[11:09] Ubit Umarov: welcome :)
[11:10] Ubit Umarov: gaving was asking if vicent notices that KDU is better than openjpeg on viewers
[11:10] Ubit Umarov: :)
[11:11] Gavin.Hird @grid.xmir.org:8002: curious minds needs to know :-)
[11:11] Vincent.Sylvester @hg.zetaworlds.com:8002: kdu runs slightly faster due to lower memory footprint, but actually loads images a bit slower
[11:11] Vincent.Sylvester @hg.zetaworlds.com:8002: openjpeg is like instantly everything has textures where kdu displays gray for a bit longer
[11:11] Vincent.Sylvester @hg.zetaworlds.com:8002: Overall though the kdu claims of 10x performance are bullshit
[11:12] Gavin.Hird @grid.xmir.org:8002: between kdu and the screen sits the LL renderer
[11:12] Ubit Umarov: ( possilbe andrew is finding he has a new naked ruth avatar here at brave new osgrid..)
[11:13] Cuga.Rajal @rajal.org:9000: I'm seeing that graphics freeze issue on 7.x release also
[11:13] Vincent.Sylvester @hg.zetaworlds.com:8002: The lower memory footprint of kdu is useful in cases with limited system memory, it only adds like 4fps from what I can tell in the same scene at the same position
[11:14] Gavin.Hird @grid.xmir.org:8002: but are teh frames better??
[11:14] Lyr.Lobo @cc.opensimulator.org:8002 chuckles
[11:14] Ubit Umarov: jp2000 decode should have minimal impact on fps
[11:14] Andrew Hellershanks: Hello, everyone. I wound up at the other place. I knew I would forget we are back here. I had to look up the name for the region again as I no longer had a landmark for here.
[11:15] Gavin.Hird @grid.xmir.org:8002: Hi Andrew
[11:15] Lyr.Lobo @cc.opensimulator.org:8002: Hello, everyone! I'm celebrating my birthday today. Sorry to be late, just popped out of class. *grins*
[11:15] Cuga.Rajal @rajal.org:9000: HB, Lyr!
[11:15] Ubit Umarov: Oh no... another birthday..
[11:15] Ubit Umarov: oops
[11:15] Lyr.Lobo @cc.opensimulator.org:8002 laughs
[11:15] Ubit Umarov: Happy Birthday
[11:15] Ubit Umarov: :)
[11:15] Gavin.Hird @grid.xmir.org:8002: Congrats!
[11:15] Lyr.Lobo @cc.opensimulator.org:8002: Thank you *grins*
[11:17] Andrew Hellershanks: Happy Birthday, Lyr.
[11:17] Gavin.Hird @grid.xmir.org:8002: I mean can you see any visible differences between KDU and openjpeg? – To me it looks like KDU produce slightly better lighting in a scne
[11:18] Gavin.Hird @grid.xmir.org:8002: scene
[11:18] Vincent.Sylvester @hg.zetaworlds.com:8002: The big reason for openjpeg is that dynamic textures, say a 1024x1024 sized texture will load much faster so you do not need to add a delay for the viewer to finish display, because for some reason that's buggered now too. KDU seems to apply some different colors to some shaders which is visible in the water mostly
[11:19] Vincent.Sylvester @hg.zetaworlds.com:8002: Not sure how that even comes about frankly, but that's the big thing I noticed
[11:19] Ubit Umarov: had idea that main diference is that KDU costs $$ :)
[11:19] Andrew Hellershanks: I'll be a cloud for todays meeting. The grid seems to know what I had been wearing but I have no inventory.
[11:19] Gavin.Hird @grid.xmir.org:8002: which is the main, main difference
[11:21] Vincent.Sylvester @hg.zetaworlds.com:8002: For how much it costs its pretty shit at dealing with the dynamic textures though, but I guess doesn't show itself much at SL seeing as they don't have those
[11:21] Vincent.Sylvester @hg.zetaworlds.com:8002: ....yet
[11:21] Ubit Umarov: well about opensim code changes..
[11:22] Ubit Umarov: think we had one.. a one char change!!
[11:22] Gavin.Hird @grid.xmir.org:8002: that is at least 8 bits that changed
[11:22] Ubit Umarov: a bad space was removed from a string in code!!
[11:22] Vincent.Sylvester @hg.zetaworlds.com:8002: cosmetics on asset connector too
[11:23] Ubit Umarov: that was causing viewers allow/exclude feature to fail
[11:23] Ubit Umarov: at robust gatekeeper
[11:25] Gavin.Hird @grid.xmir.org:8002: question: do we transmit any benefits cap at all?
[11:25] Ubit Umarov: no
[11:26] Gavin.Hird @grid.xmir.org:8002: so the benefits code is actually pointless in the opensim viewer
[11:26] Ubit Umarov: basicly...
[11:26] Gavin.Hird @grid.xmir.org:8002: good
[11:27] Ubit Umarov: maybe some grids would like to have UltraPremiumPlusHyper
[11:27] Vincent.Sylvester @hg.zetaworlds.com:8002: lol
[11:27] Ubit Umarov: but not going there
[11:27] Gavin.Hird @grid.xmir.org:8002: or subscription to use Hypergrid
[11:28] Vincent.Sylvester @hg.zetaworlds.com:8002: Don't give em ideas
[11:28] Ubit Umarov: Ohh now that is a nice idea..
[11:28] Gavin.Hird @grid.xmir.org:8002: :-)
[11:28] Ubit Umarov: pay opensim
[11:28] Ubit Umarov: ( my paypal ofc )
[11:29] Gavin.Hird @grid.xmir.org:8002: If you left opensim to Adobe there would be a forest of subscriptions to enable features
[11:29] Ubit Umarov: ohh wait for windows 12
[11:29] Gavin.Hird @grid.xmir.org:8002: not really
[11:30] Vincent.Sylvester @hg.zetaworlds.com:8002: The more you buy... wait no that's hardware
[11:30] Ubit Umarov: the more you pay, the less you get :p
[11:30] Gavin.Hird @grid.xmir.org:8002: there was this joke "ask not what Vista can do for you. Ask what you can buy for Vista"
[11:31] Vincent.Sylvester @hg.zetaworlds.com:8002: Viewer performance basically has to improve now that there isn't really a high end option for graphics cards anymore, so that'll be interesting to see
[11:32] Vincent.Sylvester @hg.zetaworlds.com:8002: Through the testing I been doing the last couple days I did notice a strange relation between cache size and framerate
[11:33] Vincent.Sylvester @hg.zetaworlds.com:8002: Bewildering that a large cache will result in slower rendering... somehow... yeah I don't know either why that is related
[11:33] Vincent.Sylvester @hg.zetaworlds.com:8002: Digging into performance unearthing the weirdest things
[11:33] Cuga.Rajal @rajal.org:9000: would have expected the opposite
[11:33] Vincent.Sylvester @hg.zetaworlds.com:8002: Right? Me too
[11:34] Ubit Umarov: hmm in old days we could have a ferrari from matchbox.. Sadly we can't have a matchbox 5090
[11:34] Andrew Hellershanks: I remember when people were experimenting with setting the limit on network traffic rates and higher wasn't always better.
[11:34] Cuga.Rajal @rajal.org:9000: yay the dino!
[11:35] Vincent.Sylvester @hg.zetaworlds.com:8002: OpenSim limits the max rate to about 4000kbit anyways, you cannot set it higher than that even if the viewer requests it
[11:35] Gavin.Hird @grid.xmir.org:8002: because the viewer spends too much time handling incoming trafic and not rendering frames
[11:35] Andrew Hellershanks: I hear a noise and thought someone had an open mic. I see now that it was the dinosaur passing by behind me.
[11:35] Cuga.Rajal @rajal.org:9000: FS Prefs only go to 3000k
[11:35] Vincent.Sylvester @hg.zetaworlds.com:8002: Yeah that's what seems to be responsible for the freezes as well given they occur loading objects from an adjacent region
[11:36] Vincent.Sylvester @hg.zetaworlds.com:8002: You can set a higher number in the debug settings, but it gets clamped on the server end
[11:36] Cuga.Rajal @rajal.org:9000: ahh ok
[11:36] Vincent.Sylvester @hg.zetaworlds.com:8002: SL appears to allow like 20000 for some reason
[11:36] Vincent.Sylvester @hg.zetaworlds.com:8002: Not sure if they clamp it
[11:36] Gavin.Hird @grid.xmir.org:8002: they do
[11:36] Cuga.Rajal @rajal.org:9000: yeah I see that freeze when I come in range of a neighboring sim
[11:36] Cuga.Rajal @rajal.org:9000: in SL
[11:37] Kayaker.Magic @grid.kitely.com:8002: The dinosaur survived the restart of OSGrid!
[11:37] Ubit Umarov: guess he only lives on this region cache
[11:37] Vincent.Sylvester @hg.zetaworlds.com:8002: Viewer code seems to lack some safety to not overload itself with processing object data and just does everything it gets in a single frame
[11:37] Cuga.Rajal @rajal.org:9000: Dino didn't lose any skin or body parts
[11:38] Vincent.Sylvester @hg.zetaworlds.com:8002: I may dig into the code on the server end to see if the outgoing data can be throttled, but ultimately seems more a viewer bug
[11:38] Gavin.Hird @grid.xmir.org:8002: probably because the overhead of such code will slow down frame rate
[11:38] Vincent.Sylvester @hg.zetaworlds.com:8002: Yeah
[11:38] Cuga.Rajal @rajal.org:9000: slow it down slower than a freeze
[11:39] Vincent.Sylvester @hg.zetaworlds.com:8002: It lacks the "only do x amount of objects per frame" type break in the processing to prevent itself from locking up
[11:39] Vincent.Sylvester @hg.zetaworlds.com:8002: Course finding the suspect code for that is not easy
[11:39] Cuga.Rajal @rajal.org:9000: it probably doesn't have a mechanism to store the unprocessed objects in a buffer for the next frame
[11:39] Cuga.Rajal @rajal.org:9000: in order to limit object processing and not freeze
[11:40] Vincent.Sylvester @hg.zetaworlds.com:8002: It may well be code that takes incoming network data and sticks it into memory as well, which is harder to throttle in the first place
[11:40] Gavin.Hird @grid.xmir.org:8002: sounds like a bit increase in memory use right there
[11:40] Gavin.Hird @grid.xmir.org:8002: big*
[11:41] Gavin.Hird @grid.xmir.org:8002: plus you have to traverse it to make sure the objects have not come out of view since the last frame
[11:41] Ubit Umarov: mb you are just filling up yr gpu .. memory
[11:41] Vincent.Sylvester @hg.zetaworlds.com:8002: Throttles are transmitted as a set of individual throttles per type, so perhaps just need to adjust that ratio to slow object rate down a lot. I did try changing the burst rate at the server end to reduce the amount it would send at once, but that didn't seem to do anything
[11:42] Vincent.Sylvester @hg.zetaworlds.com:8002: When it comes to such network stuff I prefer to hide under a rock
[11:42] Gavin.Hird @grid.xmir.org:8002: :-))
[11:42] Cuga.Rajal @rajal.org:9000: I dont think the burden to manage that should be at the server, should be viewerr]
[11:42] Ubit Umarov: guess udp has little to do with that
[11:43] Ubit Umarov: region will stop sending if there are 2 many unack packets also
[11:43] Cuga.Rajal @rajal.org:9000: VFiwere should receice, and have a mechanism to throttoe or defer processing/rendering
[11:43] Ubit Umarov: that extra control was the only way to stop having lost terrain patchs
[11:44] Vincent.Sylvester @hg.zetaworlds.com:8002: The viewer sends sort of ratios for the data types and then the server takes those and does math to figure out the actual data rate. Suppose that was implemented to spec, so yeah the viewer can send different numbers to get different data rates back
[11:44] Ubit Umarov: bahh old old newa
[11:44] Ubit Umarov: news
[11:44] Vincent.Sylvester @hg.zetaworlds.com:8002: The question remains as to why that has gotten so bad as it is a somewhat recent development that people report freezes like that in the first place
[11:44] Ubit Umarov: and some viewers try to be clever on that to be "faster"
[11:45] Vincent.Sylvester @hg.zetaworlds.com:8002: Yeah that too, FS code not exactly to spec
[11:45] Cuga.Rajal @rajal.org:9000 whispers: I don't remember seeing it on 6.x. Maybe it is an
[11:45] Cuga.Rajal @rajal.org:9000: Fast FPS" thing
[11:46] Gavin.Hird @grid.xmir.org:8002: they seem to pile on loads of code for crash prevention, but all that code has overhead in terms of processing frames
[11:46] Ubit Umarov: i didn't notice that.. but i don't go out much things days
[11:46] Vincent.Sylvester @hg.zetaworlds.com:8002: I suspect the extra layer for the materials crap has negatively impacted it, but that alone shouldn't be this massive given it's a native capability of the renderer to display that stuff
[11:46] Ubit Umarov: hmm ( note to self checp if car still starts )
[11:46] Gavin.Hird @grid.xmir.org:8002: how is it native?
[11:47] Vincent.Sylvester @hg.zetaworlds.com:8002: heh yeah
[11:47] Vincent.Sylvester @hg.zetaworlds.com:8002: A mess
[11:47] Ubit Umarov: render engine is for pbr now
[11:47] Ubit Umarov: new light model etc etc etc
[11:48] Gavin.Hird @grid.xmir.org:8002: does not mean it is particularly native
[11:48] Gavin.Hird @grid.xmir.org:8002: the renderer is a hybrid mess that also can render pbr
[11:48] Vincent.Sylvester @hg.zetaworlds.com:8002: Technically it is meant to be faster because the format is closer to what the renderer has to output anyways, but in practice that's a different story
[11:48] Vincent.Sylvester @hg.zetaworlds.com:8002: The fallback is not helping it yeah
[11:49] Ubit Umarov: think now it is more try to render pbr and emulates BP
[11:49] Vincent.Sylvester @hg.zetaworlds.com:8002: Would have to remove BP to perhaps get the speed back up, but that would be taking a nuke to everything, people would riot
[11:50] Ubit Umarov: abyways that is talk for FS support channels, not here
[11:50] Ubit Umarov: specialy if it is a sl issue
[11:50] Ubit Umarov: btw how is dayturn?
[11:50] Gavin.Hird @grid.xmir.org:8002: moving forward
[11:50] Ubit Umarov: nice
[11:51] Gavin.Hird @grid.xmir.org:8002: it logs in to opensim, renders world and all that good stuff, byut HB needs a bit more work and same with avatar baking, but getting there
[11:51] Ubit Umarov: :)
[11:52] Gavin.Hird @grid.xmir.org:8002: to apply opensim features coded 13 years ago to the new codebase is not trival at points but getting there
[11:52] Vincent.Sylvester @hg.zetaworlds.com:8002: Impressive progress considering the time investment of reading LL code, facepalming, considering doom of society and then fixing the mess :)
[11:53] Jagga.Meridith @hg.zetaworlds.com:80: dinosaur's back
[11:53] Gavin.Hird @grid.xmir.org:8002: so maybe next week I'll be here on version 3
[11:54] Gavin.Hird @grid.xmir.org:8002: that thing is running wild in there
[11:54] Jagga.Meridith @hg.zetaworlds.com:80: ouch I just get stepped on
[11:54] Gavin.Hird @grid.xmir.org:8002: here
[11:54] Ubit Umarov: :)
[11:54] Gavin.Hird @grid.xmir.org:8002: :-)
[11:54] Andrew Hellershanks: :)
[11:55] Kayaker.Magic @grid.kitely.com:8002: My Kitely me is not the owner of that dino, so I can't control it now...
[11:56] Gavin.Hird @grid.xmir.org:8002: can derender it
[11:56] Jagga.Meridith @hg.zetaworlds.com:80: but it''s our mascot
[11:56] Andrew Hellershanks: It used to stay out on the sand.
[11:57] Kayaker.Magic @grid.kitely.com:8002: It has code to avoid parcel boundaries.
[11:58] Ubit Umarov: no parcels here but that EEP test one :)
[11:58] Jagga.Meridith @hg.zetaworlds.com:80: So this afternoon I have t explain to my elderly mother "I was in a technical meeting but got stepped on by a dinosaur"
[11:59] Andrew Hellershanks: We are at the top of the hour already. Any other last minute items for today?
[12:00] Kayaker.Magic @grid.kitely.com:8002: Then back to chores in RL for me!
[12:00] Cuga.Rajal @rajal.org:9000: Me too!
[12:01] Jagga.Meridith @hg.zetaworlds.com:80: party time and reading tech manuals
[12:01] Gavin.Hird @grid.xmir.org:8002: Back to waiting for Trump's next crazy announcement
[12:01] Kayaker.Magic @grid.kitely.com:8002: Buy all!
[12:01] Andrew Hellershanks: Alrighty then. I'll call this meeting to a close for today.
[12:01] Cuga.Rajal @rajal.org:9000: *waves Have a great week everybody
[12:01] Ubit Umarov: ok have fun, cya next week
[12:01] Andrew Hellershanks: Thank you all for coming. See you again next week.
