From 8c107baf9465461d831c0cc2cdc71ccc756b574a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?St=C3=A9phane=20Tetsing?= <stephane.fosso@yahoo.fr>
Date: Mon, 2 Sep 2024 10:18:41 +0200
Subject: [PATCH] linting and handle resources

---
 .../src/lib/InferenceServerManager.ts         | 40 +++++++++++++++++--
 apps/remixdesktop/src/lib/databatcher.ts      |  2 +-
 .../src/helpers/inferenceServerReleases.ts    |  1 -
 .../src/prompts/completionPrompts.ts          |  2 +-
 4 files changed, 39 insertions(+), 6 deletions(-)

diff --git a/apps/remixdesktop/src/lib/InferenceServerManager.ts b/apps/remixdesktop/src/lib/InferenceServerManager.ts
index 02d992f700..b6faa8b4a4 100644
--- a/apps/remixdesktop/src/lib/InferenceServerManager.ts
+++ b/apps/remixdesktop/src/lib/InferenceServerManager.ts
@@ -84,6 +84,9 @@ export class InferenceManager implements ICompletions {
 
       if (this.inferenceProcess === null) await this._startServer()
 
+      // check if resources are met before initializing the models
+      this._handleResources(true)
+
       console.log('Initializing model request', model.modelType)
       switch (model.modelType) {
       case ModelType.CODE_COMPLETION_INSERTION || ModelType.CODE_COMPLETION:{
@@ -125,6 +128,8 @@ export class InferenceManager implements ICompletions {
   }
 
   async _processStatus() {
+
+    // check if the server is running
     const options = { headers: { 'Content-Type': 'application/json', } }
     const state = await axios.get(this.inferenceURL+"/state", options)
 
@@ -146,7 +151,36 @@ export class InferenceManager implements ICompletions {
       // console.log('completion is runnig', state.data?.completion)
       // console.log('general is runnig', state.data?.general)
     }
+    
+    // this._handleResources()
+  }
 
+  async _handleResources(logger:boolean=false) {
+    // check resrource usage
+    const options = { headers: { 'Content-Type': 'application/json', } }
+    const res = await axios.get(this.inferenceURL+"/sys", options)
+
+    if (res.data?.status) {
+      const max_memory = res.data.memory.total
+      const used_memory = res.data.memory.used
+      const memory_usage = res.data.memory.percent * 100
+      const gpu_available = res.data.gpus
+
+      for (const model of this.selectedModels) {
+        if (model.modelReqs.minSysMemory > max_memory) {
+          if (logger) console.warn('Insufficient memory for the model')
+        }
+
+        if (model.modelReqs.minSysMemory > used_memory) {
+          if (logger) console.warn('Insufficient memory for the model')
+        }
+        if (model.modelReqs.GPURequired) {
+          if (gpu_available.length < 1) {
+            if (logger)console.warn('GPU requiredfor desktop inference but not available')
+          }
+        }
+      }
+    }
   }
 
   async _downloadModel(model:IModel): Promise<string> {
@@ -449,7 +483,7 @@ export class InferenceManager implements ICompletions {
       console.log('model not ready yet')
       return
     }
-    if (GenerationParams.stream_result) {
+    if (params.stream_result) {
       return this._streamInferenceRequest('code_explaining', { code, context, ...params })
     } else {
       return this._makeInferenceRequest('code_explaining', { code, context, ...params }, AIRequestType.GENERAL)
@@ -461,7 +495,7 @@ export class InferenceManager implements ICompletions {
       console.log('model not ready yet')
       return ""
     }
-    if (GenerationParams.stream_result) {
+    if (params.stream_result) {
       return this._streamInferenceRequest('error_explaining', { prompt, ...params })
     } else {
       return this._makeInferenceRequest('error_explaining', { prompt, ...params }, AIRequestType.GENERAL)
@@ -481,7 +515,7 @@ export class InferenceManager implements ICompletions {
     }
     const prompt = buildSolgptPromt(userPrompt, modelOP)
 
-    if (GenerationParams.stream_result) {
+    if (params.stream_result) {
       return this._streamInferenceRequest('solidity_answer', { prompt, ...params })
     } else {
       return this._makeInferenceRequest('solidity_answer', { prompt, ...params }, AIRequestType.GENERAL)
diff --git a/apps/remixdesktop/src/lib/databatcher.ts b/apps/remixdesktop/src/lib/databatcher.ts
index 6957505f6d..a7c20521eb 100644
--- a/apps/remixdesktop/src/lib/databatcher.ts
+++ b/apps/remixdesktop/src/lib/databatcher.ts
@@ -1,4 +1,4 @@
-import {EventEmitter} from 'events';
+import { EventEmitter } from 'events';
 import { StringDecoder } from 'string_decoder';
 // Max duration to batch session data before sending it to the renderer process.
 const BATCH_DURATION_MS = 16;
diff --git a/libs/remix-ai-core/src/helpers/inferenceServerReleases.ts b/libs/remix-ai-core/src/helpers/inferenceServerReleases.ts
index b19d77f556..c97fa11c37 100644
--- a/libs/remix-ai-core/src/helpers/inferenceServerReleases.ts
+++ b/libs/remix-ai-core/src/helpers/inferenceServerReleases.ts
@@ -44,7 +44,6 @@ export async function downloadLatestReleaseExecutable(platform: string, outputDi
     console.log(`Downloading executables for ${platform}..., ${executables} `);
 
     for (const executable of executables) {
-			// sort out the correct executable
       const filePath = path.join(outputDir, executable.name);
       console.log(`Downloading ${executable.name}...`);
       await downloadFile(executable.browser_download_url, filePath);
diff --git a/libs/remix-ai-core/src/prompts/completionPrompts.ts b/libs/remix-ai-core/src/prompts/completionPrompts.ts
index 24c8c0b411..51f71e71ce 100644
--- a/libs/remix-ai-core/src/prompts/completionPrompts.ts
+++ b/libs/remix-ai-core/src/prompts/completionPrompts.ts
@@ -1,4 +1,4 @@
-import { COMPLETION_SYSTEM_PROMPT } from "../types/constants";  
+import { COMPLETION_SYSTEM_PROMPT } from "../types/constants";
 import { IModel } from "../types/types";
 
 export const getInsertionPrompt = (model:IModel, msg_pfx, msg_sfx) => {