Uh what? (Classes of land)

I have been casually flying around looking at land. I haven't bought anything because I wanted to find out everything I can about land before I do. (And I don't have the money right now either.)

However I am super confused about something that seems to be important to a lot of people here. I am looking at stuff on the mainland and people keep on saying some sims are old and some are new. How do I tell if the land I'm looking at is class 4 or class 5 or what have you? And what are the major differences? Is there more than two classes of mainland land?

Thanks in advance!

Samat

There are 3 types of server currently in existance - Type 3, Type 4 and Type 5. Type 5 are the latest and fastest, type 3 are the oldest still running.

The higher the class, the better the performance - although even type 5's can grind to a halt with crap builds, camping chairs and laggy casinos. New land (most of the newer private islands and the new continent) will all be class 5, the older mainland will be mostly 3 and 4, although it seems a bit of an arbitrary mixture as some servers are swapped around and moved as maintenance requires.

There was a site run by someone that listed the server classes for each region, but I can't find it now.

Broccoli

Class 5 sims are on shinier, newer computer hosts.

But a lot of mainland land for sale adverts are full of total lies so I wouldn't really pay much attention to it.

Bottom line is a rustic, rural class 3 is probably going to be less laggy than a mega-mall and casino 5 .. methinks

That site is at http://neighbours.maxcase.info/index.php . If you want to help provide info on servers you can pick up a Registrator-tron HUD attachment in-world.

LL broke Max's stuff a while ago - 6 months, maybe - by changing the way they name the servers. AFAIK, there's no way to get class information any more unless you happen to know that the region was created after LL ran out of class 4 servers. That said, I wear a Registrator-tron whenever I go exploring and encourage friends to put one on their land.

I wouldn't put too much into class 4 vs class 5 hardware. LL has said that there's not that big of a difference between them and you're more likely to run into infrastructure problems before you hit the limits of a regions server..

The major, major difference is script performance - class 5's win hands down with this. Also (this is really hard to test apples to apples) there seems to be far better avatar performance when you have large crowds. The biggest difference I see is between class 4 and class 5 openspace sims - you can really pack a class 5 openspace sim with avatars and not have the horrible performance hit the class 4 openspaces suffer.

So - what happens is people tend to pack more scripted junk into class 5 sims, simply because they can. Eventually you start running into other limits, having to do with say, number of scripts and so forth - all of them have to be managed and there are finite resources to do that.

All in all, class 5's are nice to have, because they are a lot easier to get back to smooth running - they can handle a bit more than the class 4's and thus a little bit of junk removal stands a better chance of nicely recovering performance.

An easy way to tell a Class 5? Keep an eye out for the "fairytale" Sims like Grimm and Gawain. They are brand-new. Other newer Sims include Pudding Hill, Busby, and their neighbors. (This is all assuming that new Sims are on Class 5s. If I'm wrong, someone can smack me around with a wet trout. Which I might in fact enjoy.)

Holds trout horizontally in left hand at eye-level .... checks grip with right hand on tail ... stretches left arm out ... takes strain with right hand ......


servers?

Not doubting you but LL did say the difference wasn't that big - is there any chance you have a source for saying that a class 5 will spank a class 4 on script performance?

<hand up> Miss .. MISS ... Meade said spank miss!! .... oops!

BULLSHIT!
Excuse my french but LL saying this has never made them more FOS, thier motive for saying this is beyond me...

Are you saying that you can't see why a hardware upgrade for a company that has thousands of machines might not be based on just performance?

I read your question at least 10 times and it still makes no sense to me... sorry.. maybe I need more coffee

Ok. Take your time.

They way I see it the Class 3 sims will be the first to be upgraded to Class 7 servers...the future is bright. And no beta testing the Class 5s.

Nope still makes no sense.

Hm.. Maybe you've had too much coffee? Have you tried alcohol? Sometimes that helps for me..

How about if I change it to say: Are you saying that you can't see why a hardware upgrade for a company that has thousands of machines might not be based on just computing performance?

For LL, there's more to buying servers than just things like processing power and memory/IO bandwidth. If, for example, a class 5 server used 20% less electricity than a class 4 server, that would be a big bonus to them. Or if the class 5 hardware had a much better MTBF, that would also be a big bonus.

The point was that they said there wouldn't be much difference between class 4 and class 5, as far as the residents could see. I don't know if that's true or not but it's certainly possible.

I am not an idiot, I understood your question, it just makes no sense why you are asking me it... it had nothing to do with what you are quoting me...
What I was saying was that there is a HUGE performance difference between a class 4 and a class 5, however LL seems to say there isn't... now what your question has to do with that is beyond me...

Thanks for finally saying what didn't make sense.

Ok.

Hi, yes actually, I've done some benchmark testing in-world with known scripts when I bought brand new class 5's for Caledon.

Now, this is where I have to be veeeeeery careful.

It is downright nearly impossible, if not for-a-fact impossible to do apples-to-apples performance testing of two sims.

The why's:

1. Depending what you test, neighbouring sims can affect stats dramatically. Things like 'child agents' - avatars you can see, but not in the same sim.

2. Colocation facilities - Class 5's are in Texas for the most part, as I understand it.

3. General 'stuff' in the sim. Empty, isolated sims the very very best for test comparison.

4. Asset server fetches. If you are doing any testing at all, you may find that the bulk of your testing has to do with how the asset server was feeling at that particular moment, which has dangerously little to do with the sim in question. Simultaneous testing is ideal, but even that is shaky.

5. Testing over hidden sim functions. Yeah, your sim will go do a backup or something for a few moments, and if you are unlucky your test will catch it. Gotta test a few times to get an idea.


Now, all that said, you can STILL tell a used VW microbus from a Ferrari regardless of most road conditions.


What some friends far smarter than I had done, was get some known scripts with generally known, error-bracketed execution times that we could measure on the order of tens of seconds.

For many of the reasons listed above, these tests didn't always come up with the same values, but you could get a sense of it. In the hardest technical terms I can use: class 5 totally pwns class 4.

I suggest this: find the emptiest class 4 and emptiest class 5 you can (totally empty is ideal) and run a test script that takes about a minute to complete in your 'home' sim, just looping. Try it various places and see the difference.

We did some 'general' testing as well, which was far less scientific but a lot of fun. Imagine getting 25+ avatars running just fine in an openspace sim - essentially unheard of prior to class 5 sims. We had 29 I think for this test; I've heard of upwards of 60 did okay - laggy but okay - in a class 5 openspace. We tried to crash the sim but couldn't, save with physics:

http://cavorite.hexaflexagon.com/blog/?p=56

Another 'easy' test for class 4 -vs- class 5 when the sims are relatively full is to click on a rento box. There's usually a bit of slowdown in the green text, as a class 4 is usually hitting its script time limits - this is a LOT more rare in full class 5 sims. Click-bam, there's all the green text in an eyeblink. Very unscientific, but your first hint that a fully inhabited sim has a lot more horsepower under the hood.

As for general lagginess - usually it's the asset server, not the sims, that rule the day. Easy thing for people to mistake. Also, unless you are on a fairly fast system and getting 40+ fps on your client you are likely having enough clientside 'object' lag to make lag isolation a very difficult proposition.

Thanks for the response.

I've sometimes wondered how much of the lag I see in my mainland home sim is due to cruft that's accumulated over time. All the prim litter with scripts that have been slowly eating memory and listening to every word, all the mystery spots in the region that you can't walk through, all the garbage that people moved underground and forgot about, etc, etc, etc..

I was just curious if you had done any real looking into class 4 vs class 5 or if it was the general perception that the new boxes were better. Usually people end up saying that the class 5 servers "just feel better" but it sounds like you've done some serious looking..

Thanks again.

LL refused to move Le Cadre' from a class 4 to a class 5 even though it was giving us tons of trouble and we were prepared to pay the cost, even if it meant paying them $1675 for the server, they refused..
We purchased a class 5 and cloned Le Cadre on to the new island and not only is script running time decreased by 40%, there is mysteriously no longer packetloss and time dialtion is steady. This was done by an exact replica, prim for prim, script for script... which included about 3800 scripts and 13,500 prims.

3800 scripts!?

Wow. That's heavy usage even for a class 5. At those numbers, the sheer number of scripts starts to matter. For a class 4 - that's backbreaking, I'm not surprised that you made the change.

Do you actually need 3800 scripts?

Best thing anyone can do is 'quad up' their textures (4 512x512 -> 1 1024x1024) and reduce scripts to maybe 1500 or so. And don't 'rotate' textures if you can help it, such as on vendors, say. No matter what sim you have, no matter what asset server conditions you are going to be much better off.

Alot of the scripts are coming from our furniture store which have sit scripts and animation scripts, we also have Blingo, Bingtirs and Greedy, things we can't and don't want to do with out hehe =)
it does sound like a lot of scripts but they are all optimized (except bingtris) to the best SL allows, everything but the games are running at .003 or less, total script time in Le Cadre currently is 20.8ish and when we moved all the content to our new class 5 it strangely dropped the script time to 10.0-12.5ish, Le Cadre currently now has 2521 as we started tearing down to prepare for the conversion and is still running at 19-20.8ish script time.