Any of you faggots have any open source projects worth contributing to? Let's see em

Imagine boning Abigail every night and having him as a BIL. Shit would be so cash.

Attached: Ben_Shapiro_(1)_810_500_75_s_c1[1].jpg (810x500, 23K)

>take small FOSS project
>incorporate it into my own project or change it a bit to make the similarity not immediately noticable
>distribute compiled binaries without source code and my own proprietary license
Now what?

wpgtk

>Any of you faggots have any open source projects worth contributing to? Let's see em

I'll shamelessly shill my own repo, if no one else will: github.com/enfiskutensykkel/ssd-gpu-dma

Rate or hate, Jow Forums

kek

Attached: 823451.jpg (460x460, 49K)

Yes?

ASHKENAZI
AREOLAS

Hey man I just wanted to say thank you so much. I've been struggling to figure out my resume and it's been causing me a ton of stress because all the templates out there are garbage, but your LaTeX CV is beautiful and I downloaded it and it's the best resume I've ever created. It seriously means a lot and I hope that your open source projects go well, I would contribute but that stuff isn't up my alley. I wish you the best of luck in life and I just wanted to let you know that you made my week, thank you :)

You are become korea
Red star OS ftw

Link???

Thanks user. That's really nice of you to say. I hope stuff goes well for you too.

fAgGoTs

Lol who is this virgin incel user? He looks like a pedophile who sells ice cream cones just so he can touch hands of kids when they give him money or take the cone.

What? It's just a over exposed photo of some dude, no need to be a prick.

>post github repo
>people discuss profile picture
Every time

If he wore some nice clothes, got a Chad haircut, lost a few pounds, and had good lighting he would probably put you to shame. The kind of person who insults others over the internet has serious self esteem issues, look in the mirror user and ask yourself is this really the best version of you there is?

I expected more from Jow Forums

Attached: 7dee6d8e879c55ccf4d07cf53c153cf061b968d2bb7a4f6034a064e3df769dad.jpg (612x780, 71K)

I'm the guy in the photo. Can we get back to discussing programming projects instead, please?

Nice repo, I wish I had real projects in mine but I really only program for small niche uses, using C++ as a scripting language essentially.

Are those nude pics real?

thanks but my project already has a logo

If I add a single function that does something trivial would you be willing to let me push it to your repo? I need something to put on my resume.

Post feet
Any nudes?

I'm talking about Abigail tho

If it does something useful and you've tested it, sure.

Thanks.

>his solution to problem 81 on project euler runs in 4 (four) minutes even though he just implemented stock dijkstra
this is your brain on college level informatics, take notice

Lol, yeah, you got me there. :)

github.com/EvilPudding/candle

I've followed you for some time in the /dpt/, user. That's really cool work you do, keep it up.

Contribute to Godot, they're doing Vulkan port now.

why even live

What userspace software uses this library?

get out

>open source projects
>worth contributing to
pick one, communist

t. microsoft employee

>What userspace software uses this library?
At the moment, I'm not sure what people do with my repo. I got an email once from somebody from Samsung, and there's one guy who's contributed stuff for 32-bit ARM, so I guess he's doing some custom NVMe stuff on embedded maybe?

A couple of engineers from Nvidia have also shown interest in it, and I know that they are making some low-level direct block access mechanism for CUDA, at least on their DGX-2 box.

There are also some startup applications that I think are interested in using it for GPU accelerated software, such as machine learning and databases (they also tried to hire me). Looking into making an extension for Tensorflow is something I plan on doing in the long run.

For my part, I made it primarily to demonstrate a couple of things:
- It's possible to read/write directly to GPU memory from the NVMe disk, without copying via RAM (zero copy) = maximum bandwidth using peer-to-peer DMA.
- It's possible to control the disk from a GPU kernel, completely eliminating the CPU in the control path as well = minimal latency for random access disk access and for long running/persistent CUDA kernels.
- I can share a disk between multiple GPUs simultaneously.

There's also a second "mode" of the library, where I use a shared-memory API for PCIe clusters. This requires PCIe interconnection hardware and the semi-proprietary driver stack that goes along with it, but it allows me to share a disk simultaneously between multiple computers (and even their GPUs) in the cluster and using the fastest possible data path for I/O, entirely eliminating the need for NVMeoF.

Using the shared-memory API, I can even do "clever" stuff like hosting command queues in memory close to the disk, and posting completions to memory close to the completer, which greatly reduces the latency. I gave a talk about this on GTC this year, if you're interested.

seething

One Yike

Slightly unrelated, but how old are you?

32, why?

Free Software > Open Source
gnu.org/not-open-source

God dammit I want to motorboat those Khazar milkers so bad

This

I wouldn't associate myself with this place if I were you, but maybe that's just me.