make pipline optional

metal/paddle-mobile/paddle-mobile/API/Net.swift

@@ -56,10 +56,10 @@ import Foundation
     /// 输入维度，按照 n h w c 方式传入
     @objc public var inputDim: Dim = Dim.init(inDim: [])
     
-    /// 是否使用 MetalPerformanceShaders 进行运算
+    /// 是否使用 MetalPerformanceShaders 进行运算, 运算精度为 32 位时不支持开启 MPS
     @objc public var useMPS: Bool = false
     
-    /// 模型精度 - 当使用模型精度为 Float 16 时 不要开启 useMPS, 暂不支持
+    /// 模型精度
     @objc public var paramPrecision: Precision = .Float32
 
     @objc public init(device: MTLDevice, inParamPointer: UnsafeMutableRawPointer, inParamSize:Int, inModelPointer: UnsafeMutableRawPointer, inModelSize: Int) {

metal/paddle-mobile/paddle-mobile/Src/Operators/Kernels/Base/Kernel.swift

@@ -38,11 +38,21 @@ protocol KernelProtocol {
 }
 
 @objc open class Kernel: NSObject{
-    let pipline: MTLComputePipelineState
-    let functionName: String
-    public init(device: MTLDevice, inFunctionName: String, usePaddleMobileLib: Bool = false, initContext: InitContext) {
-        pipline = device.pipeLine(funcName: inFunctionName, metalLoadMode: initContext.metalLoadMode, metalLibPath: initContext.metalLibPath)
+    
+    private var _pipline: MTLComputePipelineState? = nil
+    
+    var pipline: MTLComputePipelineState {
+        get {
+            return _pipline ?! " pipeline can't be nil "
+        }
+    }
+    
+    let functionName: String?
+    public init(device: MTLDevice, inFunctionName: String?, usePaddleMobileLib: Bool = false, initContext: InitContext) {
         functionName = inFunctionName
+        if let funcName = inFunctionName {
+            _pipline = device.pipeLine(funcName: funcName, metalLoadMode: initContext.metalLoadMode, metalLibPath: initContext.metalLibPath)
+        }
     }
 }
 
@@ -104,7 +114,7 @@ open class BufferToTextureKernel: Kernel {
 @objc open class CusomKernel: Kernel {
     
     public let outputTexture: MTLTexture
-    public init(device: MTLDevice, inFunctionName: String, outputDim: Shape, metalLoadModel: MetalLoadMode, metalLibPath: String?) {
+    public init(device: MTLDevice, inFunctionName: String?, outputDim: Shape, metalLoadModel: MetalLoadMode, metalLibPath: String?) {
         let textureDesc = MTLTextureDescriptor.init()
         textureDesc.textureType = .type2D
         textureDesc.width = outputDim.width

metal/paddle-mobile/paddle-mobile/Src/Operators/Kernels/ConvAddKernel.swift

@@ -145,7 +145,7 @@ class ConvAddKernel<P: PrecisionProtocol>: Kernel, Computable {
                     convDic[key] = conv
                     imageDic[identifyingKey + "_input"] = MPSImage.init(texture: param.input.metalTexture, featureChannels: param.input.tensorDim[1])
                     imageDic[identifyingKey + "_output"] = MPSImage.init(texture: param.output.metalTexture, featureChannels: param.output.tensorDim[1])
-                    super.init(device: device, inFunctionName: "place_holder", initContext: initContext)
+                    super.init(device: device, inFunctionName: nil, initContext: initContext)
                     return
                 }
             }

metal/paddle-mobile/paddle-mobile/Src/Operators/Kernels/Scale.swift

@@ -20,9 +20,9 @@ public class ScaleKernel: CusomKernel {
     public init(device: MTLDevice, shape: Shape, metalLoadMode: MetalLoadMode, metalLibPath: String?) {
         lanczos = MPSImageLanczosScale(device: device)        
         if GlobalConfig.shared.computePrecision == .Float32 {
-            super.init(device: device, inFunctionName: "scale", outputDim: shape, metalLoadModel: metalLoadMode, metalLibPath: metalLibPath)
+            super.init(device: device, inFunctionName: nil, outputDim: shape, metalLoadModel: metalLoadMode, metalLibPath: metalLibPath)
         } else if GlobalConfig.shared.computePrecision == .Float16 {
-            super.init(device: device, inFunctionName: "scale_half", outputDim: shape, metalLoadModel: metalLoadMode, metalLibPath: metalLibPath)
+            super.init(device: device, inFunctionName: nil, outputDim: shape, metalLoadModel: metalLoadMode, metalLibPath: metalLibPath)
         } else {
             fatalError(" unsupport ")
         }

metal/paddle-mobile/paddle-mobile/Src/Operators/Kernels/Texture2DTo2DArrayKernel.swift

@@ -23,15 +23,6 @@ struct Texture2DTo2DArrayParam {
 }
 
 class Texture2DTo2DArrayKernel<P: PrecisionProtocol>: Kernel, Computable{
-    func compute(commandBuffer: MTLCommandBuffer, param: FeedParam<P>) throws {
-        guard let encoder = commandBuffer.makeComputeCommandEncoder() else {
-            throw PaddleMobileError.predictError(message: " encode is nil")
-        }
-        encoder.setTexture(param.input.mtlTexture, index: 0)
-        encoder.setTexture(param.output.metalTexture, index: 1)
-        encoder.dispatch(computePipline: pipline, outTexture: param.input.mtlTexture)
-        encoder.endEncoding()
-    }
     
     required init(device: MTLDevice, param: FeedParam<P>, initContext: InitContext) {
         param.output.initTexture(device: device, inTranspose: [0, 2, 3, 1], computePrecision: GlobalConfig.shared.computePrecision)
@@ -42,6 +33,15 @@ class Texture2DTo2DArrayKernel<P: PrecisionProtocol>: Kernel, Computable{
         } else {
             fatalError()
         }
-        
+    }
+    
+    func compute(commandBuffer: MTLCommandBuffer, param: FeedParam<P>) throws {
+        guard let encoder = commandBuffer.makeComputeCommandEncoder() else {
+            throw PaddleMobileError.predictError(message: " encode is nil")
+        }
+        encoder.setTexture(param.input.mtlTexture, index: 0)
+        encoder.setTexture(param.output.metalTexture, index: 1)
+        encoder.dispatch(computePipline: pipline, outTexture: param.input.mtlTexture)
+        encoder.endEncoding()
     }
 }