Chat log from the meeting on 2012-10-16

From OpenSimulator

Jump to: navigation, search
[11:06]  VivK Lowlag: hi Sarah
[11:06]  Connecting to in-world Voice Chat...
[11:06]  Connected
[11:06]  Shelann Morane: I just guessed this was the place they could be some place else
[11:06]  Monet Fitzgerald: that does sound like a fun time Lani
[11:07]  Snow Gravois: meh?
[11:07]  Monet Fitzgerald: Maybe we can plan one ?
[11:07]  Tiffany Sicling: this place is lagging to me
[11:07]  Snow Gravois: maby it was joke?
[11:07]  Justin Clark-Casey: hello folks
[11:07]  Snow Gravois: lol
[11:07]  Nebadon Izumi is Online
[11:07]  Tiffany Sicling: hi Justin
[11:07]  Sarah Kline: HI Justin
[11:07]  Shelann Morane: Howdy
[11:07]  Snow Gravois: lol are we goign nude?
[11:07]  Richardus Raymaker: hi justin
[11:07]  Lani Global: Greetings, Gridlings!
[11:07]  Robert Adams: hi Justin
[11:07]  Richardus Raymaker: Tiffany, posisble your still loading
[11:07]  Tiffany Sicling: hi Neb!
[11:08]  Monet Fitzgerald: Good morning Neb
[11:08]  Nebadon Izumi: hello
[11:08]  Sarah Kline: hello
[11:08]  Nebadon Izumi: sorry I was late, i am still not used to going from 11am meeting times to 2am
[11:08]  Nebadon Izumi: err 2pm
[11:08]  Justin Clark-Casey: you've moved east?
[11:09]  Nebadon Izumi: ua
[11:09]  Nebadon Izumi: ya
[11:09]  Justin Clark-Casey: permanently?
[11:09]  Nebadon Izumi: ya well for the near future anyway
[11:09]  Andrew Hellershanks is Online
[11:09]  Nebadon Izumi: not moving back to Arizona anytime soon hehe
[11:09]  Tiffany Sicling: he wanted to see the sun come up before anyone else
[11:09]  Lani Global: how is your ping time?
[11:09]  Justin Clark-Casey: oh, did they kick you out? :)
[11:10]  Nebadon Izumi: hehe
[11:10]  logger sewell: that could very well be Tiffany :)
[11:10]  Andrew Hellershanks: I'm back.
[11:10]  Nebadon Izumi: i get like 100ms ping to this server or so
[11:10]  Nebadon Izumi: depends
[11:10]  Tiffany Sicling: it's 1:10pm here
[11:10]  logger sewell: wb Andrew
[11:10]  Nebadon Izumi: id say 70-100ms
[11:10]  Andrew Hellershanks: machine locked up on me again. I think I really need a new power supply.
[11:11]  Richardus Raymaker: hmm. sure its power not cooling ?
[11:11]  Andrew Hellershanks: Fans are fine.
[11:11]  Snow Gravois: use superglue
[11:11]  vegaslon plutonian: wd 40
[11:12]  Monet Fitzgerald: i have to set my laptop up on a picture frame on the table so it stays cool... else it locks up
[11:12]  Monet Fitzgerald: actually it turns off
[11:12]  Lani Global: Laptop coolers are your friend.
[11:12]  Tiffany Sicling: that sounds like it needs a cleaning in the fan dept.
[11:12]  Andrew Hellershanks: THis machine isn't a laptop.
[11:13]  Lani Global: muffin fans are your friend.
[11:13]  Richardus Raymaker: cpu cooler can still hold lots of dust
[11:13]  Richardus Raymaker: but you know what graphics card you have and what power supply
[11:14]  Andrew Hellershanks: yea. Close to limits. PSU used in this was a low wattage one.
[11:14]  Lani Global: opensim viewers make the video and cpu work hard
[11:14]  Richardus Raymaker: yes
[11:14]  Snow Gravois: They do?
[11:14]  Anya Heberle: hi
[11:14]  Shelann Morane: LOL
[11:14]  Andrew Hellershanks: I had to swap cards when my old video card died.
[11:14]  Snow Gravois: I hadn't noticed because of all the lag.
[11:14]  Nebadon Izumi: afk a sec
[11:14]  Basil Sosides: ?.?.:* HaLlIHaLlOoOoooo  *.:??
[11:15]  vegaslon plutonian: ya have to keep speedfan on with these veiwers
[11:15]  Andrew Hellershanks: oh, I still have to file that bug with patch for groups
[11:15]  Dahlia Trimble is Online
[11:15]  Anya Heberle: this Firestrorm is good better than SL2
[11:15]  Anya Heberle: firestorm
[11:15]  Justin Clark-Casey: Had to fall back to viewer 1 because SL 2 insists on repeatedly changing its settings to try and upgrade itself
[11:15]  Snow Gravois: What about SL3?
[11:16]  Andrew Hellershanks: I had FS but dropped it as I didn't like something about it but can't remember what now.
[11:16]  Justin Clark-Casey: I need to rearrange my infrastructure to switch to another viewer
[11:16]  Anya Heberle: but I crashtexture launching lol
[11:16]  Anya Heberle: never used SL3
[11:16]  Snow Gravois: it sucks.
[11:16]  Tiffany Sicling: I have SL 3.3.4 only to upload mesh
[11:16]  Erik Brun: Hi Folks
[11:16]  Richardus Raymaker: sl3 have still some problematic bugs for me,. and LL SL3 dont work here anyore default.
[11:16]  Andrew Hellershanks: I'm using CoolVL but may switch to Singularity when it gets mesh.
[11:16]  Lani Global: Phoenix Viewer 1.6.1 (1691) works very well, has mesh, doesn't self update.
[11:17]  Richardus Raymaker: ANdrew, the misplaced cursor ?
[11:17]  Andrew Hellershanks: richardus, the what?
[11:17]  Richardus Raymaker: Phoenix dont have snapshot to disk. thats problken on some systems
[11:17]  Lani Global: phoenix has snapshot to disk.
[11:17]  logger sewell: the only thing with FS is you cant see light share
[11:17]  Andrew Hellershanks: Singularity doesn't support saving passwords separately for each grid.
[11:17]  Richardus Raymaker: FS have with bigger UI the cursor at the wrong position compared to where the charachter appears
[11:18]  Dahlia Trimble: hi :)
[11:18]  Tiffany Sicling: hi Dahlia
[11:18]  Sarah Kline: etc etc
[11:18]  Anya Heberle: anyone want to try out a Rocket ship ,A my Sim , I just found one and want to show it off!!
[11:18]  VivK Lowlag: hi Dahlia
[11:18]  logger sewell: Hi Dahlia
[11:18]  Lani Global: phoenix works with lightshare
[11:18]  Snow Gravois: We realy need a better universal client.
[11:18]  Basil Sosides: oh...Big Viewer Discusion
[11:18]  Richardus Raymaker: Lani, phoenix have not snapshot to disk, you can snap till you die. nothing get saved
[11:18]  Andrew Hellershanks: yeah
[11:18]  Nebadon Izumi: back
[11:18]  Justin Clark-Casey: hi Dahlia
[11:18]  Tiffany Sicling: speaking of flying stuff... defaulting to 35m would be great for that ;)
[11:19]  Anya Heberle: what?
[11:19]  Andrew Hellershanks: SIngularity was first viewer I've ever successfully compiled (and I compiled it in Linux)
[11:19]  VivK Lowlag: run into ban lines with those and see if 35m is great
[11:19]  Richardus Raymaker: SIngularity is idneed sofar the best one. with less problems
[11:19]  Tiffany Magic: Lani... we tried your script and recommendations, but still can't cross simlines. We must be doing something wrong.
[11:19]  Tiffany Sicling: can't do flying craft in physical prims bigger than 10m
[11:19]  Andrew Hellershanks: Richardus, I have a few pending bug reports about it.
[11:20]  Anya Heberle: oh
[11:20]  Lani Global: Tiffany, the phsyical vehicle can cross, but it unsits the pilot. the pilot must re-sit.
[11:20]  Anya Heberle: whos coming its so good!
[11:20]  Andrew Hellershanks: The idea of changing to 32m for physical seems ok.
[11:20]  Richardus Raymaker: ANdrew, i have so many bug reports. you almost lose track :O
[11:20]  Anya Heberle: good luck otherwise
[11:20]  Justin Clark-Casey: I know Teravus had reservations about increasing physical prim max last meeting
[11:21]  Justin Clark-Casey: robert: do you have an opinion on this?
[11:21]  Lani Global: any prim with avatar sitting on it, when moved across a sim border, will unsit the avatar.
[11:21]  Tiffany Magic: Ok.. ours, the physical prim disappears.
[11:21]  Anya Heberle: bye for now
[11:21]  Nebadon Izumi: 64m is for sure too big
[11:21]  Richardus Raymaker: Not sure how tests ended but last thinbg i know. it did not sunded good
[11:21]  Anya Heberle: good luck otherwise
[11:21]  Nebadon Izumi: just 1 64x64x64 prim has pretty bad impact on the physics and sim fps
[11:21]  Anya Heberle: stop by sometime! :D
[11:21]  Tiffany Sicling: shouldn't the regions share enough information between themselves to keep an avatar sitting on an object when it crosses a border?
[11:21]  Andrew Hellershanks: The e-mail indicates 32m is ok
[11:21]  Snow Gravois: Hardly. I strongly beleave it should be raised to 256m not 64.
[11:22]  Nebadon Izumi: ya 32m is better for sure
[11:22]  Lani Global: Avatar Un-Sit Crossing Border
[11:22]  Pete Camino: 32 is fine
[11:22]  Tiffany Sicling: 256m is a tad too big
[11:22]  Nebadon Izumi: I think thats the high end for sure
[11:22]  Snow Gravois: Have you seen my ships?
[11:22]  Richardus Raymaker: SNow, 21 december still need to come :O
[11:22]  Tiffany Sicling: 32m sounds like a good number for phys prims
[11:22]  Lani Global: i've done some research testing on Physical Prim Size.
[11:22]  Lani Global: I can present a graph today
[11:22]  logger sewell: but it doesn't crash the sim i tried a 256 cubed to test
[11:23]  Andrew Hellershanks: SOmeone tested prim sizes and sent an email with the results. 32 was ok
[11:23]  Sarah Kline: how far are you going to fly a 256m cube on a region
[11:23]  Lani Global shows the graph to the meeting.
[11:23]  Nebadon Izumi: ya that was Lani Andrew
[11:23]  Justin Clark-Casey: which e-mail was this, Lani/Amy's e-mail?
[11:23]  Justin Clark-Casey: ok
[11:23]  Lani Global: yes
[11:23]  Snow Gravois: 4k m and up
[11:23]  VivK Lowlag: was this tested with ban lines as well?
[11:24]  Nebadon Izumi: with ban lines?
[11:24]  Justin Clark-Casey: banlines don't work properly right now anyway
[11:24]  Tiffany Sicling: how would mesh vehicles be affected with fps ?
[11:24]  Nebadon Izumi: what effect would ban lines have?
[11:24]  VivK Lowlag: still crash the viewer
[11:24]  Richardus Raymaker: hmm 20 meters is to big. sost to many frames
[11:24]  Lani Global: on the graph, the red is sim physics too slow. the green is good.
[11:24]  Nebadon Izumi: I actually made a mesh vehicle once
[11:24]  Justin Clark-Casey: lani: where is this graph?
[11:24]  Justin Clark-Casey: ahh I see :)
[11:24]  Justin Clark-Casey: I'm not thinking virtual
[11:24]  Lani Global: on the screen here in the meeting, see the graph.
[11:24]  Richardus Raymaker: left from you justin in front of fire
[11:25]  Andrew Hellershanks: wow... torus performance sucks
[11:25]  Lani Global: i tested various size prims and prim shapes.
[11:25]  Lani Global: torus is 'worst case'
[11:25]  Lani Global: sphere is 'best case'
[11:25]  Andrew Hellershanks: what about cube?
[11:25]  Lani Global: as you can see, there is quite a difference
[11:25]  Tiffany Sicling: I guess I won't be making any physical flying gates with torus
[11:25]  Lani Global: cube is near sphere performance
[11:25]  Justin Clark-Casey: this is a single sphere/torus?
[11:26]  Andrew Hellershanks: Lani, ok
[11:26]  Lani Global: this is with 1 or 2 each
[11:26]  Lani Global: dropped from 100m onto uneven ground with objects
[11:26]  Snow Gravois: Did you check this with sculpted prims yet?
[11:26]  Andrew Hellershanks: even doubling limit to 20m torus makes fps drop by almost a factor of 2
[11:26]  Lani Global: sculpty, yes. it is similar to torus performance
[11:27]  Lani Global: well, for the purposes of this test and the graph, i considered anything lower than 10FPS to be bad.
[11:27]  Tiffany Sicling: now did the flying mesh go, neb?
[11:27]  Tiffany Sicling: *how
[11:27]  Nebadon Izumi: I didnt do flying mesh
[11:27]  Richardus Raymaker: everything below 20 is bad.
[11:27]  Nebadon Izumi: just a truck
[11:27]  Nebadon Izumi:
[11:27]  Richardus Raymaker: betetr to not go below 30
[11:27]  Nebadon Izumi: here is video link
[11:27]  Justin Clark-Casey: Is this graph available at an url?
[11:28]  Andrew Hellershanks: Can't see the video wit the sign in the way
[11:28]  Snow Gravois: what video?
[11:28]  Lani Global: certainly spheres and common box prims can be up to around 45m or 50m without problem
[11:28]  Lani Global: the torus starts to be a problem around 37m
[11:28]  Snow Gravois: But sculpts cant?
[11:28]  Monet Fitzgerald: when i created my avalance on my ski sim... i used sculpty rocks. and dropped them in physical mode...
[11:28]  Lani Global: everyone is welcome to copy this graph
[11:28]  Monet Fitzgerald: but they would bounce to the other sims... and cause massive lag so had to remove them
[11:28]  Snow Gravois: What would you say the largest sculpt could be safely?
[11:28]  Andrew Hellershanks: I have to get going. RL Is calling
[11:29]  Nebadon Izumi: I put video of my mesh vehicle up on screen
[11:29]  Richardus Raymaker: bye andrew
[11:29]  Justin Clark-Casey: bye andrew
[11:29]  Nebadon Izumi: see you andrew
[11:29]  Tiffany Sicling: I see the video... nice
[11:29]  Andrew Hellershanks is Offline
[11:29]  Lani Global: sculpty is similar to torus, in its impact to sim physics.
[11:29]  Richardus Raymaker: why is the movie stoping always ? like there's lag when you made it
[11:29]  Nebadon Izumi: hmm video is lagging a bit for me
[11:29]  Nebadon Izumi: ya
[11:29]  Nebadon Izumi: here is the direct link
[11:29]  Nebadon Izumi:
[11:30]  Nebadon Izumi: you can just download the file
[11:30]  Richardus Raymaker: that gives the same problem nebadon
[11:30]  Nebadon Izumi: some reason its real laggy on the screen
[11:30]  Richardus Raymaker: aha yes
[11:30]  Nebadon Izumi: guess its not right specs for this screen
[11:30]  Tiffany Sicling: no lag here and I am on 4G
[11:30]  Tiffany Sicling: viewing the video that is
[11:30]  Snow Gravois: i see gray
[11:30]  Nebadon Izumi: but anyway, I had no problem with mesh vehicles
[11:31]  Nebadon Izumi: thats not to say there arent any though
[11:31]  Richardus Raymaker: wich size nebaon ?
[11:31]  Nebadon Izumi: not sure exactly what the max prim size is on that
[11:31]  Sarah Kline: I think that truck thing had quite a few polys )
[11:31]  Nebadon Izumi: i suspect its below 10m
[11:31]  Nebadon Izumi: ya the truck mesh itself is out at "Reticulation"
[11:31]  Tiffany Sicling: oh I notice that at times you unsit and your AV cannot turn left or right afterwards.
[11:31]  Nebadon Izumi: i dont know what the heck i did with the scripted one
[11:31]  Nebadon Izumi: i think i lost it
[11:32]  Nebadon Izumi: i'll have to make another
[11:32]  Snow Gravois: I noticed that just about every time I crossed the boundry and came back.
[11:32]  Richardus Raymaker: i have the one in inventory i think. 4x4 ... ?
[11:32]  Lani Global: So, previously, at last week's meeting, a change to the default PhysicalPrimMax=64 was proposed. However, after testing and evaluating this further (at the urging of Nebadon)... a new value of 32m is being proposed, based upon the test results.
[11:33]  Richardus Raymaker: if i look at your graph. 20 meter would be the max.
[11:33]  Lani Global: the value of 32m provides a margin of safety.
[11:33]  Justin Clark-Casey: The thing is. one isn't necessarily talking about one or two physical prims here
[11:34]  Justin Clark-Casey: there could suddenly be many large physical prims, which would seem to imply much worse performance characteristics
[11:34]  Monet Fitzgerald: I second Lani's proposal
[11:34]  Nebadon Izumi: well I beleive part of the problem with the 64m prims
[11:34]  Snow Gravois: I would say about 27 would be the proposed max.
[11:34]  Pete Camino: most small ship builds won't require more that a couple of 'oversized' premitives
[11:34]  Snow Gravois: according to the graphic.
[11:34]  Nebadon Izumi: Teravus was saying that they contact too many places or something along those lines
[11:34]  Nebadon Izumi: the region is cut up into 16m square zones
[11:34]  Justin Clark-Casey: I think he was saying there would be a big increase in the number of contacts
[11:34]  Justin Clark-Casey: yes
[11:34]  Richardus Raymaker: 32 meter and up yes
[11:35]  Nebadon Izumi: and the 64m prims enter too many of them
[11:35]  Tiffany Sicling: well, that's with 1 or two prims, and at 20m limit, would seem okay. but what about a ship or craft that is like, 20 prims linked?
[11:35]  Nebadon Izumi: this was the result
[11:35]  Justin Clark-Casey: oh, your video shows the rsult?
[11:35]  Lani Global: A proposal, in the form of "Open Letter to the OpenSim Developers" went out to opensim-dev email list yesterday.
[11:36]  Monet Fitzgerald: I have large ships i build as well... and they are over 500 prims but at least 32 is a good start
[11:36]  Simulator Version v0.5 shouts: OpenSim 0.7.5 Dev          8459b98: 2012-10-16 01:40:25 +0100 (Unix/Mono)
[11:36]  Snow Gravois: Mine are well over 500 prims as well.
[11:36]  Pete Camino: if a ship gets to large theres no point in flying it on a sim
[11:37]  Richardus Raymaker: i would go to 16meters. think thats acceptable and safe
[11:37]  Snow Gravois: There is though.
[11:37]  Snow Gravois: we can go up to 4km right? so why not?
[11:37]  Pete Camino: i'd say 20 minimum and 32 is better
[11:37]  Justin Clark-Casey: robert: are you there?
[11:37]  Robert Adams: 32 is a good number and if it doesn't work, we should make it work
[11:37]  Justin Clark-Casey: robert: how do you think 32m would work with bulletsim?
[11:38]  Robert Adams: I have too many kyboards on my desk
[11:38]  Richardus Raymaker: Bullet i think works betetr with 32m
[11:38]  Baewyn Celara: There's always megaregions for the larger ships.
[11:38]  Justin Clark-Casey: the problem with megaregions is bugs and hacks to make them work with vanilla viewers
[11:39]  Baewyn Celara: I never said it was perfect.
[11:39]  Snow Gravois: Mega prims cause lag.
[11:39]  Monet Fitzgerald: I am able to run 42 sims as one mega sim
[11:39]  Snow Gravois: expecialy the sculpted ones as I have found out.
[11:39]  Richardus Raymaker: not really
[11:39]  Monet Fitzgerald: fully loaded
[11:39]  Monet Fitzgerald: and it works great
[11:39]  Robert Adams: BulletSim handles 32m pretty well... single prims on a flat surface jump around a bit (contact points far apart) but large linksets work ok
[11:39]  Monet Fitzgerald: even used a boat with a new script that WORKS...
[11:39]  Snow Gravois: Nice video.
[11:40]  Monet Fitzgerald: like the music
[11:40]  Monet Fitzgerald: but i am getting hypnotized
[11:40]  Nebadon Izumi: heh
[11:40]  Justin Clark-Casey: then perhaps one could increase to 32m with bulletsim - or perhaps max prim size should even depend on which physics engine is enabled
[11:40]  Nebadon Izumi: ya I think 32m is a good #
[11:41]  Nebadon Izumi: I would like to maybe have a few more people test it
[11:41]  Nebadon Izumi: really put it to the test a bit
[11:41]  Shelann Morane: Shelann raises a hand
[11:41]  Richardus Raymaker: find people with loess powered sims.
[11:41]  Lani Global: using ODE, 32m is safe
[11:41]  Justin Clark-Casey: please just speak Shelaan
[11:41]  Justin Clark-Casey: everybody else does :)
[11:41]  Lani Global: 64m using ODE is unsafe
[11:41]  Robert Adams: +1 on 32m as default.
[11:41]  Justin Clark-Casey: lani: 32m looks pretty bad performance-wise with even 1torus from your craph
[11:41]  Justin Clark-Casey: graph
[11:41]  Shelann Morane: Do we change Max Prim in INI or do we wait for an update?
[11:42]  Nebadon Izumi: physical torus are pretty nasty stuff
[11:42]  Justin Clark-Casey: shelann: there is an ini config setting
[11:42]  Richardus Raymaker: well i keep mine on 10
[11:42]  Shelann Morane: yes
[11:42]  Robert Adams: there are many ways people can slow their sim down... this is just another
[11:42]  Nebadon Izumi: lots of vertices in a torus
[11:42]  Nebadon Izumi: right
[11:42]  Robert Adams: if it hurts, don't do it
[11:42]  Justin Clark-Casey: robert: that's true
[11:42]  Nebadon Izumi: if you can rez 1 prim
[11:42]  Nebadon Izumi: you can rez 1000 prims
[11:42]  Lani Global: Justin - yes, for most shapes of prims 32m is fine with ODE. however, with Torus, 32m is OK, but not perfect. I would say this is a compromise.
[11:42]  Richardus Raymaker: 1000 physics ones :O
[11:43]  Tiffany Sicling: I still remember that huge pyramid I made on that test region years ago, that I turned physical and it fall off the sim and crashed it
[11:43]  Justin Clark-Casey: lani: have you tried looking at performance with more than 1 or 2 prims?
[11:43]  Monet Fitzgerald: Quick off the subject question for next meeting..... can we increase the number of linked prims? from 256 to ?
[11:43]  Snow Gravois: I have a question.
[11:43]  Nebadon Izumi: Monet its already possible
[11:43]  Richardus Raymaker: Monet, i think you can already change that yourself
[11:43]  Monet Fitzgerald: oh cool
[11:43]  Lani Global: i've tried this with 50 prim vehicles with 32m prim size doing collisions with ground and objects also
[11:43]  Nebadon Izumi: but only a couple viewers support it
[11:44]  Justin Clark-Casey: shelann: the setting is PhysicalPrimMax in [Startup] in OpenSim.ini. It should also work with individual region config, thuough I occasionally seem to hear probems about that
[11:44]  Nebadon Izumi: only Imprudence and Singularity support it
[11:44]  Snow Gravois: Is there any development in the way towards something simular to the Havoc Engine?
[11:44]  Lani Global: sculpty vehicles
[11:44]  Nebadon Izumi: and its a bit tricky
[11:44]  Jalen Optera: Many of the viewers impose the linked prim limit
[11:44]  Nebadon Izumi: you need to go very slowly
[11:44]  Nebadon Izumi: if you select 1024 prims
[11:44]  Nebadon Izumi: and link it, you will get 4 256 prim blocks
[11:44]  Richardus Raymaker: SNow, <looks to robert> i think bullet comes close then
[11:44]  Justin Clark-Casey: lani: and? what kind of perf did you get?
[11:44]  Nebadon Izumi: you then have to wait for the region to finish datastoring that linking
[11:44]  Nebadon Izumi: what I generally do is type "backup" on the sim console
[11:44]  Robert Adams: Snow, I'm working on an engine with Bullet.
[11:44]  Monet Fitzgerald: so changing the physical prim thang you do in one or the other right? either region.ini OR opensim.ini.... not both?
[11:45]  Snow Gravois: The Havoc engine can support 1k physical prims with no problem
[11:45]  Shelann Morane: thanks Justin
[11:45]  Nebadon Izumi: that forces a datastore and then its safe to select those 4 blocks
[11:45]  Robert Adams: Havok requires paying for a commercial license
[11:45]  Nebadon Izumi: and then link the 4 linksets
[11:45]  Nebadon Izumi: you will have 1 object
[11:45]  Justin Clark-Casey: monet: should be possible in both. Individual regio nconfig should overrule what is in [Startup]
[11:45]  Nebadon Izumi: with 1000 prims
[11:45]  Tiffany Sicling: too bad we cannot put havok in here
[11:45]  Lani Global: using a 55 prim vehicle made of sculpty and torus with max size 32m, it caused sim to go to about 20FPS during collision with uneven ground.
[11:45]  Snow Gravois: If I recall LL has an agreement with Havok not to share there engine with us anyways.
[11:45]  Nebadon Izumi: ya uneven ground makes things much much worse
[11:45]  Richardus Raymaker: already happy that 140 prim physics verhicle works good. or its because windows is better.
[11:45]  Justin Clark-Casey: well, as people say, there are a lot of things you can do already to slow down a sim
[11:46]  Lani Global: i tried to test worst case
[11:46]  Nebadon Izumi: right
[11:46]  Richardus Raymaker: in the past 140 prim pushed timings up
[11:46]  logger sewell: havox wont even talk to us they told me that it's not compatable with opem sim theyu are stuck with SL i think
[11:46]  Justin Clark-Casey: raising the limit may be of most concern for people trying to share sims with lots of others, or even rent them
[11:46]  Lani Global: of course, anything you do with physics has limits.
[11:46]  Nebadon Izumi: if you setup your sim properly, then realistically there is nothing anyone can do
[11:46]  Monet Fitzgerald: ok.. that explains it Neb... thanks I remember now that i was using imprudence when i linked it last time and this time was using phoneix
[11:46]  Tiffany Sicling: and havok's 'free' for non-profit agreement is very sticky too
[11:46]  Robert Adams: BulletSim does 1000 physical prims...
[11:46]  Nebadon Izumi: right phoenix doesnt support it, nor does FIrestorm
[11:46]  Nebadon Izumi: just those 2 that I know of Monet
[11:46]  Richardus Raymaker: well, someday i replace it for mesh
[11:47]  Snow Gravois: you do thre is another engine we could use.
[11:47]  Snow Gravois: it uses the more advanced part of the graphics card to do this.
[11:47]  Baewyn Celara: PhysX?
[11:47]  Snow Gravois: Aum I think there called tesulas or something. Its a DX 12 thing.
[11:48]  Tiffany Sicling: PhysX is in most of the current nvidia cards
[11:48]  Snow Gravois: Yes it is.
[11:48]  Richardus Raymaker: i wuld say cuda
[11:48]  Snow Gravois: It would be a huge advance over Havoc.
[11:48]  Shelann Morane: Neb/Justin...are you comfortable enough to raise Max prim to 32 in the next update?
[11:49]  Robert Adams: Bullet has OpenCL versions that would be accelerated on NVidea, AMD, ..
[11:49]  Nebadon Izumi: anything is possible, but your probably talking about a few years of work
[11:49]  Nebadon Izumi: hehe
[11:49]  Richardus Raymaker: OpenCL good to
[11:49]  Sarah Kline: Its interesting a mesh vehicle only counts as 1 prim...therefore you can have over 30 passengers sitting on it and it still drives
[11:49]  Justin Clark-Casey: I want to reply to the lani's ml post first - I still have concerns about perf. ideally I would test myself
[11:49]  Sarah Kline: (sorry bit behind)
[11:49]  Snow Gravois: Im not sure how making the engine works but on the test program I got to use it had over 10k physical objects. and its was perfectly smoth.
[11:49]  Nebadon Izumi: ya I would like more people to test and report
[11:49]  Justin Clark-Casey: but there would be nothing to stop osgrid raising the limit in its own config files ;)
[11:49]  Nebadon Izumi: maybe add their results to Lanis wiki page as well
[11:49]  Nebadon Izumi: can you paste that Wiki page Lani
[11:50]  Lani Global: yes
[11:50]  Monet Fitzgerald: WEll i wouild like to raise my hand and say... "I make yachts for OpenSim and need larger size physical prims"
[11:50]  Richardus Raymaker: above 20m physics looks tricky on the cart. because framedrop
[11:50]  Monet Fitzgerald: 32M is my vote
[11:50]  Lani Global:
[11:50]  Richardus Raymaker: Monet, well you can change it yourself in the ini file
[11:50]  Monet Fitzgerald: oh\
[11:50]  Monet Fitzgerald: i thought we needed approval here?
[11:51]  Lani Global: Content Creators for OpenSim Community must build to the defaults.
[11:51]  Richardus Raymaker: it dont need to be set for everybody. if you want bigger monet you can try
[11:51]  Justin Clark-Casey: monet: if it's your sim you can change what you want
[11:51]  Tiffany Sicling: Monet, you can change your own... if it breaks, you have your own pieces
[11:51]  Robert Adams: a lot of the slowdown is due to sending 10K prim updates
[11:51]  Snow Gravois: Aum we do?
[11:51]  Richardus Raymaker: but remeber that your shi posisble dont work outside your sim
[11:51]  Monet Fitzgerald: well it will be interesting to see how it does with megas
[11:51]  Justin Clark-Casey: robert: generated by the physical movement?
[11:51]  Monet Fitzgerald: i got back up so np
[11:51]  Dahlia Trimble: when testing something like this, it may be a good idea to have lots of moving avatars and lots of stationary content with smaller prims
[11:52]  Snow Gravois: The wey the engine works that I am talking is they are rendered on the users grpahics card not the server.
[11:52]  Baewyn Celara: That sounds like pathfinding...
[11:52]  Dahlia Trimble: if you only have a few avatars you might not see any effects from a single large object generatting lots of collisions
[11:52]  Robert Adams:
[11:52]  Tiffany Sicling: yeah, a slow computer with an old video card will take the biggest hit
[11:52]  Robert Adams:
[11:53]  Pete Camino: i release my products under creative commons license free of charge. they can be copied and modified but not transferred, sold, or given to others.
[11:53]  Snow Gravois: Who even plays this with a old video card?
[11:53]  Pete Camino: i have released more than 30 vehicles on the grid and have a dozen or more waiting to be released.
[11:53]  Justin Clark-Casey: reobrt: the question is how they look in motion ;)
[11:53]  Justin Clark-Casey: pete: that's not what the CC license says
[11:53]  Pete Camino: now that we have the beginnings of physical flight available to us it has come to my attention, and others as well, that the 10 meter size limit has effectively tied our hands creatively. i don't build flying boxes. i try to build interesting and aerodynamically designed vehicle shapes that people will want to own and enjoy.
[11:53]  Robert Adams: Pictures of BulletSim doing >3000 physical objects
[11:53]  Tiffany Sicling: there are some that try to use a laptop with the 950 chipset, it's SLOW
[11:53]  Richardus Raymaker: oh you made a new one robert. looks funny
[11:53]  Pete Camino: the ten meter maximum primitive size limit keeps me from using larger individual primitives such as torus, spheres, and cylinders that have been cut, shaped, and or, stretched to make more interesting components parts to my vehicles so that they don't look like flying boxes or have reparative less desirable designs. allowing us to increase this individual primitive size within the whole will allow us more creative options and make a better product for the end users.
[11:54]  Nebadon Izumi: there are a few different variations of Creative Commons
[11:54]  Nebadon Izumi: that do disallow commercial use and resale
[11:54]  Justin Clark-Casey: well, the physical prim limit has remained for a reason, because of performance issues
[11:54]  Justin Clark-Casey: esp with ode
[11:54]  Richardus Raymaker: i think robert wherey eathing chipitoes at the time :)
[11:54]  Justin Clark-Casey: I don't think any CC prohibits copyuing though
[11:54]  Nebadon Izumi: right
[11:54]  Nebadon Izumi: or transfering
[11:54]  Justin Clark-Casey: we don't keep these limits for fun you knwo
[11:55]  Richardus Raymaker: Pete, you can change it on your own sim
[11:55]  Nebadon Izumi: well thats why we are also asking people to test it
[11:55]  Robert Adams: we keep them for control ;-)
[11:55]  Nebadon Izumi: to help us find out faster if its really worth it
[11:55]  Pete Camino: i don't run my sims servers but i have requested the ini change
[11:56]  Nebadon Izumi: in all likely hood we will probably increase the limit above 10
[11:56]  Nebadon Izumi: we just need to be sure we don't go to far
[11:56]  Lani Global: I've shown that 32m is good compromise, while still providing a safe margin. This enables the needs of the OpenSim community to have content while keeping simulators from crashing.
[11:56]  Richardus Raymaker: i think when it change. its good to make it someay clear to everybody. (news?)
[11:56]  Dahlia Trimble: there's nothing magic about 10 meters. If youhave sparse large, large objects in a sim its not really different than small physical objects with small content
[11:56]  Dahlia Trimble: its relative
[11:57]  Lani Global: yes, i agree with dahlia
[11:57]  Justin Clark-Casey: lani: I don't think you have shown that - what you have shown is that for 1 prim the dropoff can be okay or fairly bad (down to 10 fps)
[11:57]  Pete Camino: them maybe increase it inn stages starting at 20
[11:57]  Justin Clark-Casey: lani: what one really needs is a sense of how performance is affected with a few more prims
[11:57]  Dahlia Trimble: if all you have is a terrain, yuo probably wont see any effects from huge physical objects if they are distant from the terrain
[11:57]  Lani Global: my tests were done with 2 prims bonking together on uneven ground, dropped from 100m. It did not crash the sim
[11:57]  Lani Global: sure, things slow down
[11:57]  Tiffany Sicling: How about a chart showing 1 prim up to 20 prims linked and physical.
[11:57]  Robert Adams: might also reduce maximum prim mass... large physical objects are sometimes bad just because they get too heavy
[11:57]  Richardus Raymaker: Nebadon, is the test betetr when the terrain is not flat...
[11:57]  Lani Global: but no crash
[11:57]  Dahlia Trimble: what else was in the sim?
[11:58]  Nebadon Izumi: Lanis sim is pretty busy
[11:58]  Nebadon Izumi: but I agree, we need to test on a wide variety of hardware
[11:58]  Dahlia Trimble: I dont know what "pretty busy" means
[11:58]  Nebadon Izumi: varying degrees of building and scripting
[11:58]  Robert Adams: set it to 32m and if baddness happens, revert it
[11:58]  Nebadon Izumi: I dont know the exact numbers
[11:59]  Tiffany Sicling: Lani has lots of running scripts and lots of prims large and small on her region(s)
[11:59]  Nebadon Izumi: but I think she has a bunch of animated prims and scripts
[11:59]  Dahlia Trimble: how many avatars present? did any fall thru what they were standing on?
[11:59]  Nebadon Izumi: and her terrain is very hilly bumpy
[11:59]  Justin Clark-Casey: the problem is it can take a long time for badness to manifest
[11:59]  Justin Clark-Casey: or be reported, rather
[11:59]  Nebadon Izumi: ya
[11:59]  logger sewell: I'll ask Andrew to do some testing in VH and record it I'll bring it in next week
[11:59]  Robert Adams: nebadon has never been quiet about baddness :)
[12:00]  Nebadon Izumi: ok well I suggest for the people who are testing and reporting back to the wiki information
[12:00]  Nebadon Izumi: you please include information about your server hardware
[12:00]  Nebadon Izumi: resources etc..
[12:00]  Nebadon Izumi: what kind of cpu, total memory
[12:00]  Nebadon Izumi: that would be helpful
[12:00]  logger sewell: i sure will neb
[12:00]  Snow Gravois: Do you want namebrand as well?
[12:00]  Lani Global: yes, if anyone needs help with the testing, give me a call
[12:00]  Nebadon Izumi: what we really need is people with lower end hardware
[12:00]  Monet Fitzgerald: well... let me jump in.... i made a tidal way the size of my sim width. stretched a cylander and had ti roll across my sims with targets type script...
[12:00]  Nebadon Izumi: no, name brand is not needed
[12:00]  Nebadon Izumi: just want to get a feel for what kind of hardware and how it performs
[12:01]  Nebadon Izumi: be specific about your testing
[12:01]  Nebadon Izumi: what you did, experienced
[12:01]  Tiffany Sicling: my server is an old dell poweredge 1850, is that old enough ?
[12:01]  Nebadon Izumi: maybe if you make a video or something paste a link to it
[12:01]  logger sewell: well we have a test server that is a 2 gig machine we can use
[12:01]  Dahlia Trimble: also record how many avatars and if they were walking, sitting, or standing
[12:01]  Dahlia Trimble: and if they notice any effects
[12:01]  Snow Gravois: and mine are 1759
[12:01]  Snow Gravois: 1750*
[12:01]  Nebadon Izumi: ya those are good servers to test on
[12:01]  Nebadon Izumi: I was testing on a pretty fast machine
[12:02]  Nebadon Izumi: it was a VM
[12:02]  Tiffany Sicling: I have only one region on it too lol
[12:02]  Snow Gravois: I have 12.
[12:02]  Nebadon Izumi: but the VM had 2 cores of a i5-3550
[12:02]  Nebadon Izumi: and 4gb ram
[12:02]  Nebadon Izumi: thats pretty fast setup
[12:02]  Richardus Raymaker: and im sure linux have a harder job with physics then windows
[12:02]  Nebadon Izumi: not really Richardus
[12:02]  Nebadon Izumi: my money is on Linux
[12:02]  Nebadon Izumi: heh
[12:02]  Dahlia Trimble: ª4
[12:02]  Justin Clark-Casey: yes, I wouldn't have though there would be a big difference
[12:03]  Justin Clark-Casey: thought
[12:03]  Tiffany Sicling: 2 x xeon 3.0GHz dual cores with 3GB RAM
[12:03]  Dahlia Trimble: my sims run in a dual core atom
[12:03]  Justin Clark-Casey: lol
[12:03]  Dahlia Trimble: I probably wont be raising the limits :)
[12:03]  Nebadon Izumi: heh ya thats on the low side of things for sure
[12:03]  Snow Gravois: 2x Xeon 3.6 GHz Dual core with 4 GB Ram on mine with scsi 15k rpm hard drives.
[12:03]  Nebadon Izumi: Atom is basically a Pentium 4
[12:03]  Richardus Raymaker: Well, i see opensim run much betetr on windows then on the same system with linux. linux start todo crazy weird
[12:03]  Snow Gravois: times 40
[12:04]  Nebadon Izumi: or Maybe a Pentium D
[12:04]  Tiffany Sicling: I only got 10K scsi drives
[12:04]  Richardus Raymaker: you all have older hardware that linux likes much more
[12:04]  Justin Clark-Casey: richardus: there is a different because of better GC on windows - nothing to do with physics per se
[12:04]  Tiffany Sicling: rpm
[12:04]  Dahlia Trimble: mono GC *shudders*
[12:04]  Tiffany Sicling: windows would run better than linux on physics?
[12:04]  Richardus Raymaker: Well, linux really dont like amd X6. it works realy starneg start to lag fast etc.
[12:04]  Nebadon Izumi: Ive not really seen any evidence of that really
[12:05]  Tiffany Sicling: ahh, I have intel
[12:05]  Justin Clark-Casey: dahlia: I tried with the sgen GC the other day and things are ... different
[12:05]  Nebadon Izumi: the simulator in general might work slightly better on WIndows in certain cases
[12:05]  Richardus Raymaker: and maby ODE is beter but in the past i ended with phyics time of 180ms. i now have a new train but on windows and its 0ms
[12:05]  logger sewell: thats all we have ever run is windows we just started withh it
[12:05]  Nebadon Izumi: but I doubt there is much difference in physics behavior
[12:05]  Justin Clark-Casey: oddly, much higher lagging behaviour with a high load, though I'm wondering if that's something to do with our use of sleep to control the scene loop
[12:05]  Tiffany Sicling: yeah, but the overhead OS will take up more memory though
[12:05]  Dahlia Trimble: Justin, I have problems with sgen as well
[12:05]  Justin Clark-Casey: which many things I've read say is a bad idea since it's not very accurate
[12:06]  Tiffany Sicling: I just don't see windows needed in a headless server
[12:06]  Dahlia Trimble: I have some huge memory leaks on mono with my pathfinding stuff, on windows, perfect :)
[12:06]  Tiffany Sicling: ahhh... pathfinding
[12:06]  Baewyn Celara: I need to remember to try running opensim from wine with .net installed
[12:06]  Justin Clark-Casey: dahlia: urgh nasty
[12:06]  Justin Clark-Casey: dahlia: is that all VM code or some native as well?
[12:07]  Justin Clark-Casey: baewyn: ha
[12:07]  Dahlia Trimble: the version with the memory leaks is all managed code
[12:07]  Baewyn Celara: bad idea?
[12:07]  Justin Clark-Casey: baewyn: actually - it might not be much different since wine uses mono
[12:07]  Justin Clark-Casey: to run .net stuff
[12:07]  Justin Clark-Casey: dahlia: that's disturbing
[12:07]  Baewyn Celara: I meant with the .net runtime installed
[12:07]  Richardus Raymaker: Linux just dont support new pc hardware good, thats my experience with many tried on intel and linux
[12:07]  Richardus Raymaker: uhmm intel & amd
[12:07]  Justin Clark-Casey: baewyn: I suspect it would still use mono
[12:08]  Dahlia Trimble: ya and mono people cant figure it out either
[12:08]  Richardus Raymaker: still next time i go for intel. seems to work anyway betetr
[12:08]  Baewyn Celara: :/
[12:08]  Justin Clark-Casey: rira: really - linux supports new hardware before many other things
[12:08]  Justin Clark-Casey: 64 bit for instance
[12:08]  Richardus Raymaker: Possible justin. but amd x6 the dotn get right with opensim.
[12:08]  Tiffany Sicling: mine works well, and total server memory usage (with xwindows) is just under a gig.
[12:08]  Baewyn Celara: yeah that worked out swimmingly
[12:09]  Baewyn Celara says running a 32bit viewer on 64bit linux with no stream support due to lib incompatibilities
[12:09]  Richardus Raymaker: besides scripting runs better on windows. less event problems
[12:09]  Justin Clark-Casey: anyway, I need to leave now
[12:10]  Tiffany Sicling: I still think that raising the physical prim limits in default .ini to 32m would be nice, as I do have some craft that is waiting to be deployed that are bigger than 10m prims in it.
[12:10]  Richardus Raymaker: not sure linux used between 4-8GB windows is hanging around 8GB
[12:10]  Dahlia Trimble: I think its more mono vs .net than linux vs. windows
[12:10]  logger sewell: Justin tc and have a good evening
[12:10]  Sarah Kline: Bye Justin
[12:10]  Justin Clark-Casey: tiffany: I don't have an issue with it as long as we're reasonably sure it isn't going to have a very bad effect on fps
[12:10]  Dahlia Trimble: bye
[12:10]  Baewyn Celara: bye Justin
[12:10]  Tiffany Sicling: why release them if nobody can use it?
[12:10]  Richardus Raymaker: I think dahlia is right. possible mono is chocking on my system
[12:10]  Richardus Raymaker: i runb linux with robust and 1 region on old hardware very nice
[12:11]  Justin Clark-Casey: there's no point having these nice big craft if more than a couple of prims on the same sim bring it to its knees
[12:11]  Justin Clark-Casey: alright, see you folks later
[12:11]  Tiffany Sicling: Friday, I will have 30Mbps down, 5Mbps up on my Internet, so I would welcome anyone to come there for flying craft testing (greater than 10m physical prims)
[12:11]  Richardus Raymaker: bye justin
[12:11]  Nebadon Izumi: see you Justin
[12:11]  vegaslon plutonian: well the point of flying is to not crash into the ground
[12:11]  Dahlia Trimble: need more testing across the community. I think if the community wants it the limit should be raised
[12:12]  Justin Clark-Casey waves
[12:12]  Nebadon Izumi: I agree Dahlia
[12:12]  Tiffany Sicling: bye Justin!
[12:12]  Richardus Raymaker: aha nice, shame its not 5down and 30 up :)
Personal tools
About This Wiki