4.0,3.0,2.0,1.0,]varC:[Float32]=[Float32](repeating:0.0, count: m * n)varaPointer:UnsafeMutableRawPointer?varbPointer:UnsafeMutableRawPointer?varcPointer:UnsafeMutableRawPointer?defer{
_ = aPointer.cudaAndHostDeallocate()
_ = bPointer.cudaAndHostDeallocate()
_ = cPointer.cudaAndHostDeallocate()}letf32Size= MemoryLayout<Float32>.stride
_ = aPointer.cudaMemoryAllocate(m * k * f32Size)
_ = bPointer.cudaMemoryAllocate(k * n * f32Size)
_ = cPointer.cudaMemoryAllocate(m * n * f32Size)
_ = aPointer.cudaMemoryCopy(fromRawPointer:&A, numberOfBytes:A.count * f32Size, copyKind:.cudaMemcpyHostToDevice)
_ = bPointer.cudaMemoryCopy(fromRawPointer:&B, numberOfBytes:B.count * f32Size, copyKind:.cudaMemcpyHostToDevice)lethandle=CUBLASHandle()varparams=CUBLASParams<Float32>(
fromRowMajor: aPointer!.assumingMemoryBound(to:Float32.self), B: bPointer!.assumingMemoryBound(to:Float32.self),
C: cPointer!.assumingMemoryBound(to:Float32.self), m:Int32(m), n:Int32(n), k:Int32(k), alpha:1.0, beta:0.0)letstatus= handle.sgemm_v2(params:&params)
#expect(status.isSuccessful)C.withUnsafeMutableBytes{ rawBufferPointer invarpointerAddress= rawBufferPointer.baseAddress
letoutStatus= pointerAddress.cudaMemoryCopy(
fromMutableRawPointer: cPointer, numberOfBytes: m * n * f32Size, copyKind:.cudaMemcpyDeviceToHost)
#expect(outStatus.isSuccessful)}cudaDeviceSynchronize()letcExpected=matrixMultiply(m, n, k, A, B, isRowMajor:true)
#expect(cExpected ~= C)}

For more examples check test cases or cublas docs

Testing

Current version of SwiftCUBLAS is tested on Swift 6.0 development branch using swift-testing package and CUDA v12.5