I personally have an account on Slayn and Amber. I can say without a doubt that Slayn is a better server than the particular server I have an account on with Amber. This is not to say that slayn is more pleasant to work with or request things from so much as the computer itself that is the server is of very high quality and reliable.
I agree $6, $11, $15 aren't absurd. Though frankly I'm the type of person who would offer to host MUDs for free and probably will make that offer when I get my new computer in a few weeks. (Intel Core Duo 2 2.0 GHz 4 MB Cache 800 667 MHz Cache, 2 GB RAM).
As far as speed I'd rather see compiling speeds than speeds that a speedtest will show myself. Is there perhaps a standard codebase people with accounts on any of these sites should compile and post the results of the time it took?
I know that tbaMUD which is essentially CircleMUD under new developers has no errors with new versions of gcc (I assume that's what you meant when you said gdb) as it is still being updated and paid attention to unlike the now defunct circleMUD.
21 Sep, 2007, Rojan QDel wrote in the 28th comment:
Votes: 0
Please note that if you're testing on the moocowpenguin servers, that we have ccache installed and that it may be used to increase compile speed after the first time it is run. You may use it by adding ccache to the Makefile, it is to be inserted before the gcc command on whatever codebase you're running. IE: ccache gcc mycode.o, etc.
21 Sep, 2007, David Haley wrote in the 29th comment:
Votes: 0
Seems to me like a way of cheating the benchmark, even though it is incredibly useful if you sit around waiting for compiles to finish. Is compiling time really the best way to judge how fast a MUD will run on a given server?
21 Sep, 2007, Rojan QDel wrote in the 30th comment:
Votes: 0
I don't think it really counts as "cheating" the bench-mark, as I don't expect him to use it when comparing to other servers, or if he does, to use it on the other servers as well. It was merely to inform him that it was available and that because of this, on an actual running mud the compile time will most likely feel like less.
What I DO consider a fair judge though is to use something like 'make -j2' or 'make -j3'.
21 Sep, 2007, David Haley wrote in the 33rd comment:
Votes: 0
Sort of. That's a fine metric in that it measures the CPU power and things like having a dual-core processor, but that's not really that relevant for single MUDs since they're not multithreaded. Of course, having one or more dual-core processors can speed up shared hosting environments, so it does have its uses.
And there's still the problem that compiling is just a very different task from running a MUD. Better tests would be to run servers in parallel and simulate heavy load, like lots of people connecting and disconnecting, walking around and so forth. Of course, that's a harder test to set up. :wink:
Nods, ccache can be useful but shouldn't be used as a comparison on the speed of a server in relation to others.
I've used ccache before. I also see someone using it on my server. I don't like it. I stopped using it a few days after trying it. For a source that is 3.6MB, the ccache takes up nearly 1GB of data.
Oh, I'm not saying compiling is the only benchmark we could create, but it is important, at least for muds in heavy developmental stages, whereas a MUD that compiles maybe once a month or less might not care
21 Sep, 2007, David Haley wrote in the 36th comment:
Votes: 0
Zeno said:
I've used ccache before. I also see someone using it on my server. I don't like it. I stopped using it a few days after trying it. For a source that is 3.6MB, the ccache takes up nearly 1GB of data.
Precompiled headers are usually more than sufficient to speed up compilation immensely, because every SMAUG source file includes the massive mud.h file. That's an awful lot of wasted parsing over and over again. One way to help would be to split it into more logical sub-parts instead of a single massive file, but precompiled headers do help. Unfortunately, they're a little finnicky in gcc and don't always work, depending on which version you have, the system you're on, etc.
Fizban said:
Oh, I'm not saying compiling is the only benchmark we could create, but it is important, at least for muds in heavy developmental stages, whereas a MUD that compiles maybe once a month or less might not care
That's true. I guess I tend to not think much about compile time on the server since I do all of my development locally (be it on my own Linux machine, or under Cygwin back when I ran Windows).
I've used ccache before. I also see someone using it on my server. I don't like it. I stopped using it a few days after trying it. For a source that is 3.6MB, the ccache takes up nearly 1GB of data.
You have to set your ccache limit. It can get scary if you don't have a default value for it. Usually keeping it around 100-200megs and clearing the cache every once and awhile does the trick. Shouldn't take up 1gb at all.