Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Jiub]

I stared at the code for a few minutes and tinkered with it a couple more, so here's some conclusions I came to. Firstly, there's a little problem with naming. At the begin you create the two lists, "curGen" & "nexGen", which is ok, no problem. But then you also use these names as arguments for all the methods. I haven't experimented with/am not experienced enough to say for sure which one the method uses. An example of what I mean, you can do this:

value = 10
def add_value(number):
    return number + value

And it does it fine. What your code is doing, is using the same name twice (in the example, replace "number" with "value"), which obviously does something unwanted/unexpected.

So what I ended up doing to make the code run fine was:

1. Change all the argument names to something a little different.
2. Create the empty list "nexGen" inside, at the beginning of step() (no need for it to be global)
3. Change line 64 to "curGen = step(curGen)"

And that should pretty much do it.

[da_nang]

Python doesn't distinguish between formal and actual parameters?

[Vattic]

Hurrah it works! Thank you Jiub.

I still wonder why it was working fine with the list swapping line, and I thought it could handle the way I named things.

Thanks again.

[Orange Wizard]

Python doesn't distinguish between formal and actual parameters?

Python doesn't distinguish between a lot of things, like constants and... not constants. It makes it very easy to learn, but a nightmare to use anything someone else has written.

[Skyrunner]

Profiling for C# is so painful.

Reason: all the profilers seem to be like 500$ for a license. ._.

Also, while I'm sure MSVS also has a profiler, you need a 600$ copy (the Ultimate version, I think) of it to actually use it.

edit:
Thanks, ANTS, you've been very helpful. Thanks for telling me things I already knew, then not provide extra information.

 Also, your MSVS plugin didn't actually install correctly. You're supposed to put it in 2013's folder, but you put it in 2010 and have no option for changing it. Sigh. ;_; Someone help me figure out how to convince ants to give me more info?

edit2: Ok, revealing hidden methods at least gives me the name of the methods. woot!

[Rose]

If you compile your own copy of substrate, and use that, rather than the included dll, then it will show you that most of the time is spent on recalculating the light.

This is also confirmed by the fact that adding caves resulted in a significant speed decrease in generation.

[Skyrunner]

Interestingly, disabling fluid checking makes it slower. Kinda counter-intuitive.

[miauw62]

Ah, SS13 code.


That's a function that returns and calls the parent function. And then returns again. At the end of said function.
Because why not.

[Orange Wizard]

Ah, SS13 code.
<image snipped>

That's a function that returns and calls the parent function. And then returns again. At the end of said function.
Because why not.

Ew.

[Skyrunner]

So... is there a way to force C# to release memory? And by force, I don't mean by calling Collect() and WaitForPendingFinalizers() and hoping that the GC attempts to clean it up

[MagmaMcFry]

So... is there a way to force C# to release memory? And by force, I don't mean by calling Collect() and WaitForPendingFinalizers() and hoping that the GC attempts to clean it up

I don't think so. But if you feel like you're burning through memory too fast for the GC to clean up behind you, you could try reusing your data structures instead of de- and reallocating them.

[Skyrunner]

I have a temporary Bitmap that takes up 1.2GB of space. I kinda need the GC to remove that 1.2 GB immediately after I'm done with processing it into a very much smaller HashSet, or I kinda run out of memory :/

Times like this are when I slightly miss C++... though I have a mild suspicion that the GC actually does free RAM but the Task Manager doesn't reflect it. Perhaps the CLI VM (I think?) Holds onto the 1.2 GB of memory in case I need it later.

[Telgin]

Yeah, you really can't trust Task Manager for that.  What it reports for memory usage includes a lot more than what your program may be actively allocated and using.

When you say you run out of memory, is that actually happening, as in you get exceptions when trying to allocate something new (I think C# does that)?  I'd be really surprised if the runtime environment wouldn't collect memory you're no longer using when it came time to allocate something new, if it didn't have enough free memory in the first place.

[Mego]

I have a temporary Bitmap that takes up 1.2GB of space. I kinda need the GC to remove that 1.2 GB immediately after I'm done with processing it into a very much smaller HashSet, or I kinda run out of memory :/

Times like this are when I slightly miss C++... though I have a mild suspicion that the GC actually does free RAM but the Task Manager doesn't reflect it. Perhaps the CLI VM (I think?) Holds onto the 1.2 GB of memory in case I need it later.

Theoretically, that *should* get paged out to the disk whenever you're not using it, considering its size. If it's really a problem, you could use a random-access file instead of an in-memory data structure, and force it to be on disk. But, it sounds like it's not a huge (huehuehue) problem, especially considering Task Manager lies, as Telgin noted.

[Skyrunner]

Yep, I get exceptions when trying to make another Bitmap of the same size

More research tells me that it has to do with allocating large bitmaps more than running out of memory--continuous, large memory structures are problematic.