Does LL have any professionals?

I purchased a parcel today..a 1024 for 12300 lindens and divided it for resale....I split it into two 512 parcels...and then went to put them both up for sale for 7150 each...I definitely had a green line down the center of my parcel...dividing it into two parcels...I put one up for sale for 7150, and the second that I clicked the confirmation for putting it up for sale...the green line disappeard...the parcels reverted into a single parcel of 1024 again, and a landbot swooped it in half a second A landbot run by Zor Zeddmore, of course both the bot and Zor ignore my request to make things right...so I am out 5150 lindens...and so I called the concierge support, to fix this obvious problem...

I got hold of Guy Linden, and he came to the parcel and did some sort of "examination" while I waited..then told me he could do NOTHING. There is obviously an undocumented bug in the system...I am hardly in the habit of buying land and selling it for much less than I buy it for...and I sure saw the green line on the parcel.

So...basically, the lindens are supporting the "right" of the landbots to rip us off, because of a bug in THEIR systems????...I want everyone to know...buy land at your own risk in this game...you have a better chance keeping your money if you just flush it...

Naaaaw............it must be your graphics card, or your firewall. Or you have a 4 years old eMachine running a Celeron P3 processor with 256 megs ram.

You can see from the many posts on this thread that Linden Labs has a top notch platform using the ultimate architecture with coding so flawless no one else in the entire universe can match them. It just could not be a bug or even a minor glitch..........it simply has to be on your end. That's what Guy Linden was trying to say..........he can't do anything because it wasn't LL's fault.

Clear your cache, update your driver, get more memory and try again. If that doesn't fix it then reinstall with the most current viewer. All your problems are solved. "Issue resolved"...........heheheheheh, I'm a linden now. LOL

Like Nectere, I appear to be one of the 'lucky' ones.

I don't teleport around as much as Darien... maybe up to 10 times a night... I did see one TP fail a month or so ago.

I've had approximately 3 viewer crashes in the last 3 months.

When I'm on, I usually stay on for *hours*. Weekends, mornings, evenings... doesn't seem to make much difference.

Honestly, I haven't checked to see if SL is leaking memory... because I'm not having viewer "problems".... and I've only got 1gb of ram in my computer!

I used to get logged off due to a sim crashes more often than I thought reasonable (used to be a few times a night) but it's down to once every 3-5 days now, which I find "reasonable" at the moment. Wouldn't mind if it were better.

Temporal - add your feedback to VWR-235 in jira. It sounds like a related symptom of a known problem!


Carl- "If you can answer those and about 1000 other questions, great, we can agree to disagree or debate the finer points of each. If not then back to ur armchair."

Carl, dude... Sindy wasn't making any boasts about her on professional expertise, only calling your posturing into question. Likewise... I call your bluff:

Show us you know how LL could architect something better than they have... as you seem to imply you're so very capable of... what with your "professional posturing"

Or... just sit there in that safe little armchair and show that your opinion is nothing but fluff and wind.

Why not? You continually bluff with the factoid that 40 AVs in a sim means (n^2) bandwidth output which is simply, demonstrably wrong.

Is LL paying you for the fanboy posts? They should.

hmm seems the 40x40 formula is about right to me... 40 avatars receiving updates for 40 avatars (yourself included since the server correct your client's predictions)

-add to this of course 40x A where A is the number of primitives getting an update at this frame. (badly coded blings for example)
-add also 40x b where B is the updates of the physic engine for every objects nearby.

more stuffs i imagine

yeah for Carl, all i have been reading so far is ego centric wind blowing, it looks impressive on the paper but it solve nothing.

Nope, I'm strictly on a volunteer basis.

And I'm not bluffing... I've stated my line of thinking and my conclusion.

You COULD say that in a sim with 40 people bouncing around that the lag is so bad that they can't do as MUCH as one person in a sim all by themselves. Fine, so it may be n^1.9. That's in the same ballpark as "Squared".

Don't just bluff and say my claim is "demonstrably wrong".

Back up your accusation with something solid.

And thank you =)

Most blings are, I hope, set-and-forget particle effects, not constantly updated, so that's just a client-side rendering issue.

Animation Overriders however, 40 people with bad scripts asking "What's he doing now?" four times a second, resulting in streams of animations to 40 viewers (more if there's folks are near enough to a sim boundary). Hopefully, the animations get cached on the client rather than streamed each time... I've never gotten a good sense of whether they are or not.

And then there's the "oo, I have a bracelet that changes colors randomly as fast as it possibly can" spamming everyone with updates.

Fortunately, seated people are treated as non-physical, so... thank poseballs for helping reduce the sim processing overhead to some degree.

I'm getting a bit tired of repeating this as you seem to forget as soon as the thread is over. But, once more: you assume that every AV in a sim affects every other client in that, and as the maximum number of agents in a sim is 40 you get a 40^2 traffic multicast burden.

But your assumption is more often than not incorrect. Look at the pattern of green dots on a sim. Those further than 20m from each other have chat culled before it is sent out the door. Those outside the draw distance of each other will have their "updates" similarly culled from view by others. Many clients are set to draw distances of 96m and distribute themselves in small clumps. Yes, I'm intentionally ignoring distribution patterns in clubs as they are far from ubiquitous, indeed probably less than one club per 3 sims.

I could go on, but that's all you get as this is as rewarding and productive as debating theology with Kevn.

Malachi Petunia: I'm getting a bit tired of repeating this as you seem to forget as soon as the thread is over.

Hm. Refresh my memory? I can't find it in search... where did you correct me on this point before? My apologies, if I did honestly miss or forget it.

Malachi Petunia:But, once more: you assume that every AV in a sim affects every other client in that, and as the maximum number of agents in a sim is 40 you get a 40^2 traffic multicast burden. ... But your assumption is more often than not incorrect.

I assume no such thing.

You're welcome to assume I'm too stupid to be aware of this... I won't take that personally. Much.

Okay, so I didn't hedge and clarify my statement like a lawyer, fine:

In cases where 40 ACTIVE residents are in a sim each WITHIN VIEWING DISTANCE OF EACH OTHER, the amount of network output from the sim increases by squares, and 40 people means 40*40 as many updates being communicated out to client viewers.

Chat traffic, by the way, is probably inconsequential. A packet when someone starts typing, a few packets if they have something really long to say when they hit return. Keep in mind that the typing animation and the sound that accompanies it is seen/heard well beyond the 20m radius, so these kinds of updates, few as they are relatively, aren't completely culled at 20m.

(and the sim still has to expend the effort to figure out whom to notify and whom not to notify, where with "viewable" updates, it may just be easier to send them all and let the client figure out whether to render them or not. Who knows? Not I. I haven't seen the server code, but it could buy processor cycles at the expense of network bandwidth, and LL has shown a tendency to make that choice before.)

It would be far far worse for 40 residents to press and hold the "move forward button" (several updates per second from every avatar) all at once than it would be for them all to say "Norm!" all at the same time.

I'm not going to debate the frequency or likelyhood of 40*40 situations... I'm just pointing out that for each person in a gathering the number of people needing updates grows to unsupportably numbers very quickly. Maybe 30 people here, that's 30*30 times as much output as a solitary person... fine. Enough groups of 30 around the grid is still going to spill over and affect other sims.

For each "busy" sim, there are likely 3 other sims on the same server fighting over the same network interface.

For each rack of 20 servers or 80 sims, there's probably one network switch (or blade in a switch more likely).

Of 4000 sims... 1000 servers, 50 racks, in two or more co-locations... all that network traffic is probably sifting through a small number of redundantly deployed bottlenecks before spilling out into the internet.

Is there any wonder that there's packet loss problems?

It will be a long time before we see LL allow 100 people in a sim.

Even the most clever architecture is going to have difficulties accomodating that.

And if Carl thinks LL can "do better" ... I'd like to see him pony up some of this "professional expertise" of his to show that it's even remotely possible.

Well ... why on earth do you thnk its 40*40 ??? I have a real-time system with 40 inputs and 40 outputs ... at each interval X i need to update the state of that system and then send that to all the 40 participants ...

There is *no* square involved here ... its a simple linear relationship ... the servers job is to reconcile the 40 inputs and send a single packet of data for all the updates to all 40 AVs to each client at once ... no big deal at all ...

Of course you *could* design a system that scaled N^2, but since it can be done O(N) why on earth would you?

--Carl

Because when it *does* run for longer the client VM footprint continually edges higher and higher ... i starts at 300M approx and continues upward forever ... usually getting to 2G and then crashing ...

If u can justify a 2G memory footprint for the client go ahead ...

And 3ds max is a reasonable comparison since it has a large full featured 3D graphics subsystem, physics engine, and animation system. I can load scenes that are VASTLY more complex into the app and still run in say 400-500M ..

--Carl

...and as am FYI on this thread ... as of this post the SL billing system has been down for over 24 hours and is still not running ...

But of course SL is the best that can be and so this is excusable ... as is every other flaw in the system. I'm not expecting SL to be perfect, but its sad that so many people here rush around to defend stuff so strongly with apparently no justification.

--Carl

Are all 40 of your participants are on the same subnet and you're using broadcast? Are you using multicast?

No i dont think so, nor do i think that would help or hinder much ...

All these simulation engines tend to work in the same way ... they have phases for physics, anims, graphics updates etc ...that means that the sim runs the various scripts etc for each AV, possibly (hopefully) several at once on an SMP system, then once the sim state is updated it can compose a single global update packet/packets and send then to the clients.

Sending that update to 40 clients isnt a biggie even w/o multi-cast, and in fact should really be a candidate for off-loading.

Frequently one optimization that helps such systems is to split each phase onto a thread and then force thread affinity such that a physical CPU is dedicated to one task. Normally this is bad as it causes excess thread switching but in face in these tight coding models the loss caused by thread switching is easily made up for processor cache "warming" that keeps the cache rate very high ... by far the biggest factor in CPU thruput today.

Carl: "Sending that update to 40 clients isnt a biggie even w/o multi-cast, and in fact should really be a candidate for off-loading."

There may be useful merit to your suggestion.

But it would be little more than a spit in the rain when it comes to helping SecondLife deal with its scaling problems more gracefully.

I was expecting something a little more - effective - from someone who is "not just venting in ignorance guys ... ive been designing complex systems not unlike SL for years"

Sure, I agree... SL has problems aplenty. Some are growing pains, some are transitional pains, some are the results of difficult no-win compromises. There is nothing else in this world like SecondLife, and flawed or not, there's no call for accusing LL of being incompetent.

Maybe LL's goals are ultimately unattainable with today's technology. Good for them for having the gusto to give it their best shot.

You may be a brilliant software developer in your field, (at least I hope so! I'd hate to fall from the sky some day! ) but with regards to SecondLife's architectural design, I think your "Expert Witness" credibility is bit lacking.

Unless, of course, you have something more insightful to suggest?

Ok. I haven't looked at 'max for a while.. How many people get virtually instant access to what you're doing, as you're doing it?


Aside from you, who said SL is the best that can be?


What's sadder, and what I think was a big reason why the forums that were for interacting with LL were closed, is the number of people who come here and do nothing but whine. Not talking about posting on legitimate issues - I have no problem speaking up when LL does something I don't agree with or breaks things - but the constant whining. "LL sucks," "LL doesn't care about anybody," "LL is a bunch of monkey weasels," "LL is unprofessional," blah, blah, blah, etc, etc, etc..

That's all this thread is really about, Carl. You started whining and people asked if you could actually do anything else.

I'm on 1.14.0.0, I think so far its crashed twice since i made the update, running sl almost constantly (yes i'm a total addict) more than acceptable for me.



I'm on 2gig too, I agree it can eat a lot, but not so much to cause problems with all the other simultaneous stuff i have running like Live messenger, winamp, firefox etc



The client download is tiny considering the world we explore (because its stored their end of course). Unless this is another reference to memory usage, in which case, why call this one 3) when its 2) again?



My tp's seem to work about 90% of the time, guess i'm just lucky.



The systems you compare to have been in development for donkeys years and by multiple developers throughout the world. Linden Labs is one company, doing something no-one else has done before (and still hasn't, have yet to see any clear rival. If your understanding of such complex systems is so great, by all means start your own Second Life)



You're venting, and clearly very ignorant. 'not unlike' there has been NO system like Second Life till now, period. Are you really that stupid to think they aren't hiring some serously good programmers to sort all this stuff out. It's in their interests!!. DUH!

And if its soo clear cut to you, how come no-one else has made a rival to Second Life yet hmm? *looks around* I don't see anyone. There's one trying to start up in Australia, still not even off the ground yet, wonder why?. could it be... ITS DIFFICULT?!? *shock*

*shakes head*



It's about time you realised thats what they've been doing , and mainting Second Life is a damn site harder than you clearly have any idea of comprehending.

Ignorant ranting like this is as bad as the ass-lickers.

i DO agree SL could be a lot better, i do NOT agree that its because they dont know what they're doing.

So we'll have to agree to disagree

I have less of a problem when someone displays the depth of their ignorance by screaming "FIX IT!" without realizing that "it" is a delicate balance of speed, affordability and reliability. In wild-west frontier-like systems, such as SecondLife, you just don't get all three.

But when someone spouts off that "I'm a professional, and I hereby declare this to be a scatological nightmare" or some such pundit-level nonsense... it rather chaps my hide.

Few enough people have a sufficient grasp of software, hardware, networking, databases, client server design and the business of providing a receational service in a new market to speak with that kind of authority. Those that DO are likely too busy to stop and make such a comment in a forum space such as this.

Of course, those that THINK they have that kind of knowledge are more likely to be here mouthing off. Safer that, than risking actually trying to put that knowledge to productive use only to find out there are more gaps and holes in that understanding than previously thought.

And, before anyone beats me to it... I will quite emphatically suggest I'm in that "I only think i have that kind of knowledge" category. I wouldn't be here on the forums mouthing off otherwise.