Constructive Criticism

Here in this thread post your program and other programmers will review the code and give you constructive criticism, please be civil and do not be rude, the purpose of this thread is to help you become a better programmer and not to make you feel bad about yourself.

Attached: 1517970949338.jpg (500x375, 70K)

Other urls found in this thread:

github.com/enfiskutensykkel/ssd-gpu-dma
gitlab.com/luanerdus/orbital
pastebin.com/xgbJMdav
pastebin.com/ymqhydgx
pastebin.com/dqg0C4Em
pastebin.com/NHhA3d7m
github.com/zerrain/LibraryApplicationJava
twitter.com/NSFWRedditImage

It's shit.

don't be shy, show us your code

#include #define S "Hello, World\n" main(){exit(printf(S) == strlen(S) ? 0 : 1);}

nice forkbomb

Attached: you.jpg (621x702, 16K)

Neat is that actually yours?

Yes.

Not sure why it was deleted, that wasn't me.

post it again? i didnt see >:*(

CUDA support is interesting. Any plans on adding opencl or spir v?

Sure.

github.com/enfiskutensykkel/ssd-gpu-dma

>Any plans on adding opencl
If there's enough interest, I might look into adding ROCm support for it, it seemed pretty straight forward. My immediate plans is to simplify the driver path though, currently I have two different code paths depending on if you use my driver or the SmartIO one, which is painful to maintain.

Huh. I only use cuda/opencl a few times for my parallel computing class so I only have a basic knowledge. Seems neat though. What applications do you have in mind?

Well, to be honest, I started out just wanting to see if it was possible to control an NVMe disk directly from a GPU without involving the CPU at all. But based on the projects of people who are showing some interest in it, I think building some sort of framework for loading your machine learning training set onto your GPU as efficiently as possible is one possible application. There's obviously potential for sharing an NVMe disk between multiple nodes in a cluster as well, and/or between multiple VMs. I've been approached by a couple of companies that are interested in this.

>Not sure why it was deleted, that wasn't me.
Probably a fluke in hiromoots programming

Pretty cool. That sounds like job security. Machine learning is very big right now. Despite what Jow Forums says, it works as good or better than humans in done cases. I also took a class on machine learning. Tried to use reinforcement to drive a car across a procedurally generated bumpy 2d terrain. My conclusion, 5% off the time, it would work. Were hoping it would work most of the time. Nope. That was my fail project for the class. Teacher loved it though. Didn't get 100% because i used bellow instead of below...

Final project. Not fail

Here's what I'm currently working on, criticism would be appreciated: gitlab.com/luanerdus/orbital

I wrote for myself a rss reader in Go that fetches addresses in parallel and serves single-page feed with built-in http server and templates. The thing has 195 lines (166 loc) from which 30 lines are hard-coded addresses and 18 is the html template. I haven't worked on it since first time I made it work, the TODO list would be pretty big. Using 3rd party library for parsing the RSS/Atom feeds and doing some manual transformation for youtube channels.
I've posted it here before and allegedly the interface is very ugly :^)

main.go pastebin.com/xgbJMdav
item.go (only to have them sortable, wonder if this stupid boilerplate could be solved with closure instead) pastebin.com/ymqhydgx
config.go (stripped from most subscriptions) pastebin.com/dqg0C4Em

planning to rewrite it in Perl and coreutils but still not sick of it enough to touch it

Attached: rss-reader.png (675x640, 256K)

pastebin.com/NHhA3d7m

Is it an IRC server implementation?

Yes

moon is something related to Lua?

Moonscript is a language that compiles to lua

Then this sentence is a bit silly, considering how IRC already is a decentralized service:
>Orbital promotes the decentralizaton of IRC by making it easy to run and maintain a private IRC server.

IRC clients and servers are pretty much babby's first networking program. Even though I don't know what language it is in, it seems logically structured based on file names and module names.

Sure it's technically decentralized, but many people have channels for their projects on servers that other people are running, not themselves.

>but many people have channels for their projects on servers that other people are running
You mean clusters of servers, any EFNet channel is available through a myriad of EFNet servers that are maintained and ran by different people.

I do mean servers, because it becomes a privacy concern that anyone running a server can see your private conversations. It's safest to run your own, and that's the goal of orbital. To make it easy to run and maintain a private IRC server.

>because it becomes a privacy concern that anyone running a server can see your private conversations
Then you should rather rephrase and talk about how to make it easy to run a *private* IRC server, if that is really your goal.

do you have some shareable resources on server-side IRC protocols and implementations? or tip on some simple reference implementation

The IRC RFC is pretty straight forward and almost acts as a reference implementation.

I used modern.ircdocs.horse as a reference when developing orbital. RFC is a bit dated nowadays.

Bump

This is actually legit and has potential. Is this personal or for work, user?

Hey Jow Forums, ive been working on this project in java which was a past assignment at uni. Essentially it is just a basic library system with all the features shown in the readme as a list of the menus. I still have a few wip parts for it such as saving to and reading from file to store the data and once thats done to then build a complete gui with some more functionality like a special set login key for admin access. Would appreciate any feedback at all, especially on the code itself and how everything is set out, since i feel like it is very poorly made and dont want to get any bad habits.

Main method is in library.java
github.com/zerrain/LibraryApplicationJava

I haven't looked at it, but if it's a lib why does it have a main?

Nevermind, just realized you meant an actual library.

>I started out just wanting to see if it was possible to control an NVMe disk directly from a GPU without involving the CPU at all.
If I understand what you've made correctly, I think you've actually beaten NVIDIA in implementing NVMe support. There has been some speculations if there is going to be NVMe support in CUDA 9.3/9.5 later this year or early next year, seeing how NVMe drives have become tightly integrated in their DGX compute boxes.

I would actually recommend you taking contact with them, user. This is the real deal.

bump

bumgp