Added code for comparison between x86_64 with arm64 (graviton2) lambda functions

Author: eitansela

img/xgboost_x86_64_arm64_improvement.png

@@ -8,6 +8,7 @@ This project contains source code and supporting files for a serverless applicat
 - train-code - Code for training XGBoost model based on breast cancer dataset.
 - events - Invocation events that you can use to invoke the function.
 - template.yaml - A template that defines the application's AWS resources.
+- invoke_x86_64_arm64_lambdas.py - Code to invoke x86_64 and Arm64 (Graviton2) Lambda Functions and compare performance.  
 
 The application uses several AWS resources, including Lambda functions. These resources are defined in the `template.yaml` file in this project. You can update the template to add AWS resources through the same deployment process that updates your application code.
 
@@ -63,6 +64,20 @@ xgboost-inference-arm64-docker-lambda$ sam local invoke XGBoostInferenceArm64Fun
 
 ![arm64 Architecture](../img/xgboost_arm_64_arch_view.png)
 
+## Compare performance of x86_64 Lambda Function with Arm64 (Graviton2) one
+
+You can achieve up to 34% better price/performance with AWS Lambda Functions powered by AWS Graviton2 processor.
+
+To test the performance improvements of the XGBoost x86_64 Lambda Function with an Arm64 (Graviton2) one yourself, please execute the following steps:
+1. Deploy the XGBoost x86_64 Lambda Function, as described [here](../xgboost-inference-docker-lambda/)
+2. Open `invoke_x86_64_arm64_lambdas.py` file.
+3. Replace `<XGBoost_x86_64_Lambda_ARN>` with XGBoost x86_64 Lambda Function ARN.
+4. Replace `<XGBoost_arm64_Lambda_ARN>` with XGBoost Arm64 (Graviton2) Lambda Function ARN.
+5. Run the following command: `python invoke_x86_64_arm64_lambdas.py`
+
+![xgboost x86_64 vs. arm64 improvement](../img/xgboost_x86_64_arm64_improvement.png)
+
+**We can see that the Arm64 (Graviton2) Lambda Function has performance improvements of 36% over the x86_64 one!**
 
 ## Testing your Lambda function in the Cloud
 

xgboost-inference-arm64-docker-lambda/README.md

@@ -0,0 +1,74 @@
+import boto3
+import json
+import base64
+import re
+import sys
+
+client = boto3.client('lambda')
+
+f= open('./events/event.json')
+json_input = json.load(f)
+
+
+def call_XGBoost_x86_64_Lambda():
+    response = client.invoke(
+        FunctionName='<XGBoost_x86_64_Lambda_ARN>',
+        InvocationType='RequestResponse',
+        LogType='Tail',
+        Payload=json.dumps(json_input)
+    )
+    return(find_duration(response))
+
+
+def call_XGBoost_arm64_Lambda():
+    response = client.invoke(
+        FunctionName='<XGBoost_arm64_Lambda_ARN>',
+        InvocationType='RequestResponse',
+        LogType='Tail',
+        Payload=json.dumps(json_input)
+    )
+
+    return(find_duration(response))
+
+
+def find_duration(response):
+    log_result = base64.b64decode(response['LogResult']).decode('utf-8')
+    result = re.findall(r'Duration: (\d+\.\d+) ms', log_result)
+    return (float(result[0]))
+
+# Warm the two Lambda Functions first
+print("Warming x86_64 and arm64 Lambda Functions")
+call_XGBoost_x86_64_Lambda()
+call_XGBoost_arm64_Lambda()
+
+print("Warming x86_64 and arm64 Lambda Functions - Done")
+print("Sending events to x86_64 and arm64 Lambda Functions")
+
+counter=1
+num_calls=100
+total_duration_x86_64=0
+total_duration_arm64=0
+
+while counter < num_calls:
+    duration_x86_64 = call_XGBoost_x86_64_Lambda()
+    # print(f'call_XGBoost_x86_64_Lambda duration: {duration_x86_64}')
+    total_duration_x86_64 = total_duration_x86_64 + duration_x86_64
+
+    duration_arm64 = call_XGBoost_arm64_Lambda()
+    # print(f'call_XGBoost_arm64_Lambda duration: {duration_arm64}')
+    total_duration_arm64 = total_duration_arm64 + duration_arm64
+
+    sys.stdout.write(('=' * counter) + ('' * (num_calls - counter)) + ("\r [ %d" % counter + "% ] "))
+    sys.stdout.flush()
+
+    counter = counter + 1
+
+print("\nSending events to x86_64 and arm64 Lambda Functions - Done")
+
+avg_duration_x86_64 = total_duration_x86_64/num_calls
+avg_duration_arm64 = total_duration_arm64/num_calls
+improvement_percentage= "{:.0%}".format(1 - (avg_duration_arm64 / avg_duration_x86_64))
+
+print('Average duration x86_64: {:.2f} ms'.format(total_duration_x86_64/num_calls))
+print('Average duration arm64: {:.2f} ms'.format(total_duration_arm64/num_calls))
+print(f'*** Improvement of arm64 (Graviton2) over x86_64: {improvement_percentage} ***')