This projects project contains showcases on GreenCoding in java

an initiative to create software creating less carbon emissions

The project consists of two different types of showcases.

1. Measurement of software carbon emissions
2. Good (and bad) GreenCoding practices

The measuring examples are (currently only) designed for Windows machines running on Intel CPU only and rely on Intel's Power Gadget Intel Power Gadget. Please install it first to get the measuring showcases running. The examples could be executed by calling GET Request ( e.g in browser).

The result will look similar to this.

{
  "co2": {
    "1        times": "0,000002 CO2(g)",
    "10       times": "0,000019 CO2(g)",
    "100     times": "0,000194 CO2(g)",
    "1.000    times": "0,001942 CO2(g)",
    "10.000   times": "0,019421 CO2(g)",
    "100.000 times": "0,194212 CO2(g)"
  },
  "times": 1,
  "cumulativeEnergyMwh": "0.004004377417090408",
  "cumulativeEnergyJoule": "0.014415762311856333",
  "averagePower": "12.825902",
  "elapsedTime": "0.373580",
  "cpuEnergyMwh": "0.003167836678514516",
  "dramEnergyMwh": "5.218191713693626E-4",
  "gpuEnergyMwh": "1.831340330557368E-5",
  "avgCpuFrequency": "2.112"
}

1. The energy consumption of CPU, GPU and DRAM is measured using Intel Power Gadget during code execution (in Joule and milli watt hours [mwh])
2. It observes the total CPU utilization and the JVM CPU utilization of this process
3. It applies the ratio of total CPU utilization / process CPU Utilization to measured energy consumption of 1)
4. It multiplies the measured energy with local carbon grid emissions per kwh. This can be configured by setting property

greencoding.emissionsPerKwh=485

or programmatically

EnergyResult.setAvgCo2PerKwh(485);

Default is set to Germany 485gCo2e/kwh

5. The measured Co2-equivalent emissions are linear extrapolated with factors 10/100/1.000/10.000/100.000

simulate and measure the serialization and deserialization of documents in Avro and in Json format to see the difference.

Start the Spring Boot application (GreenCodingShowcaseApp) and execute

GET http://localhost:8080/serialize/json
or
GET http://localhost:8080/serialize/avro

The following request parameters are supported

• includeFilesystem: specify if the data should be serialized from and to disk (=true) or from and to memory ( =false) (Default:false)
• times: specify how often the simulation should be executed (default=1)

ℹ️ the data for this serialization is located in src/main/resources/serialization-testdata.json. In Json its size is 6,16KB and in Avro 2,37KB

simulate and measure a ssl handshake and first request encryptions on server and decryption on client side with an 777KB sized index.html.

Start the Spring Boot application (GreenCodingShowcaseApp) and execute

GET http://localhost:8080/ssl

The following request parameters are supported

• rsaKeyLength: specify the key length of rsa keypair (default=1024)
• aesKeyLength: specify the key length of symmetric AES key (default=128)
• times: specify how often the simulation should be executed (default=1)

Learn by example the impact of memory leaks and bad object creations on performance, energy and carbon emissions.

Open the Unittests in class GreenCodingTests and execute them. After observing the bad behaviour try to fix them.

Learn to use a profiler and optimize some simple Java Code to improve performance and carbon intensity of a simple file import.

A file with 10 million can be generated using MapGenerator utility class. Afterwards, run the MapReaderTests in package profiling.demo1 To understand how and why this is implemented use a profiler (for example build in IntelliJ).

Most likely you should have a look at src/main/resources/application.propreties and check Power Gadget Properties are configured correctly