WHY GO ? …. Go has first class support for concurrency.

The primary motive for designing a new language was to solve software engineering issues at Google. They also mention that Go was actually developed as an alternative to C++.

Rob Pike mentions the purpose for the Go programming language:

“Go’s purpose is therefore not to do research into programming language design; it is to improve the working environment for its designers and their coworkers. Go is more about software engineering than programming language research. Or to rephrase, it is about language design in the service of software engineering.”

I- Concurrency:

Concurrency is one of the major selling points of Go.

The language designers have designed the concurrency model around the ‘Communicating Sequential Processes’ paper by Tony Hoare.

The Go runtime allows you to run hundreds of thousands of concurrent goroutines on a machine.

A Goroutine is a lightweight thread of execution. The

Go runtime multiplexes those goroutines over operating system threads. That means that multiple goroutines can run concurrently on a single OS thread. The Go runtime has a scheduler whose job is to schedule these goroutines for execution.

There are two benefits of this approach:

i) A Goroutine when initialized has a stack of 4 KB. This is really tiny as compared to a stack of an OS thread, which is generally 1 MB. This number matters when you need to have hundreds of thousands of different goroutines running concurrently. If you would run more than thousands of OS threads in parallel, the RAM obviously will become a bottleneck.

ii) Go could have followed the same model as other languages like Java, which support the same concept of threads as OS threads. But in that case, the cost of a context switch between OS threads is much larger than the cost of a context switch between different goroutines.

Since I’m referring to “concurrency” multiple times in this article, I would advise you to check out Rob Pike’s talk on ‘Concurrency is not parallelism”. In programming, concurrency is the composition of independently executing processes, while parallelism is the simultaneous execution of (possibly related) computations. Unless you have a processor with multiple cores or have multiple processors, you can’t really have parallelism since a CPU core can only execute one thing at a time. On a single core machine, it’s just concurrency that’s doing its job behind the scenes. The OS scheduler schedules different processes (threads actually. every process has atleast a main thread) for different timeslices on the processor. Therefore, at one moment in time, you can only have one thread(process) running on the processor. Due to the high speed of execution of the instructions, we get the feeling that multiple things are running. But it’s actually just one thing at a time.

Concurrency is about dealing with lots of things at once. Parallelism is about doing lots of things at once.

f) Interfaces: Interfaces enable loosely coupled systems. An interface type in Go can be defined as a set of functions. That’s it. Any type which implements those functions implicitly implements the interface, i.e. you don’t need to specify that a type implements the interface. This is checked by the compiler automatically at compile time.

This means that a part of your code can just rely on an interface type and doesn’t really care about who implements the interface or how the interface is actually implemented. Your main/controller function can then supply a dependency which satisfies the interface (implements all the functions in the interface) to that code. This also enables a really clean architecture for unit testing (through dependency injection). Now, your test code can just inject a mock implementation of the interface required by the code to be able to test if it’s doing its job correctly or not.

While this is great for decoupling, the other benefit is that you then start thinking about your architecture as different microservices. Even if your application resides on a single server (if you’re just starting out), you architect different functionalities required in your application as different microservices, each implementing an interface it promises. So other services/controllers just call the methods in your interface not actually caring about how they are implemented behind the scenes.

g) Garbage collection: Unlike C, you don’t need to remember to free up pointers or worry about dangling pointers in Go. The garbage collector automatically does this job.

h) No exceptions, handle errors yourself: I love the fact that Go doesn’t have the standard exception logic that other languages have. Go forces developers to handle basic errors like ‘couldn’t open file’ etc rather than letting them wrap up all of their code in a try catch block. This also puts pressure on developers to actually think about what needs to be done to handle these failure scenarios.

i) Amazing tooling: One of the best aspects about Go is its tooling. It has tools like:

i) Gofmt: It automatically formats and indents your code so that your code looks like the same as every Go developer on the planet. This has a huge effect on code readability.

ii) Go run: This compiles your code and runs it, both :). So even though Go needs to be compiled, this tool makes you feel like it’s an interpreted language since it just compiles your code so fast that you don’t even feel when the code got compiled.

iii) Go get: This downloads the library from GitHub and copies it to your GoPath so that you can import the library in your project

iv) Godoc: Godoc parses your Go source code — including comments — and produces its documentation in HTML or plain text format. Through godoc’s web interface, you can then see documentation tightly coupled with the code it documents. You can navigate from a function’s documentation to its implementation with one click.

You can check more tools here.

j) Great built-in libraries: Go has great built-in libraries to aid modern development. Some of them are:

a) net/http — Provides HTTP client and server implementations

b) database/sql — For interaction with SQL databases

c) encoding/json — JSON is treated as a first class member of the standard language 🙂

d) html/templates — HTML templating library

e) io/ioutil — Implements I/O utility functions

There is a lot of development going on in the Go horizon. You can find all Go libraries and frameworks for all sorts of tools and use cases here.



deep learning frameworks :: Tensorflow, Theano, Caffe, Pytorch, CNTK, MXNet, Torch, deeplearning4j, Caffe2 among many others.

Deep Learning is a branch of AI which uses Neural Networks for Machine Learning. In the recent years.

Deep Learning became a household name for AI engineers since 2012 when Alex Krizhevsky and his team won the ImageNet challenge. ImageNet is a computer vision competition in which the computer is required to correctly classify the image of an object into one of 1000 categories. The objects include different types of animals, plants, instruments, furniture, Vehicles to name a few.

This attracted a lot of attention from the Computer vision community and almost everyone started working on Neural Networks. But at that time, there were not many tools available to get you started in this new domain. A lot of effort has been put in by the community of researchers to create useful libraries making it easy to work in this emerging field. Some popular deep learning frameworks at present are Tensorflow, Theano, Caffe, Pytorch, CNTK, MXNet, Torch, deeplearning4j, Caffe2 among many others.