What is the most exciting field in CS?

>sucking cock 101
>indianese 101

File: can you feel it.jpg (56KB, 552x510px) Image search: [Google]

56KB, 552x510px

>>8107892

It's not a field per se, but if you include computer engineering FPGAs are really fascinating to work on, if a bit frustrating.

pic related, me useing xilinx ISE

File: 1461427277933.jpg (86KB, 500x598px) Image search: [Google]

86KB, 500x598px

>>8107892
Creating exotic dildos for you mom.

>>8107901
so wtf do those dumb acronyms stand for exactly?

>>8107909

FPGA stands for field programmable array, basically think programmable logic gates. Instead of writing a program for a processor you program THE processor. You can implement a standard processor on an FPGA, but the real fun comes when you make something unique like a special purpose processor.

ISE stands for something that can't be pronounced without going completely insane. It is written by the most evil people to ever walk the earth to torment programmers.

File: 1434645861229.gif (1MB, 300x225px) Image search: [Google]

1MB, 300x225px

>>8107912
>programmable logic gates
wow

AI, almost by definition

>>8107917

what happened to vsauce ? ;_;

>>8107917

If you want to check it out I recommend trying chisel, spinal or cλash rather than starting out with verilog or VHDL.

I use chisel which features a rudimentary simulator that I do most of my work in since it usually works on an actual FPGA if it works on the simulator, unlike VHDL

File: 41d58c253866e9b3b3fe7994331a5fa4.jpg (145KB, 1280x720px) Image search: [Google]

145KB, 1280x720px

3D graphics

>>8107932
>it usually works on an actual FPGA if it works on the simulator, unlike VHDL
If you know which parts of VHDL are synthesizable and which aren't this shouldn't be a problem, right?
I programmed some stuff on FPGAs with VHDL in Uni (a simple processor etc.), I agree it was fun.
I always had the feeling that it was basically just for quick prototyping of digital electronics.
Programm your stuff, test it in simulation, test it on FPGA, then produce some actual chips from your code.
Although, I think my prof said that FPGAs are also used in actual products like mixers and shit.

>>8107954

The thing is that it's often not very obvious when you've missed a signal creating something unsynthesizable, and it will often fail silently, at least on the godawful ISE tools made by the likes of Pol Pot and Stalin.

You can write PHP that is free of vulnerabilities, but you'd rather use a sane language that helps you leverage the abstractions when working with digital circuits.

I believe FPGAs have a pretty bright future, personally I've used them for a system that reconfigured hardware layout from on chip memory to adapt to different tasks, basically changing the entire architecture. This was for a processor performing 2d image convolution, and we could synthesize different designs for different kernel sizes and swap layout in the field.

>>8107959
>I believe FPGAs have a pretty bright future, personally I've used them for a system that reconfigured hardware layout from on chip memory to adapt to different tasks, basically changing the entire architecture.
Yeah, at my uni there is a research group doing similar stuff. Reconfiguring FPGAs at runtime. They also hold lectures about that topic, but I haven't looked into it more.

I'm also a CS/CE student, but I'm currently more into control theory and machine learning.
There is another research group here that solves complex non-linear control problems (mostly in robotics) with machine learning methods (often based on self organizing maps and similar techniques).
I like that because it combines several interesting fields and is applicable to a shit ton of other stuff too. You have to know the current developments in ML (and often new methods are created for these tasks) and you also have to model these problems with classical control theory first, to later compare it to the ML results.
Also, with ML robots can learn really complex stuff, that we really couldn't program without it.
Like catching half filled water bottles that are thrown at them.

>>8107978

Which university are you at? I'm at NTNU, and we have some real wizard shit with FPGAs going on here, pretty underrated. Really exciting field

>>8107892
Maybe NLP or computer vision because of the resulting applications scaring normies or type theory if you are a smug category theory shitter. Surely not embedded systems.

>>8107954
> Although, I think my prof said that FPGAs are also used in actual products like mixers and shit.
FPGAs are used if the total number of units required is too low to absorb the up-front costs of an ASIC.

They're commonly used for military applications (the gov't gets to decide how many units you sell), scientific equipment (you're often creating a single unit), and high-frequency trading (the algorithm will probably change every week).

>>8108025
>Maybe NLP or computer vision because of the resulting applications scaring normies

Industrial NLP research is my job at a tech startup and my current project is literally to automate the work being done by 4 people at our company right now.

Luckily I hate 3 of them so I don't feel guilty right now. Want a laugh? 3 of them majored in "journalism" and now sit all day flagging spam on our system.

Programming theory with hardcore theoretical mathematics applied to it.

>>8108056
>Programming theory with hardcore theoretical mathematics applied to it.

What hardcore theory is that? Most math in comp sci I've seen is basic induction, pigeon hole arguments, or graph traversal.

>>8107978
That sounds really neat. My eyes have been opened lately to how interesting and widely applicable control theory is, same for machine learning. Most recently I saw Google demonstrating a robotic arm learning to pick up arbitrarily shaped objects efficiently, for example by pushing nearby blocks out of the way first.

I study mechanical engineering but I'm getting interested in CS and control systems, and I consider changing direction. I don't know what to go for anymore, it's all pretty interesting.

Seconding the question, where do you study? Once I figure out what path to take I want to do an exchange semester or two.

I'm actually at NTNU as well.

>>8107987
Hello brother. Never thought I'd come across someone from NTNU just like that. Question - if I've only taken ITGK, do you know what else I'd need to take to start understanding FPGAs?

>>8107892
pajeet studies

>>8107892
It doesn't matter because you're going to end up with a job doing web development.

>>8108060
Category theory. I like thinking about programming languages from a categoric framework.

The mathematics you mention is learned in discrete math 101.

Mathematical Logic, Graph theory, etc. is much more rich and difficult than what you see in your intro to discrete math coursework.

See

http://mathoverflow.net/questions/239871/why-is-this-new-result-such-a-big-deal

for example

AI and distributed systems are pretty comfy imo

>>8107892
Algorithms

>>8108138
The sort that sorts

Probably proof systems.

They're writing mathematical proofs that are way beyond human scope.

http://www.nature.com/news/two-hundred-terabyte-maths-proof-is-largest-ever-1.19990

The previous largest was 13GB. The 13GB might have been a little more useful.

>>8107892
Rendering foliage in video games.

After that probably correct hand and foot placement in the environment.

File: ut.jpg (1MB, 5312x2988px) Image search: [Google]

1MB, 5312x2988px

>>8108064

If you're still there, I'll offer you the path to enlightment:

Data GK -> Datamaskinkonstruksjon -> Datamaskinprosjekt.

Datamaskinprosjekt is the best course I have ever taken, if you have questions just ask away, or send me an email (my name, the one with two a's in pic related + gmail.com with no extra symbols)

I highly urge you to either shoot me a mail or talk to Gunnar Tufte.

>>8108199
Thanks a lot. I won't take it for granted that those three are all I need to know - I see data GK already builds on Krets og digitalteknikk and OOP. I'm planning on learning the latter this summer. Apart from Matlab there's no programming in MechE, so I've barely gotten my feet wet.

There's two names with aa/å in your picture, by the way. Your surname starts with å?

>>8108199
I realise I've read about your project before, in this article.
https://gemini.no/2015/12/tre-maneder-pa-a-gjore-det-umulige/
That sounded like a very rewarding project.

Thanks again for the tips, I have to get back to studying for my exam now.

>>8108248

fug, you're right. okay, just shoot me a mail at [email protected]

I'm currently applying for a ph.d with Gunnar Tufte working on the cyborg project, and that's pretty much all thanks to datamaskinprosjektet.

If you want to have a chat I usually sit @ glos, just send me a mail or something and I'll show you how to get into FPGA stuff and who to talk to/which courses you need.

>>8108258

dont worry, the beard is long gone.

>>8107892
diybio

>>8108260
Thanks again, you're very kind. I'll see how far I've come at the end of the summer. I'm scatterbrained.

>>8107946
fast inverse square root changed the world

>>8107909
It's important to remember that technology related acronyms don't actually stand for anything.
They're composed of tech-related words, but the words themselves are always super vague.
You just have to memorize the acronyms along with the rest of the jargon.

Xillinx ISE -> Integrated Synthesis Environment
It's a HDL, which is a fancy way of saying it's a markup syntax used when designing chips.
ISE is proprietary, and made by Xillinx.

FPGA -> Field-Programmable Gate Array
A chip containing a giant, inefficient shit-sprawl of logic gates and registers.
A large mess of gates checks its registers to figure out what to do with its inputs.
The end result is that it simulates a less shitty circuit, based on what is stored in it's registers.
The manufacturers of FGPAs are also the designers of HDLs, proprietary bullshit ahoy!


Anyways,
the two together can be used to design and prototype circuits without having to build shit constantly.

>>8107901
Why do you still use those and not just program using software

>>8108603
because to program something with software, you need something to program and to run the program.
Most of the time, using tailor-made electronics can save you a lot of money/space/other.
You're not gonna buy a computer to run a toy car.
Also you can actually control everything, as opposed to being limited to functionalities provided for you.

>>8108603
It's faster

File: mobile-15164-1412713382-24.jpg (20KB, 563x582px) Image search: [Google]

20KB, 563x582px

>>8107912
That's fucking awesome

>>8107901
Are you angry frog or feels man?

>>8107912
>hyping up FPGAs this much
m8 it's not all that

unless you're doing hardcore signal processing you're better off just using a normal microprocessor

>>8109084
FPGAs are one of the nails on the EE job market coffin.

>>8109886
if you can lose your job because of that, you weren't that good in the first place.

>>8109921
if you were that good in the first place you wouldn't have been an EE

>>8109886
this is one of the most bizarre things I've ever read

FPGAs ARE a significant part of the EE job market

>>8109927
kek
I wonder what you are

>>8109936
>>8109927
so? Don't be shy.

>>8107901
Does the phrase "smart robot challenge" mean anything to you? If it doesn't, just ignore this.

File: DickVanDykeLeL.jpg (584KB, 1280x960px) Image search: [Google]

584KB, 1280x960px

AI, Machine Learning, and IR dummy

>>8109035

both, using xilinx tools is sadness and anger at the same time.

>>8109084

If you want to make something cool it has to be energy efficient and fast, both wrt rtt and throughput. FPGA tools are becoming better, as soon as programming FPGA code becomes effective enough there's gonna be some hype shit going in.

>>8107901
Y u no use yosis?