Looks perfect for VR porn

Doing some machine learning interests.

While I can get behind some Rocket League in this case I meant reinforcement learning specifically as that will be a large chunk of the research / experimentation / but mostly education and I just wanted a nice machine to be able to play as much locally at a decent clip.

It has been a long time since I really messed with PCs and Linux (been mainly on macbook pro laptops for a number of years) and I have to say it has been enjoyable throwing it together and mucking about getting everything ready both on the hardware and software stack for what I am planning on working on.

Original plan was to go all-out and have 4 Titan X Pascals and really just have an absurd machine that matched (and out-specced) the nvidia DIGITS devbox for about 65% the cost they were asking (~15k)

But I started to be more realistic as to how it is going to be utilized, and a lot of the uncertainty of what the nut machine for doing dl work locally would even be 12-18 months from now, with companies like AMD even just the other day announcing their new chips, Intel acquiring Nervana in Aug (brilliant), TPU's potentially lingering out there, other custom hardware, etc.

So I opted to go with just a single Titan-X for the time being which allowed me to save money on the cpu (i7 6800k) and motherboard (asrock taichi).

Looking forward to the new Samsung 960's to ship and the gpu to arrive.

why not rent compute time on a cloud service?

EDIT: nm, that would probably get crazy expensive at 128G ram

I think cloud is probably still cheaper. You can usually get a machine with at least those specs for < $0.50/hour (spot prices on AWS) and you don't need to run it all the time (especially if you can do a lot of development and early testing on a smaller/cheaper machine).

But the savings are probably minimal at best, and it wouldn't be as fun or cool.

Yeah I looked into various cloud solutions, and played around a bit with AWS just to learn how to do it.

The pricing and power are not too bad, but for what I plan to be messing with it just seemed preferable to have my own machine.

Also the other benefits of just having a full base machine for person use such as archiving and managing old information from old computers through the years, and family stuff is nice.

Plus the enjoyment from being an old timey computer nerd at heart.

Full build specs for anyone interested or that may be planning to build a similar machine:



Prices are off on a number of things, and it seems like there are always nice deals popping up for various components. Some nice Black Friday deals made me go ahead and jump on some things when originally I had been planning to build in January.

Very nice!

I'm looking forward to seeing AMD's new stuff next year, it's been a long time coming.

Thanks dave! Yeah I am looking forward to seeing how they end up pricing things and just how aggressive they end up going.

Also really interested to see what Intel has in store with Nervana, and if there is any chance of google releasing their TPU to the public in any fashion.

I think we are going to see a lot of interesting things in the hardware space over the coming 2 years directly in regards to neural nets and ML.

anyone doing adventofcode.com? the times this year are crazy fast. i did day 16 in 15m and still just missed the leaderboard.

also been learning J, which i find to be a mind-bendingly beautiful language. my solution for day 16, keeping in mind this is still probably too long cause i'm just a newb:

I'm confused, is that the "beautiful" part or did you submit that to an obfuscated code contest?

Seeking General Career Advice:
Some context, I've been employed as a GIS Analyst for almost 6 years. Found some tasks in my day to day that I could automate using Python and that kind of lead me to getting interested in web development. I decided to enroll in an online bootcamp and graduated in early October and have been on the job hunt since.

I'm now sitting with two job offers.

Position Uno - Front End Developer working in a small company (~15 people) that has it's business split between IT services and custom website development. Work environment seems to be very laid back and seems comfortable to me. (Note: CEO has a placard on his desk that says "I am a ****ing professional", which I think is hilarious.) All of the clients are local and either small businesses or town/city gov't with real estate companies their biggest focus. I'd mostly be working by myself on the front end building out webpages with HTML, CSS and JS working along with PHP devs using Laravel.

Position Dos - Ruby on Rails Developer also working in a small company (~10 people) which is a holding company for several other ventures (not quite sure how large each of these other companies are). Also a very comfortable environment with everyone I saw being younger, I'm 32. Most of my work here would consist of building web scrapers which get public records data (criminal and traffic charges), munging that data, and selling direct to attorneys in the form of direct mailers to these alleged criminals. They have some other businesses, but description above seems to be a lot of my tasks.

--

I'm not very good on the front end, so I like that with Uno I'd be spending time learning all the intricacies of various browsers which I don't understand yet and also generally improving my JavaScript skills while getting paid.

With Dos, the fun/learning part would be interacting with all these websites with web scrapers and figuring out how to make them work correctly. Also, all their servers are on AWS so I'd end up picking up a lot of skills there as well.

Assuming compensation is similar, which of these positions seems like a better first developer position? Meaning, which skills and work will translate the best to any position I take from here on out. I kind of have a general lean, but I'd welcome any thoughts. Thanks for reading this tl;dr post.

Im kind of biased against general front end web dev, but even with that I still lean #2. It reads like better professional experience through and through both with Ruby experience and the actual work being a bit deeper than making a web page look pretty.

IOW, it feels like #1 could be learned a lot more on your own time and #2 reads like more professional experience which is both more interesting and builds your resume better (working with data, using APIs and scraping to collect data, etc all very important things right now in the industry)

Also, some exp working with local clients web pages and its generally "Hey make this button fit better on mobile" and other minutia

I think the biggest downside to #1 is that you'll be working solo. Ideally you want plenty of opportunity to learn from more experienced folks.

Thanks for weighting in here. I was already learning towards 2 just because I enjoy Ruby and feel like it'd be a slightly better fit. I think you make a real good point that I could learn the browser stuff on the side if I'm really interested in it. It's unlikely that I'll just start making web scrapers on the side unless I come up with a good idea.


Thanks for reading my post. That's a good point that I didn't fully consider. #1 I assume I'd be leaning on the PHP back end guys for some help, but those 2 probably aren't that great at front end so it'd be mostly flying solo. For #2, they do have contract employees which have built some of their current software and it does sound like I would be able to lean on these folks for learning more best practices regarding scraping, etc.

Go with 2.

At my last company we scraped oil and gas data from state and pipeline websites. Writing scrapers can be soul crushing. Every time you think everything is good somebody will change their site around and you have to go back and fix. Some sites are downright evil in trying not to get scraped. One pipeline company would routinely change the column order on some of their tables, ****ing up a week or two of data before we caught it.

But it was rewarding solving the puzzle of getting around all the **** they threw up there trying to prevent the scraping. It would have been a lot less work for them to just provide a nice zip file of the data and throw it up on a server somewhere, but no, lets spend all of our time making it hard for people to scrape.

You know what's more soul crushing than scraping? Building a cheap ass website for some schmuck who will give you directions like "needs to be more fun" or "it needs to be more vibrant" or "needs more animated gifs"

http://theoatmeal.com/comics/design_hell

The Agile process in action.

Agile Manifesto

Quote:

Our highest priority is to satisfy the customer
through early and continuous delivery
of valuable software.

Welcome changing requirements, even late in
development. Agile processes harness change for
the customer's competitive advantage.

Business people and developers must work
together daily throughout the project.

The most efficient and effective method of
conveying information to and within a development
team is face-to-face conversation.

What could go wrong?

Agile is still better than trying to define every requirement and design the perfect system up front. Design your system to be flexible to new requirements, which ALWAYS come up in large projects, and you have a much better chance of success. Big reason why I like component-based over framework-based, and composition over inheritance.

That oatmeal comic doesn't show that changing requirements is bad. It shows that having clueless people change your requirements is bad.

We're getting a box at work with 24 terabytes of RAM, obviously not for gaming!

Details?

+1

A single machine and not some kind of cluster? I'm pretty curious about that