llama.swiftui : add bench functionality (#4483)

* llama.swiftui : add bench button

* llama.swiftui : initial bench functionality

* force to use n_gpu_layers on simulator

* add download buttons & expose llamaState.loadModel

* update project.pbxproj

* comment #Preview & fix editorconfig check

* gitignore : xcode stuff

* llama.swiftui : UX improvements

* llama.swiftui : avoid data copy via "downloadTask"

* llama.swiftui : remove model from project

* llama : remove "mostly" from model infos

* llama.swiftui : improve bench

---------

Co-authored-by: jhen <developer@jhen.me>

examples/llama.swiftui/llama.swiftui/Models/LlamaState.swift

@@ -3,24 +3,26 @@ import Foundation
 @MainActor
 class LlamaState: ObservableObject {
     @Published var messageLog = ""
+    @Published var cacheCleared = false
 
     private var llamaContext: LlamaContext?
-    private var modelUrl: URL? {
-        Bundle.main.url(forResource: "q8_0", withExtension: "gguf", subdirectory: "models")
+    private var defaultModelUrl: URL? {
+        Bundle.main.url(forResource: "ggml-model", withExtension: "gguf", subdirectory: "models")
         // Bundle.main.url(forResource: "llama-2-7b-chat", withExtension: "Q2_K.gguf", subdirectory: "models")
     }
+
     init() {
         do {
-            try loadModel()
+            try loadModel(modelUrl: defaultModelUrl)
         } catch {
             messageLog += "Error!\n"
         }
     }
 
-    private func loadModel() throws {
+    func loadModel(modelUrl: URL?) throws {
         messageLog += "Loading model...\n"
         if let modelUrl {
-            llamaContext = try LlamaContext.createContext(path: modelUrl.path())
+            llamaContext = try LlamaContext.create_context(path: modelUrl.path())
             messageLog += "Loaded model \(modelUrl.lastPathComponent)\n"
         } else {
             messageLog += "Could not locate model\n"
@@ -31,7 +33,7 @@ class LlamaState: ObservableObject {
         guard let llamaContext else {
             return
         }
-        messageLog += "Attempting to complete text...\n"
+
         await llamaContext.completion_init(text: text)
         messageLog += "\(text)"
 
@@ -42,4 +44,42 @@ class LlamaState: ObservableObject {
         await llamaContext.clear()
         messageLog += "\n\ndone\n"
     }
+
+    func bench() async {
+        guard let llamaContext else {
+            return
+        }
+
+        messageLog += "\n"
+        messageLog += "Running benchmark...\n"
+        messageLog += "Model info: "
+        messageLog += await llamaContext.model_info() + "\n"
+
+        let t_start = DispatchTime.now().uptimeNanoseconds
+        await llamaContext.bench(pp: 8, tg: 4, pl: 1) // heat up
+        let t_end = DispatchTime.now().uptimeNanoseconds
+
+        let t_heat = Double(t_end - t_start) / 1_000_000_000.0
+        messageLog += "Heat up time: \(t_heat) seconds, please wait...\n"
+
+        // if more than 5 seconds, then we're probably running on a slow device
+        if t_heat > 5.0 {
+            messageLog += "Heat up time is too long, aborting benchmark\n"
+            return
+        }
+
+        let result = await llamaContext.bench(pp: 512, tg: 128, pl: 1, nr: 3)
+
+        messageLog += "\(result)"
+        messageLog += "\n"
+    }
+
+    func clear() async {
+        guard let llamaContext else {
+            return
+        }
+
+        await llamaContext.clear()
+        messageLog = ""
+    }
 }