[IMG]http://i.imgur.com/ivEnu.png[/IMG]
Fully-fledged attack system.
Well, only attack works for now.
Messy code here:
[url]http://pastebin.com/YgsW6ByT[/url]
[QUOTE=Kepler;33938710][IMG]http://i.imgur.com/ivEnu.png[/IMG]
Fully-fledged attack system.
Well, only attack works for now.
Messy code here:
[url]http://pastebin.com/YgsW6ByT[/url][/QUOTE]
Fight :: Pokemon
Bag :: Run
[QUOTE=robmaister12;33938633][url]http://tirania.org/blog/archive/2008/Nov-03.html[/url]
[editline]28th December 2011[/editline]
there's a nice chart at the bottom of the post comparing speeds of a direct physics simulation written in C++, ported directly to C# and compiled with the new Mono.SIMD optimizations.[/QUOTE]
Besides the other post here stating that the test was flawed:
Did they use SIMD operations in the C++ version? Because there's a huge difference between not using SIMD and using SIMD, so it's not fair to compare a C# application using SIMD to a C++ one not. It's a hardware thing at that point.
[QUOTE=Chris220;33938821]ftfy[/QUOTE]
whatever
[editline]28th December 2011[/editline]
depends on the version etc. etc.
red generation is best generation
[editline]edit[/editline]
I can't really "show" this off yet, but I figured people might want to use lopensteamworks for fast osw goodness, so I'm going to clean up the code some more and push it to Google Code.
Jookia do you really have to get involved in every argument here?
[QUOTE=BlkDucky;33938896]whatever
[editline]28th December 2011[/editline]
depends on the version etc. etc.[/QUOTE]
As far as I can remember, none of the versions have it with those words in that order :v:
[QUOTE=Lord Ned;33938857]Besides the other post here stating that the test was flawed:
Did they use SIMD operations in the C++ version? Because there's a huge difference between not using SIMD and using SIMD, so it's not fair to compare a C# application using SIMD to a C++ one not. It's a hardware thing at that point.[/QUOTE]
yeah, that would be the link that artanis posted. But that still includes the time it takes for the JIT compilation. I'm assuming that if you ran these benchmarks for several minutes as opposed to .03 seconds the results would me much closer than the C++ version taking 5 minutes and the C# version taking 4 hours and 10 minutes.
[editline]28th December 2011[/editline]
I'm looking at the source code of this benchmark... he's using DateTime.Now, which is only precise to 10ms: [url]http://msdn.microsoft.com/en-us/library/system.datetime.now.aspx[/url]
I was expecting System.Diagnostics.Stopwatch...
Procedural graphics generated with Fibonacci;
[img]http://i51.tinypic.com/xfx1fd.png[/img]
[QUOTE=robmaister12;33939400]yeah, that would be the link that artanis posted. But that still includes the time it takes for the JIT compilation. I'm assuming that if you ran these benchmarks for several minutes as opposed to .03 seconds the results would me much closer than the C++ version taking 5 minutes and the C# version taking 4 hours and 10 minutes.[/QUOTE]
Actually, I don't think that guy changed anything other than compiling with an optimization flag on the command line. However, I didn't like the fact that that source was on windows, vs deIcaza's linux test, so I ran the test for myself on a Gentoo box. Here is the modified makefile that I used: [url]http://pastebin.com/4BiS45ku[/url]
Here are the results:
(linear...spring)
g++ no optimization
2.021s...45.496s
g++ -O2
0.039s...0.039s
Mono simd
1.401s...2.387s
Mono no optimization
3.309s...57.390s
It appears that g++ can perform simd optimization just as well (better) than the mono compiler with no modification to the c++ application.
Benchmark comparisons! The most useless shit in the world.
[QUOTE=artanis;33939543]Actually, I don't think that guy changed anything other than compiling with an optimization flag on the command line. However, I didn't like the fact that that source was on windows, vs deIcaza's linux test, so I ran the test for myself on a Gentoo box. Here is the modified makefile that I used: [url]http://pastebin.com/4BiS45ku[/url]
Here are the results:
(linear...spring)
g++ no optimization
2.021s...45.496s
g++ -O2
0.039s...0.039s
Mono simd
1.401s...2.387s
Mono no optimization
3.309s...57.390s
It appears that g++ can perform simd optimization just as well (better) than the mono compiler with no modification to the c++ application.[/QUOTE]
The difference is that sometimes you have to compile an app to run on platforms that don't support SIMD as well but still take advantage of SIMD on supported platforms. A JIT compiler does this at runtime whereas an compiled-ahead binary has to check whether or not said technologies are supported and deal differently if they aren't.
So my Mac has kernel panic'd twice today. This isn't good at all, but I am working on a project and I need to continue working.
Every time I reboot my fucking Mac, apple make me log into the Documentation. I don't remember ever having to log into the documentation. And I use LastPass to manage my passwords, so I don't know my Apple Developer password. I have access to it, but it is a 20 letter alphanumeric string which I dont feel like memorizing. Fuck you, Apple.
[img]http://i.imgur.com/uKx8X.png[/img]
Does anyone know a workaround to this or something???
[QUOTE=Dotmister;33939122]Jookia do you really have to get involved in every argument here?[/QUOTE]
I'm a very lonely person with nothing else to do and I'm pretty sure if I stopped programming I'd hurt myself.
Anyway, If you didn't notice, I stepped out of that argument.
[QUOTE=i300;33939634]So my Mac has kernel panic'd twice today. This isn't good at all, but I am working on a project and I need to continue working.
Every time I reboot my fucking Mac, apple make me log into the Documentation. I don't remember ever having to log into the documentation. And I use LastPass to manage my passwords, so I don't know my Apple Developer password. I have access to it, but it is a 20 letter alphanumeric string which I dont feel like memorizing. Fuck you, Apple.
[img]http://i.imgur.com/uKx8X.png[/img]
Does anyone know a workaround to this or something???[/QUOTE]
I heard Win7 doesn't have kernel issues
[QUOTE=high;33939600]Benchmark comparisons! The most useless shit in the world.[/QUOTE]
thisthisthisthis.
Even if one language is marginally faster in one particular benchmark, it's completely meaningless in real scenarios even assuming the difference is any more than negligible, which it won't be.
I dunno man, even with your gigahertz processors and gigabytes of RAM and terabytes of hard disk space I still think we need to make sure things run faster.
[QUOTE=DangerStranger;33939648]I heard Win7 doesn't have kernel issues[/QUOTE]
I have two computers: A Windows 7 Desktop Machine and a Mac Laptop.
I have a Mac Laptop because, for the most part, they last and are very stable. I also happen to develop iOS Apps and Mac Software, which you can [I][B]only do on a Mac[/B][/I].
The project I am working on also happens to be a iOS App. Trust me, if I could make iOS and Mac software on my Windows Machine, I would. But I can't, so I'm stuck on my Mac right now.
[editline]a[/editline]
Found a solution. In XCode, go to XCode > Preferences > Downloads > Documentation and click Check and Install Now. It will make all the documentation available offline!
[img]http://i.imgur.com/CEaOa.png[/img]
[QUOTE=BlkDucky;33939654]thisthisthisthis.
Even if one language is marginally faster in one particular benchmark, it's completely meaningless in real scenarios even assuming the difference is any more than negligible, which it won't be.[/QUOTE]
I do consider a 60 times speedup significant. This particular benchmark [I]is[/I] a realworld scenario of doing math against vectors for a physics simulation (and not breaking your brain to write the code that does it). Benchmarks aren't worthless just because they are benchmarks. Real world scenarios can be benchmarked.
[QUOTE=i300;33939634]I use LastPass to manage my passwords[/QUOTE]
Open LastPass and copy your password?
[QUOTE=CDeansy;33939846]Open LastPass and copy your password?[/QUOTE]
They have paste disabled on the dialog. Who the fuck does that, especially for passwords?
[QUOTE=artanis;33939737]I do consider a 60 times speedup significant. This particular benchmark [I]is[/I] a realworld scenario of doing math against vectors for a physics simulation (and not breaking your brain to write the code that does it). Benchmarks aren't worthless just because they are benchmarks. Real world scenarios can be benchmarked.[/QUOTE]
But they can't be [B]generalised[/B]
[QUOTE=artanis;33939543]Actually, I don't think that guy changed anything other than compiling with an optimization flag on the command line. However, I didn't like the fact that that source was on windows, vs deIcaza's linux test, so I ran the test for myself on a Gentoo box. Here is the modified makefile that I used: [url]http://pastebin.com/4BiS45ku[/url]
Here are the results:
(linear...spring)
g++ no optimization
2.021s...45.496s
g++ -O2
0.039s...0.039s
Mono simd
1.401s...2.387s
Mono no optimization
3.309s...57.390s
It appears that g++ can perform simd optimization just as well (better) than the mono compiler with no modification to the c++ application.[/QUOTE]
I'm not saying that he modified any part of the C# source code, JIT compilation is part of the process of running C# code, and it takes time. That time is added to the benchmark time. If you want the same statically compiled optimizations, you'll want to AOT compile the C# code: [url]http://www.mono-project.com/AOT[/url]
So, I'm thinking that I should probably move on to trying git as like a more full-time dealio, and use github because it's the cool thing to do apparently. Plus, broadening my horizons, etc, etc.
I'm gonna use TortoiseGIT because I'm a git (teehee) who can't be bothered to do things The Right Way[sup](TM)[/sup] on Linux. Also considering, afaik, there isn't a good Windows solution for git yet.
Is there anything I should know specifically to get my head around git that would be meaningful to someone who's worked with subversion for years?
I used git a little bit once, and it didn't seem too hard at all. There's something about public-key jazz that's a little new to me, but besides that, the whole pushing factor doesn't throw me off as being radically different to a subversion commit.
I have begun to follow the philosophy that if I can run a program well on my laptop, anyone who doesn't have a beige computer should be able to run it.
[QUOTE=artanis;33939543]Here are the results:
(linear...spring)
g++ no optimization
2.021s...45.496s
g++ -O2
0.039s...0.039s
Mono simd
1.401s...2.387s
Mono no optimization
3.309s...57.390s[/QUOTE]
Sounds like you're comparing dick sizes. It doesn't matter how fast/large your language is; it's how you use it that counts.
I can see a C# program outperforming a poorly written C++ program and vice versa.
[QUOTE=Naelstrom;33940832]Sounds like you're comparing dick sizes. It doesn't matter how fast/large your language is; it's how you use it that counts.[/QUOTE]
TIL programming languages are basically peni
[QUOTE=amcfaggot;33940506]So, I'm thinking that I should probably move on to trying git as like a more full-time dealio, and use github because it's the cool thing to do apparently. Plus, broadening my horizons, etc, etc.
I'm gonna use TortoiseGIT because I'm a git (teehee) who can't be bothered to do things The Right Way[sup](TM)[/sup] on Linux. Also considering, afaik, there isn't a good Windows solution for git yet.
Is there anything I should know specifically to get my head around git that would be meaningful to someone who's worked with subversion for years?
I used git a little bit once, and it didn't seem too hard at all. There's something about public-key jazz that's a little new to me, but besides that, the whole pushing factor doesn't throw me off as being radically different to a subversion commit.[/QUOTE]
If you want to hear about some of the reasons it's different (better) than SVN, watch Linus Torvalds' talk he did at google a few years ago. Managed to convert me and never looked back since.
[media]http://www.youtube.com/watch?v=4XpnKHJAok8[/media]
If you are working on stuff alone, it might make sense to stick with SVN for now. But when you are working on stuff collaboratively, it makes the process of branching, forking and merging much much easier. You can set up chains of repositories. Infact, I made a post about it in the HOLLY forum yesterday -
[quote="aarong11"]Just as an example, i've set up a repository on GitHub at [url]https://github.com/aarong11/holly_public[/url]. I've directly imported the SVN including all the previous history and stuff, and people are free to make new branches (either local or central). When you make some changes, you are welcome to make a pull request, and if the request doesn't break something in the central repository then it can be merged into the main tree. If your branch has conflicts with the main one, then it is either up to the main branch owner or individual tree owner to fix these and then perform a new pull request.
This is all much easier when you are using just GIT rather than pulling the main branch from an existing SVN. Say for example you wanted to work with a friend in order to add a new codec, you could have the following setup -
[img]http://i.imgur.com/zVPoz.png[/img]
1. Bob, Fred and Dave decide they want to work on a new codec, but each working on smaller individual parts of it.
2. They set up a remote repository on GitHub that forks from Garry's Branch called "Codec 1"
3. They then each set up a local Git repository on their own machine and every time they add a new feature they make a pull request to the "Codec 1" repository.
4. If there are any conflicts, it is up to the individual users to fix them.
5. Once the new codec is implemented, a pull request is made to Garry's Branch. If there are no conflicts, Garry is able to merge the code with the main branch, otherwise he sends it back to the team to fix.
This might seem pretty complicated, but it's fairly trivial to set up using GitHub and makes collaborative open source projects 100x easier to manage rather than having to rely on patches and the like.
[/quote]
[QUOTE=amcfaggot;33940506]So, I'm thinking that I should probably move on to trying git as like a more full-time dealio, and use github because it's the cool thing to do apparently. Plus, broadening my horizons, etc, etc.
I'm gonna use TortoiseGIT because I'm a git (teehee) who can't be bothered to do things The Right Way[sup](TM)[/sup] on Linux. Also considering, afaik, there isn't a good Windows solution for git yet.
Is there anything I should know specifically to get my head around git that would be meaningful to someone who's worked with subversion for years?
I used git a little bit once, and it didn't seem too hard at all. There's something about public-key jazz that's a little new to me, but besides that, the whole pushing factor doesn't throw me off as being radically different to a subversion commit.[/QUOTE]
The big difference between SVN and git is that SVN tracks files and git tracks commits. Branching in git is extremely cheap (no visible difference in repo size) because a branch just points to a specific commit identifier (a SHA1 checksum of the changes to be exact, metadata like commit messages and the checksum of the previous commit are also stored).
When you commit in git, you are generating a new identifier, and your branch is updated to point to the new commit. There's also this fancy thing called fast-forward merging. Let's say you want to try out a new feature, but don't want to touch the stable code. So you create a new branch and start committing a bunch of changes. When it's all good and working, you checkout the main branch again, and merge with the experimental branch. Because there were no changes that the stable branch had that the experimental one didn't, git determines that the merge is a fast forward, and simply points the stable branch to the commit that you want to merge with.
Also you CAN use git in a centralized way by just having a server with SSH access and a git repo identical to the one on your computer. The public key stuff is just for creating a secure SSH connection, not specific to git. You can communicate between repositories (even if they're just two folders on your computer) by adding "remotes" to a repo. A remote is just the URL of another repository. When you push to a remote repo, you're sending the server all the commits that it doesn't have for your branch. When you fetch (pull) from a remote repo, you're sending the server the SHA1 checksum of the commit each branch is attached to, and if it doesn't match with the server, the server will send over all of the branch's new commits based on the checksum you sent it.
If that didn't make too much sense to you, I'm sure this lovely guide with pictures will: [url]http://progit.org/book/ch1-3.html[/url]
Sorry, you need to Log In to post a reply to this thread.
