At 11/11/13 11:12 PM, Innermike wrote: Just nitpicking nonsense, anyway @egg82 since you have such a beefy machine how does this run for you? http://www.simppa.fi/experimental/as3/2011/molehillparticle/1_4_million/

Not fullscreened, I can do 15-20ish fps.

At 11/11/13 11:12 PM, Innermike wrote: Just nitpicking nonsense, anyway @egg82 since you have such a beefy machine how does this run for you? http://www.simppa.fi/experimental/as3/2011/molehillparticle/1_4_million/

With recording, 59-60 (I had to use FRAPS, so the framerate is really HDD-dependent)
Without, of course, never below 60. Fullscreen all the way.
I have a gaming rig that's meant to play and record things like BF4. You aren't going to kill it with Flash :P

Anyway, yeah. If you want to keep throwing challenges my way i'm completely okay with that, I kinda want to see what I can handle and what I can't. If you wan "optimization/how's this" advice, I think PS is your guy to talk to.

Three cheers for YT atrifacting and copyright system, eh?

So after watching the entire 1.4 million particle thing, I really wanted to record it for you guys so you could watch it (it really is beautiful) - so far the best FPS i'm getting is from FRAPS in FireFox, which records at 60 but dropped as low as 30 during one bit of the recording. I'm experimenting with some different stuff, but yeah. It's really nice and you should see it in its full 60 FPS glory :3

Here's a quick benchmark while not recording, if anyone's interested. Probably not, but I did it anyway :P
2013-11-12 14:40:50 - FlashPlayerDebugger
Frames: 22485 - Time: 374746ms - Avg: 60.001 - Min: 59 - Max: 61

At 11/12/13 07:03 PM, Innermike wrote: Well I couldn't tell a difference in framerate between your recording and running it (fullscreen) on my low end macbook air so I think PS may be the only guy to really benefit, but yeah upload the full thing to your YT anyway, it's pretty cute.

hmm, I suppose I can see why. Anyway, yeah. I think I figured out how to record it at true settings with true quality, I just gotta make sure the final mp4 (after render) isn't artifacted, even though YT will artifact it anyway. Don't want to compound the problem.

Sidenote, since when does youtube officially support 60fps? I had friends that used to upload gameplay and they said youtube forcibly cuts it down to 30

29.97, actually. And probably never, though usually you can still tell the different between a 60 FPS recording and a 30 FPS recording, even when the final video is cut down.

Recorded, edited, encoded, and uploaded.
Framrate for the most part was anywhere between 57-60, though a few times during intense sections it dropped to about 45. Once it dropped to as low as 35 :(
Encoded it with blu-ray quality settings and compressed it down into an mp4.
Uploaded to both YouTube and MediaFire because YouTube re-encodes videos with really poor-quality settings.
Original file is on MediaFire, I seriously recommend watching it there. The YT video looks awful in comparison.

Enjoy?

YouTube: https://www.youtube.com/watch?v=1GPdvy-NlC4
MediaFire: http://www.mediafire.com/watch/114dffkprmw7jt8/Particles.mp4

Just checking in. Been very busy with full time job. I really miss having time to devote to passion projects. I do have a side project going on, and it's pretty neat; but nothing really more to say about it right now. I am also pumping some effort into my Javascript Pathfinding library EasyStarJS: https://github.com/prettymuchbryce/easystarjs. I am redoing the website for it with node.js/express/foundation/pixi.js. I am also experimenting with implementing another algorithm called Jump Point Search into the library. Otherwise last night I downloaded Unity 4.3, and played around with the new 2D stuff a little bit. It's very clever and fun to use. The thing that is great about Unity is just how fast you can get going and making content. It reminds me of building games in flash with the IDE. Super fun.

I've been playing RimWorld, a super fun strategy-type game. You should check it out. I also played rogue legacy, but I felt it was a little too grindy for my tastes. Yeah.

At 11/13/13 10:10 PM, PrettyMuchBryce wrote: I am also pumping some effort into my Javascript Pathfinding library EasyStarJS: https://github.com/prettymuchbryce/easystarjs.

Looks pretty neat. I might add it to my thing if I ever end up using a* for something other than rpg movement. On average, nothing is going on besides rendering, so I can get away with unoptimized grid searches.

The full story of MSGHero for throwback Thursday.

Java GUI development is the shittest thing I've ever had to do.

Java development is the shittest thing I've ever had to do.

I come with a challenge.

You have to convert one word to another, changing only one letter at a time. Every step between must be a real word.

For example cat -> dog:

cat
cot
dot
dog

You must return the shortest possible path, or detect if there is no path. You must do this in the quickest time (shortest number of branches) possible.

Test word for you: short -> bread

No googling for solutions

No one will try this, but if anyone is bored then you have no excuse to be :)

At 11/14/13 03:53 PM, Rustygames wrote: I come with a challenge.

You have to convert one word to another, changing only one letter at a time. Every step between must be a real word.

For example cat -> dog:

cat
cot
dot
dog

You must return the shortest possible path, or detect if there is no path. You must do this in the quickest time (shortest number of branches) possible.

Test word for you: short -> bread

No googling for solutions

No one will try this, but if anyone is bored then you have no excuse to be :)

Generate a graph from a dictionary (link all words which differ by 1 letter to each other).
Run Dijkstra's algorithm on it.

i dont feel like coding it but that works

At 11/14/13 03:57 PM, Glaiel-Gamer wrote: Generate a graph from a dictionary (link all words which differ by 1 letter to each other).
Run Dijkstra's algorithm on it.

i dont feel like coding it but that works

I like it. Why not A* though?

At 11/14/13 02:12 PM, MSGhero wrote: The full story of MSGHero for throwback Thursday.

Bards of the future will sing this story, provided that the future brings back bards.

At 11/14/13 03:53 PM, Rustygames wrote: I come with a challenge.

The lazy way to do it is to calculate the Levenshtein distance, keeping track of the changes made. I'm too lazy to program even the laziest of solutions though.

At 11/14/13 05:40 PM, Archawn wrote:

At 11/14/13 03:53 PM, Rustygames wrote: I come with a challenge.

The lazy way to do it is to calculate the Levenshtein distance, keeping track of the changes made. I'm too lazy to program even the laziest of solutions though.

Oh, silly me, they need to be words in between. Glaiel had it right then. Go Glaiel

At 11/14/13 04:34 PM, Rustygames wrote: I like it. Why not A* though?

Cool problem. A* would be faster. Use the current difference in letters from the target as your heuristic.

Here is one. Given two lists of unsorted unique integers, find the nth lowest of both in the lowest time complexity. No google.

Hint: Solution can be achieved in linear time

Example
[0, 10, 2, 5]
[4, 8, 1, 7]

3rd lowest is 2

At 11/14/13 03:53 PM, Rustygames wrote: I come with a challenge.

I remember this challenge from a book of logic puzzles I had (I had quite a few as a kid)

Test word for you: short -> bread

short
sort
sot
bot
boat
bat
brat
brad
bread

At 11/14/13 04:34 PM, Rustygames wrote:

At 11/14/13 03:57 PM, Glaiel-Gamer wrote: Generate a graph from a dictionary (link all words which differ by 1 letter to each other).
Run Dijkstra's algorithm on it.

i dont feel like coding it but that works

I like it. Why not A* though?

dijkstra's easier to implement and you can do it with a queue instead of a priority queue. optimize later if its actually slow (probably wont be slow)

At 11/14/13 07:03 PM, PrettyMuchBryce wrote:

At 11/14/13 04:34 PM, Rustygames wrote: I like it. Why not A* though?

Cool problem. A* would be faster. Use the current difference in letters from the target as your heuristic.

Here is one. Given two lists of unsorted unique integers, find the nth lowest of both in the lowest time complexity. No google.

Hint: Solution can be achieved in linear time

Example
[0, 10, 2, 5]
[4, 8, 1, 7]

3rd lowest is 2

There was some weird sort we talked about in java where you place numbers into "buckets" based on the last digit, then based on the 2nd to last, etc. 10 buckets for 10 digits it can end with, 10 buckets in each of those for 10 digits that can be in the tens spot, etc. It was linearish I think, maybe some coefficient based on the max length of a number. Then iterate through that thing to find the nth.

maybe

At 11/14/13 08:33 PM, MSGhero wrote: There was some weird sort we talked about in java where you place numbers into "buckets" based on the last digit, then based on the 2nd to last, etc. 10 buckets for 10 digits it can end with, 10 buckets in each of those for 10 digits that can be in the tens spot, etc. It was linearish I think, maybe some coefficient based on the max length of a number. Then iterate through that thing to find the nth.

maybe

That sounds like a hash table with a very basic hash. Also sounds like there'd be a lot of memory overhead with potentially a lot of partially filled buckets.

or maybe I'm just going mad with this small amount of sleep I've had

At 11/14/13 08:33 PM, MSGhero wrote:

At 11/14/13 07:03 PM, PrettyMuchBryce wrote:

At 11/14/13 04:34 PM, Rustygames wrote: I like it. Why not A* though?

Cool problem. A* would be faster. Use the current difference in letters from the target as your heuristic.

Here is one. Given two lists of unsorted unique integers, find the nth lowest of both in the lowest time complexity. No google.

Hint: Solution can be achieved in linear time

Example
[0, 10, 2, 5]
[4, 8, 1, 7]

3rd lowest is 2

There was some weird sort we talked about in java where you place numbers into "buckets" based on the last digit, then based on the 2nd to last, etc. 10 buckets for 10 digits it can end with, 10 buckets in each of those for 10 digits that can be in the tens spot, etc. It was linearish I think, maybe some coefficient based on the max length of a number. Then iterate through that thing to find the nth.

maybe

Radix sort. In any practical situation I would say you're right. This is a good solution. But just for fun what if each integer has a unique key length (number of digits) ? Maybe they are 64 bit floats instead of ints with many decimal places. We then have an average case of at least linearithmic time. How we do achieve an average case of linear time ?

I will post the solution I'm thinking of tomorrow if nobody gets it. I'll give you a hint. It involves the partition step of quicksort.

At 11/14/13 11:32 PM, PrettyMuchBryce wrote: Radix sort. In any practical situation I would say you're right.

Stopped reading after here.

At 11/14/13 11:34 PM, MSGhero wrote:

At 11/14/13 11:32 PM, PrettyMuchBryce wrote: Radix sort. In any practical situation I would say you're right.

Stopped reading after here.

Typically these sorts of problems have very little practical application. That is not really the purpose. I mean.. when do you need find the shortest list of words by single letter permutation of "short" to"bread" ?

At 11/14/13 11:52 PM, PrettyMuchBryce wrote: Typically these sorts of problems have very little practical application. That is not really the purpose. I mean.. when do you need find the shortest list of words by single letter permutation of "short" to"bread" ?

In a terrible Saw movie.

At 11/14/13 07:26 PM, egg82 wrote:

At 11/14/13 03:53 PM, Rustygames wrote: I come with a challenge.

I remember this challenge from a book of logic puzzles I had (I had quite a few as a kid)

Test word for you: short -> bread

short
sort
sot
bot
boat
bat
brat
brad
bread

I should have mentioned you aren't allowed to take away or add letters, just change them

At 11/14/13 07:34 PM, Glaiel-Gamer wrote:

At 11/14/13 04:34 PM, Rustygames wrote:

At 11/14/13 03:57 PM, Glaiel-Gamer wrote: Generate a graph from a dictionary (link all words which differ by 1 letter to each other).
Run Dijkstra's algorithm on it.

i dont feel like coding it but that works

I like it. Why not A* though?

dijkstra's easier to implement and you can do it with a queue instead of a priority queue. optimize later if its actually slow (probably wont be slow)

The difference in implementation between Dijkstra's and A* is tiny. I suspect some googling may have gone on here ;)

At 11/14/13 11:52 PM, PrettyMuchBryce wrote:

At 11/14/13 11:34 PM, MSGhero wrote:

At 11/14/13 11:32 PM, PrettyMuchBryce wrote: Radix sort. In any practical situation I would say you're right.

Stopped reading after here.

Typically these sorts of problems have very little practical application. That is not really the purpose. I mean.. when do you need find the shortest list of words by single letter permutation of "short" to"bread" ?

Was an interview question here and Dijkstra's was given as the best answer. I disputed this by saying A* was far superior and proved as much. I was wondering if anyone here fell into the same trap :)

At 11/15/13 05:47 AM, Rustygames wrote: I should have mentioned you aren't allowed to take away or add letters, just change them

Proper nouns?

At 11/15/13 12:21 PM, egg82 wrote:

At 11/15/13 05:47 AM, Rustygames wrote: I should have mentioned you aren't allowed to take away or add letters, just change them

Proper nouns?

Any dictionary you find and use I will accept :)

Once again I'm left with the apt to myself for the weekend. Sucks that there isn't a game jam going on, no hw and no tests to interrupt either. We have a big LAN party going on tomorrow and Sunday, so I might go there to play some dota. I participated in the tournament last year, but I was put in a random team, and they weren't that good. So I'll just play solo or watch the tournament games.

At 11/15/13 04:02 PM, Innermike wrote: And so your sig is fulfilled yet again. Speaking of sigs, I need to start a catalog of all my potential MSG post sigs.

One guy was just at a meeting, but he leaves tomorrow morning.

at least he didn't walk in on me singing naked in the lounge

At 11/15/13 07:48 PM, MSGhero wrote:

at least he didn't walk in on me singing naked in the lounge

Pretty sure he'd be used to it at that point :P