• What are you working on? v67 - March 2017
    3,527 replies, posted
I posted about how I accepted a job offer for $54,000 the other day. Well, I got a job offer for $65,000 with a non-profit (government funded). I talked to my parents, and my dad thinks I should weigh it based off my priorities, but my mom says go for it. Also, based off of what you guys said in my last post, it seems like I'm being underpaid for my current role. So, I've decided to renege on my offer and go with the $65,000 one. After all, I'm all about them Benjamins
65k in Richmond means you basically don't have to worry about money.
Well one of my goals is to send money home to my girlfriend's mom to help her with rent. She works like 7 days a week
What kind of taxes are you going to buy? Sounds like a nice chunk of money.
https://www.youtube.com/watch?v=TyHqjF35k4c Implemented bullet physics's 2d pipeline (CPU sadly) and integrated it with navier stokes. Unfortunately, all these gpu reads are killing performance. I need to bundle them up into one big read (ie submit a request for things to be compacted into a linear buffer and read back) or investigate bullets opencl rigidbody pipeline, which unfortunately seems to be extremely unsupported
I always factor like, 25% for taxes
I needed a very simplistic scripting language for my game so I made one https://i.imgur.com/8yWX4h6.png It's not advanced and the interpreter code is bad but it does what I need it to do and I can say I wrote a language so I don't care.
Do you have a complete instruction set? I love stuff like this
Alright. I gave in and looked into bullets OpenCL rigid body pipeline. To anyone who's ever wondered what the state of it is and whether or not its worth using, the answer is no: Its about as undocumented as you can possibly get, and the examples don't seem to compile. I had to dig through a lot of source code With the uniform grid solver, its fast but low velocity collisions between objects with a radius of more than 2 causes them to gain speed. I don't know why, but i thoroughly tested this. Objects with a radius less than this work fine With the SAP solver or parallel linear, performance is terrible for some reason (5ms/frame) Traditionally in bullet, static objects are objects with 0 mass. This causes really bad performance. This may be me using the api incorrectly The above is all tractable, but the real killer: perturbing an objects velocity in OpenCL seems to cause some sort of big ol recalculation to happen. This brings the performance to 5ms/frame ish with the grid solver which is unacceptable So overall its completely unusable without investing a lot of time into fixing it which i'm not willing to do. I may also be using the API incorrectly, its extremely hard to tell although I am relatively sure that i'm using it how the examples frame it This left the other option of async read bundling. The way this works is by abusing the crap out of callbacks. Basically: Queue an asynchronous write for a list of points you want to query on the gpu, and on its callback Queue a kernel to fetch the points you wrote to the gpu in step 1, and on its callback Queue an asynchronous read asynchronously to read back the points you got in 2, and set its callback to signal the main thread that data is available The other option that might seem appealing is to simply queue a write, use an event to queue a kernel and tie it to the write, and then queue a read and use an event to tie it to the kernel, and finally a callback. This is super bad as it creates a huge pipeline bubble with events waiting for each other, asynchronous callbacks work significantly better. The other thing of note is that for some reason performance is bad if you use a dedicated read queue, so for some reason using the main render/sim queue works best. Who knows Anyway I know you're all in this for the youtube videos https://www.youtube.com/watch?v=uoMr8P7tor0 Overall this appears to have 0 overhead, which is way *way* better than I was hoping for
What do you mean by complete? Here's a full list of the instructions and you can tell me if it's complete or not enum {     OP_NON=-1, // not an operation     OP_MDS,    // memory location     OP_MDX,    // second memory location     OP_IFE,    // if equal to     OP_IFL,    // if less than     OP_IFG,    // if greater than     OP_ELS,    // else     OP_ENF,    // end if     OP_SET,    // set mds     OP_ADD,    // add     OP_MUL,    // multiply     OP_DIV,    // divide     OP_NEG,    // negate     OP_PRN,    // print     OP_MEM,    // print memory (debug)     OP_DEF,    // define label     OP_JMP,    // jump to label };
I didn't mean "complete" in any sense other than "can I see all the instructions"
This isn't an emulator or anything as advanced as that, It's just a tiny interpreter for my game's scripting language. I enjoyed TIS-100 and shenzhen i/o and thought that style fit the game so that's the syntax I decided to go with.
man, those both look obscenely low to me since in silicon valley the median starting salary is like 100k or so for people just out of college with a bs in cs or software eng or w.e fucking housing costs more than balances it out though
Yeah like I said, a 1br in my area, on the high end, is $1,000/mo
Reverse engineering GTA V attempting to synchronize task over the network Working in project named Fivenet Multiplayer [https://streamable.com/q4woc] Some experiments found here: Fivenet Multiplayer / Archive / librg
Got some (internship) interviews coming up with some big players, a bit worried but more worried about my midterms. Hopefully it goes well (and they have corporate housing).
Is this all your own engine? This is some really impressive stuff!
First of all, classic WAYWO using a thread that's over a year old Second, I've been working on the art for MLS5 (play about with it here: My Little Starship 5) and adding gamey stuff, like music (muted by default), sound, animations, etc. It's a portfolio piece for non-gamedev developer job applications but I feel like presentation is important, right? So make it look nicey nice nigey nige. https://files.facepunch.com/forum/upload/133252/92fdd8c3-d8cb-4e41-9444-093c87c0946e/screen.png I also want to start ticking boxes on my portfolio, so it's got some basic unit testing stuff in there as well. I'm also running a google sheets document for other people to contribute content for story events, and I'm gonna put together a little JS applet with a web UI for parsing that into JSON and bringing it into the game. My main issue at the moment is that the event text can only be so long because the UI library I'm using doesn't have scrollbars. It's really nice other than that, though, would recommend: SlickUI on GitHub. (There's a fork of it that claims to have scrollbars but I'm not that au fait with grunt yet and that's needed to work on it.) I'm looking for projects to contribute little bits to so I can show off my github profile in interviews, so hmu if you have any tips or want a hand with some simple JS or C# work! Also all the code for MLS5 is open source on GitHub so if you fancy a poke about it's here: MLS5 on GitHub. Please don't fork it if you're going to do it better than me and make me look like a chump thx. Sorry for rambling anyway luv u all byee xx
I almost forgot the joy of learning a new language. Learning some basic powershell scripting for some small tasks at work and I am in that amazing phase where completely messy piles of text turn into something you can understand.
https://www.youtube.com/watch?v=7Wh0DZtlS48 I got bullet physics to interact nicely with the ground, and also fixed the order of rendering to work correctly (which was a huge pain, as OpenCL and OpenGL are really tediously refusing to work together correctly, where the physics objects in white are opengl/bullet, and everything else is rendered with opencl)
That certainly would be impressive! It is Unity though.
I decided I wanted an editor for my language so I wrote one https://cf-e2.streamablevideo.com/video/mp4/uf9pt.mp4?token=1518016434-Imv7eHzFS93IDW9Fbx%2FQQj%2BQXRW3n70RfsiAy0y1tOA%3D also some rudimentary functions are implemented
Microsoft interview went really well, wasn't too sure with the first interviewer but the second one liked me and I raised some things no one had earlier so hopefully I'll know soon.
Nawww it's still impressive. Doesn't take a you over a year to get where I've gotten, which is still building backend systems
SO that 65k was a quote from my recruiter but turned out not to be accurate. The job offer is for $60k, so now i have to choose between the $54k that I already accepted or the $60k in front of me. Apparently the $60k job has excellent benefits as well (federal contractor)
Probably Redmond. There's corporate housing right? I mostly applied due to the corporate housing.
I once accepted a paid internship over an unpaid internship and regretted it a lot. Yesterday I accepted a game development job for $30k. Compare which one would make you the most happy to do and which one could give you the better opportunities instead.
This is pretty cool, for you pixel art on mobile problem you could keep all the elements the same size and just move them further apart perhaps. It would be cool too if you made story lines, so when it generates a new event there is a chance it "starts" a new story line then on subsequent event fires it would have a chance of playing out the next event in that story. Simple concept for a game but it's cool and I sorta want more. Will you be adding more or it gonna remain a bare bones portfolio piece?
I believe so. I never interned or worked for them, but my friends did/do and they found housing just fine. The closer you live to the campus, the better.
yeah it's corporate housing that's fully furnished. if you have any Q's let me know -- my 2 close friends (1 of them is still there) and myself interned at MS a few years ago. what team was it for?
Sorry, you need to Log In to post a reply to this thread.