Jump to content
Rohan Dockal

Why is Second Life so laggy now compared to the past?

Recommended Posts

2 hours ago, TiaShaw said:

4gb graphics Vram is barely enough even at 720p... 32gb ram only helps IF you log multiple accounts at the same time. 4.2gb per instance of firestorm. I have tested probably more than i have actually just ran around and explored.a Gtx1060 6gb is the bare minimum i would suggest to anyone. For 1080p  for a good result.  Just my opinion. I bought and tested everything in graphics from  gtx 570 up to my twin gtx1080ti’s (2nd card is dedicated stream encoder) with a intel I9-9900k locked at 5ghz. I have spent alot of money in the wrong places. Sl doesnt care if you have more than 4cores but the faster the better. 

SL and other MMOs run just fine on my single GTX 1050 2GB.   ¯\_(ツ)_/¯

Share this post


Link to post
Share on other sites
9 hours ago, TiaShaw said:

32gb ram only helps IF you log multiple accounts at the same time

Also, for a RAM Disk/cache. Which, I hope most with such resources consider. I've used a RAM Disk for caching since around 2012. I recently upgraded from 32GB to 48GB--have a 20GB RAM Disk, and set up a Squid Proxy Cache. It's quite silly, how fast loading textures is. The leftover 28GB is more than enough for anything/everything else that I do (gaming, audio/music production and virtual machines). 

Share this post


Link to post
Share on other sites
On 8/6/2019 at 3:29 AM, Extrude Ragu said:

...

It's not the technology, it's the economics.

Commercial creators are heavily incentivized to cram as much detail as possible, into as low an LI as possible. They do this by 'cheating' the Level of Detail System by supplying a single tri to Medium, Low and Lowest Models(A mesh has different levels of detail based on zoom level) to get lower LI, whilst still having poorly optimized highest model and demanding users set their LoD factor up to max or even changing debug settings in their viewer.

With the right economics, mesh is less laggy than Prims

...

What needs to be done is we need to find a tangible way to incentivize commercially motivated creators to optimize their products. The average consumer generally does not consider efficiency in their purchase, only how good the product looks. Personally I'm not 100% sure what we could do, but I know this is the area that needs solving.

this question comes up now and again.  What would LL need to do to incentivize creators to use all LOD levels in the manner intended ?

my thought on this is that every level has a minimum baseline cost, baseline being some percentage of the top level, a cost charged whether used or not. The thought is that when a LOD level is going to be charged for then the creator will use it in the manner intended

 

Share this post


Link to post
Share on other sites
9 hours ago, Mollymews said:

this question comes up now and again.  What would LL need to do to incentivize creators to use all LOD levels in the manner intended ?

my thought on this is that every level has a minimum baseline cost, baseline being some percentage of the top level, a cost charged whether used or not. The thought is that when a LOD level is going to be charged for then the creator will use it in the manner intended

 

That's how animesh handles costs.

Most SL users aren't tech savy, might have never seen a texture map or don't even know that lod models exist.

Most people want things that look pretty, the problem is how to showcase problems in the products they purchase, and how to make it clear that those problems are not bugs/errors brought up by SL itself, but mistakes and/or intentional shortcuts taken by the creator.

Share this post


Link to post
Share on other sites
1 hour ago, Kyrah Abattoir said:

That's how animesh handles costs.

For all practical purposes it's how all objects are treated. It's just that the baseline for other objects is too low. TH animesh LoD baseline on the other hand, seems to be way too high to encourage real optimisation but it may be neccessary to compensate for other flaws in the system.

 

27 minutes ago, Kyrah Abattoir said:

Most people want things that look pretty, the problem is how to showcase problems in the products they purchase,

How about 3D displays of the actual model for sale on MP?

 

29 minutes ago, Kyrah Abattoir said:

and how to make it clear that those problems are not bugs/errors brought up by SL itself, but mistakes and/or intentional shortcuts taken by the creator.

I would say the problem is to a large degree caused by four important mistakes LL made in the past.

One is the client-side adjustable LoD factor. It overrides the little resource we have of resource management functions and without a reliable reference point, it's very hard for cotnent creators to optimise effectively. I'm not sure why the LoD factor exists at all. From something Grumpity once said, I got the impression that it was to allow full high res rendering for still photos but I may have misunderstood her.

The second error was to not include an object based LoD factor - at least for mesh. We could save a lot of lag if it was possible to strengthen the LoD of individual items (of course also adjusting their LI accordingly) rather than for entire scenes. This is a crucial function in most modern game engines and I can't understand why LL chose to leave it out.

The third mistake was their choice of algorithm for creating LoD models in the uploader - and also to some degree the way it was implemented and presented. The one they chose, GLOD (Geometric Level Of Detail) is built for speed and nothing but speed. The idea behind it was to generate LoD on the fly. This quality is of course totally irrelevant for the uploader where if hardly matters if it takes a few extra milliseconds to generate the models and it means GLOD can't possibly do the thorough analyzis algorithms made for pre-generated models do.

The fourth mistake is LL's consistent inability to deliver relevant and transparent documentation.

I have to add that I don't really blame the current LL for any of this. These are all sins of the past and none of the people responsible for them work here anymore. Even so, one thing SL needs to become a lean and mean 3D paltform for the next decade or two, is a thorough code cleaning. LL is perfectly aware of this, Ebbe Linden has mentioned it in a tleast one interview. But everybody from the CEO to the lowest code monkey hate code cleaning. It takes a lot of hard, dull and difficult work, all the manpower it needs costs a lot and the result doesn't show up in the bottom line right away.

  • Like 1

Share this post


Link to post
Share on other sites

Most sims don't lag for me, and if they do I turn down my settings and things run considerably better. I don't even have an elite gaming pc. It's very basic entry level. But yeah, I'd say for me SL is less laggy these days because even if my gaming pc is entry level it's still better than ones I've the in the past, and I think creators are more aware of what makes sims and avatars lag like crazy these days and try to be mindful. Remember that blingy jewelry fad? I think everyone suffered at least one casualty from those and sims would grid to a halt if more than one person showed up with them. 

Share this post


Link to post
Share on other sites
22 hours ago, Kweopi said:

From Firestorm help pages: "One aspect of SL that is often underestimated is the importance of your network bandwidth setting. Having this too high or too low can lead to all kinds of problems, from simple TP failure to being disconnected - among other things. So it is advisable to not rely on the bandwidth stated by your provider, but to check for yourself.

SpeedTest will give some indication of the quality of your connection. But another very useful indication is ping and packet loss.You can determine these from within the viewer by accessing the statistics bar (press Shift-Ctrl-1). Look for Packet Loss and Ping SIM. Ideally, these should be 0.0% and under 150ms, respectively. The stats bar gives data based on your present location in the SL grid.1)Why the location matters. Second Life has data centers in Phoenix, Arizona, USA. This is why you need to select that city when measuring your bandwidth. If you live far away and do the default test, it will be to a city close to you, which will not give a good indication of your connection to Second Life servers.2)NOTE: The city recommended is not exactly where SL has servers, but is close enough to serve as a guide in determining a good value for bandwidth."

i posted that in other section too : i have a lousy i3 with 2gb vram and 8 gb ram. Constantly i got 100 fps or more ( around 150 in totaly empty sims - like animesh test ones, and  around 20 in crowded events. it might not be the only thing who save the day net helps a lot

net speed.png

yet another reason why Linden has a hard to keeping new residents lol. You almost need to be a computer scientists just to log onto SL and use it properly! lol

But yeah Ive checked the connection back to Tuscon

  • Confused 1

Share this post


Link to post
Share on other sites
5 hours ago, ChinRey said:

the result doesn't show up in the bottom line right away.

There won't be a result if it doesn't get done. If anything can kill off SL that will be the one thing.

Share this post


Link to post
Share on other sites

Core 0-10 <5%, core 11 100%. Maybe I’m being unrealistic and SL’s fundamental nature is incompatible with significant multi-threading, but it seems like some optimization for CPUs made in the last decade is appropriate. I can’t believe that would be more difficult than dealing with almost 20 years of bad content and creators throwing more garbage onto that pile every day.

Share this post


Link to post
Share on other sites
On 8/1/2019 at 9:23 PM, Kyrah Abattoir said:

I'm not cherry picking. You are dead set on the idea that some things are simply not your problem and this irrational "zero sum game" that you won't be able to enjoy SL to its fullest with your fancy hardware if creators where forced to be more reasonable.

My position is that SL doesn't incentivize good content, there is no real mechanic in place that forces creators into the balancing act that should be expected from them.

 

Ay, there's the rub. You are dead set on the idea that creators needs to be forced 'into the balancing act,' that they should be reasonable, and restrain themselves, resource-wise. And I've even heard arguments, in this thread, that are tantamount to "Creators simply don't care about restrictions, and purposely try and use high-res textures, because they just want to sell a house that looks better than that of the competition." To which I can only say: duh! See, there's an sich absolutely nothing wrong with a creator wanting to sell me a house that's better looking than that of the competition. And, in fact, I want to BUY said home! :) If one home has all edgy curves, and the other doesn't, I'll go for the one with more polygons. And this is neither sleazy on the creators' part, nor on mine.

You are simply confusing two separate issues: 'pretty' versus 'costly' (resource-wise). 'Costly' may be a nasty side-effect of wanting 'pretty,' but it's really a rather unreasonable position to blame creators for wanting to sell me 'pretty' (or blaming me, for that matter, for wanting to buy 'pretty'). If better looking homes require an overall increase in resources, well, then so be it! In fact, one could cogently argue, that 'the best looking stuff' is not just okay, but the highest goal for any virtual world. If more resources are required to get there, then the onus is on LL to provide/optimize, and not on the creators to restrain themselves (all the latter accomplishes anyway, is that it invites laziness on LL's end).

I'm not arguing the case that creators should be sloppy, or lazy themselves. If only X polygons are required to make a curve look good, then it's pointless -- and simply wrong -- to use X times 10. But a gorgeously baked house, that uses 1024 textures (for its main parts), will simply look better than an old one using 512 textures. The creators know it, and I know it. And then I will buy the 1024 one. Shoot me.

  • Haha 1

Share this post


Link to post
Share on other sites
7 minutes ago, kiramanell said:

You are dead set on the idea that creators needs to be forced 'into the balancing act,' that they should be reasonable, and restrain themselves, resource-wise. And I've even heard arguments, in this thread, that are tantamount to "Creators simply don't care about restrictions, and purposely try and use high-res textures, because they just want to sell a house that looks better than that of the competition."

I have nothing against "pretty", we all want that. I would praise everyone who can do "pretty" if it wasn't drowned by the inner voice screaming "WHAT WHERE THEY THINKING?".

And yeah a good chunk of them don't care because it's not about 'good' it's about 'fast', and it's also about making the most money from the least amount of work.

8 minutes ago, kiramanell said:

If more resources are required to get there, then the onus is on LL to provide/optimize, and not on the creators to restrain themselves (all the latter accomplishes anyway, is that it invites laziness on LL's end). 

You're not new in this topic and despite all the explanations that where given by me, and people who are far better at this than me you are still blaming LindenLab?

And what if they can't?
What if there is no solution that they can implement because your fantasy of making SL look like a Pixart movie, and doing that at a reasonable framerate is simply not possible?

Disney is still using render farms that grills top of the line hardware 24/7, to produce a single frame of movie, in a few HOURS.

 

The only people whose content would become worse with sane limitations are those who refuse to get better at what they are supposed to be good at. Everyone else will do just fine.

  • Like 1

Share this post


Link to post
Share on other sites
1 minute ago, Kyrah Abattoir said:

And yeah a good chunk of them don't care because it's not about 'good' it's about 'fast', and it's also about making the most money from the least amount of work.

You're not new in this topic and despite all the explanations that where given by me, and people who are far better at this than me you are still blaming LindenLab?

 

I did wut?! No, all I said was, that asking creators to restrain themselves invites laziness on LL's end (after all, what better way for LL to not try and optimize stuff, when they got ppl telling each other to go low on resources?).

But yes, there's plenty room for LL to improve on. Textures are only a small factor contributing to possible lag. Thousands of idle (!) scripts alone, for instance, are reportedly enough to use up as good as all script resources on a sim. Linden has even more-or-less acknowledged this issue -- and they're allegedly looking to it (with low priority, far as I understood). Greatest drain on SL resources, IMHO, is not the use of high-res textures, but Sansar.

Share this post


Link to post
Share on other sites
Posted (edited)
43 minutes ago, kiramanell said:

If better looking homes require an overall increase in resources, well, then so be it!

Better looking content do not generally require an overall increase in resources. Nearly all the content I've seen in SL have been obviously loaded down with serious amounts of "dead weight" that can be eliminated without altering the appearance or performance in any way.

I made an example of this about a year ago. I looked at a fairly typical SL mesh (a dead tree in this case) and made a very similar one using some fairly basic optimisation techniques. I had no problems cutting the land impact down to a third and the render weight down to half of the original. Those are fairly typical figures and if all SL content had been made with the fairly moderate level of optimisation I went for in this case, we'd save a lot of gpu and bandwidth load with no negative side effects whatsoever.

LittleMe Jewell was kind enough to post a picture of the two trees side by side. They are not exactly identical of course - I had to respect the original creator's IP rights - but I hope you agree they are simialr eough to be comparable. The one to the left is mine, the one to the right the original. https://community.secondlife.com/forums/topic/428493-current-premium-gift-camp-fire-tent/?tab=comments#comment-1801775

I still don't like to blame content creators for the lag though. Most of the ones I've met - and that includes a few of the "worst offenders" - have tended to be quite positive to optimisation. The main problems are the lack of relevant information and the abundance of myths and misinformation.

Edited by ChinRey
  • Thanks 1

Share this post


Link to post
Share on other sites
3 minutes ago, ChinRey said:

Better looking content do not generally require an overall increase in resources. Nearly all the content I've seen in SL have been obviously loaded down with serious amount of "dead weight" that can be eliminated without altering the appearance or performance in any way.

I made an example of this about a year ago. I looked at a fairly typical SL mesh (a dead tree in this case) and made a very similar one using some fairly basic optimisation techniques. I had no problems cutting the land impact down to a third and the render weight down to half of the original. Those are fairly typical figures and if all SL content had been made with the fairly moderate level of optimisation I went for in this case, we'd save a lot of gpu and bandwidth load with no negative side effects whatsoever.

LittleMe Jewell was kind enough to post a picture of the two trees side by side. They are not exactly identical of course - I had to respect the original creator's IP rights - but I hope you agree they are simialr eough to be comparable. The one to the left is mine, the one to the right the original. https://community.secondlife.com/forums/topic/428493-current-premium-gift-camp-fire-tent/?tab=comments#comment-1801775

I still don't like to blame content creators for the lag though. Most of the ones I've met - and that includes a few of the "worst offenders" - have tended to be quite positive to optimisation. The main problems are the lack of relevant information and the abundance of myths and misinformation.

And neither of them are something I would consider good enough to use

  • Haha 1

Share this post


Link to post
Share on other sites
1 minute ago, ChinRey said:

Better looking content do not generally require an overall increase in resources. Nearly all the content I've seen in SL have been obviously loaded down with serious amount of "dead weight" that can be eliminated without altering the appearance or performance in any way.

I made an example of this about a year ago. I looked at a fairly typical SL mesh (a dead tree in this case) and made a very similar one using some fairly basic optimisation techniques. I had no problems cutting the land impact down to a third and the render weight down to half of the original. Those are fairly typical figures and if all SL content had been made with the fairly moderate level of optimisation I went for in this case, we'd save a lot of gpu and bandwidth load with no negative side effects whatsoever.

LittleMe Jewell was kind enough to post a picture of the two trees side by side. They are not exactly identical of course - I had to respect the original creator's IP rights - but I hope you agree they are simialr eough to be comparable. The one to the left is mine, the one to the right the original. https://community.secondlife.com/forums/topic/428493-current-premium-gift-camp-fire-tent/?tab=comments#comment-1801775

I still don't like to blame content creators for the lag though. Most of the ones I've met - and that includes a few of the "worst offenders" - have tended to be quite positive to optimisation. The main problems are the lack of relevant information and the abundance of myths and misinformation.

 

For the record, I completly agree with you. :) As I said in the last paragraph of my previous post, using more resources than is strictly required for (almost) the same visual quality, is simply wrong. So, in that sense, I can agree with everyone else, that creators should strive to seek said optimum: best looks for the least amount of resources.

Share this post


Link to post
Share on other sites
1 minute ago, KanryDrago said:

And neither of them are something I would consider good enough to use

That may well be. I obviously wasn't going to spend the time needed for a more elaborate build just to prove the point. It does apply to SL meshes at all levels of complexity though.

  • Like 1

Share this post


Link to post
Share on other sites
Just now, ChinRey said:

That may well be. I obviously wasn't going to spend the time needed for a more elaborate build just to prove the point. It does apply to SL meshes at all levels of complexity though.

Yes I wasnt expecting you to put a lot of time into it and to be fair it may just be the way they were photographed. The point I was trying to make however is merely that when you can show this reduction with something that people consider to be usable then it would carry more weight.

For me at least I find SL runs fine virtually everywhere with everything like shadows turned on and alm at ultra. Very occasionally at a hugely crowded sim I turn down to high, maybe once a month. Most of those that I see complain seem to have machines that would have been considered outdated 5 years ago. While I accept there is a balance between pitching sl at machines that are less able I personally don't think that a ten year old machine struggling so we should do something to make it work a reasonable balance. At some point you have to turn round to people and say....well yes its going to run like crap with that and the answer is upgrade your machine

  • Like 1
  • Haha 1

Share this post


Link to post
Share on other sites
12 hours ago, Kyrah Abattoir said:

Most SL users aren't tech savy, might have never seen a texture map or don't even know that lod models exist.

i think that more information available at time of purchase is always a good thing

i think also that most people understand LI.  While the way in which LI is calculated can always be improved, most people I think are comfortable that there is an LI measure. In the same way we are comfortable with RL things like Energy Rating of household appliances and other products.  How Energy Rating is calculated isn't something we, the larger body of consumers, particularly concern ourselves about, just that there is some kind of easily understood way for us to compare similar products

LI serves this purpose - a simple easily understood measure for/by consumers

were Linden to retrospectively apply a more accurate LI measure based on min. LOD costs, texture size, etc, then it will push a lot of current content up in LI.  Which will cause residents some distress

what could Linden do to counter/ameliorate this distress ?  The simple answer would be to increase LI capacity.  This was done before when the current LI measure was introduced, it could be done again

a less simple answer might be that the LI measure does not change when a mesh object is resized.  From a resident perspective this is something that I think would go down well

Share this post


Link to post
Share on other sites
5 hours ago, KanryDrago said:

Yes I wasnt expecting you to put a lot of time into it and to be fair it may just be the way they were photographed. The point I was trying to make however is merely that when you can show this reduction with something that people consider to be usable then it would carry more weight.

For me at least I find SL runs fine virtually everywhere with everything like shadows turned on and alm at ultra. Very occasionally at a hugely crowded sim I turn down to high, maybe once a month. Most of those that I see complain seem to have machines that would have been considered outdated 5 years ago. While I accept there is a balance between pitching sl at machines that are less able I personally don't think that a ten year old machine struggling so we should do something to make it work a reasonable balance. At some point you have to turn round to people and say....well yes its going to run like crap with that and the answer is upgrade your machine

I think the issue facing Linden is people aren't buying "machines" they have laptops and tablets nowadays. Except for hard core gamers who probably aren't likely that interested in SL anyways. And i think it would be hard for someone to justify just because they like SL to spend 2 or 3 thousand (or more) on a machine powerful to enjoy SL in all its glory

Share this post


Link to post
Share on other sites
4 minutes ago, Jackson Redstar said:

I think the issue facing Linden is people aren't buying "machines" they have laptops and tablets nowadays. Except for hard core gamers who probably aren't likely that interested in SL anyways. And i think it would be hard for someone to justify just because they like SL to spend 2 or 3 thousand (or more) on a machine powerful to enjoy SL in all its glory

well see thats where you exaggerate, my machine is 4 years old, it cost 900 at the time. You can pick up the same spec now for much less and it runs sl perfectly fine

Share this post


Link to post
Share on other sites

A halfway decent laptop should be able to run SL fine. It can't because of the content. Here's an article I wrote comparing the performance of an old desktop and a relatively modern laptop in some decently optimized environments.

I have been optimizing contents and environments in SL for over a decade. I get better performance when in those environments than people with much more powerful computers have ever experienced in SL. And I'm able to run SL at higher graphics settings than them to boot. You know how SL freezes up when you turn your camera around? Yeah, that almost never happens to me. You know how things take forever to rez? My environments load lightning fast every time. You know how your computer says you're getting X framerates, but in reality there's lots of stuttering? I solved that years ago. You know texture thrashing? When textures keep going blurry, reloading, then going blurry again over and over? Apparently that happens to most SL users, but I almost never see it. When I do see it, it's generally pretty easy for me to fix it.

Your mileage may vary, of course, because there's three main issues with content in SL.

  • Textures. Textures are by far the biggest performance killer in SL right now. You only have so much video memory, and everything you see on screen needs to be held in your video memory, and once the content onscreen exceeds that framerates plummet. Textures are why your computer freezes when you spin your camera around too fast. Too many textures is why texture thrashing occurs. I've seen sims using upwards of 10GB worth of textures. And a lot of these textures are wasteful, full of blank, unused space that is just eating up memory and not contributing at all to how things look on screen.
  • High-poly content. A lot of people are uploading meshes that have no business being rendered realtime. They typically get around the high LI cost such an object should have by gutting the LOD models and telling their customers to go into the SL debug settings and push Object Detail absurdly high. Well made content with proper LOD models will look good even if your Object Detail is set around 3, or even 2. And it's easy to make good LOD models that will get you a smaller LI cost when you import to SL. When you crank Object Detail up higher than 3 you are making SL laggier for yourself.
  • Avatars. The only reason I don't rank avatars higher than textures as far as SL's biggest performance killer is that the problem with avatars is a combination of the above two points, pushed to an extreme. For all intents and purposes, there's no real limits on how much you can plaster onto an avatar. I've seen avatars wearing over a GB of textures and more polygons than the sim they're residing in has rezzed altogether. Worse yet, rigged mesh only ever displays the higher LOD, making avatars even more render intensive. And if your avatar is laggy, everywhere you go you're bringing your own lag with you to share with everyone else.

The issue facing LL is that they never hired a content creation TD who understands all this, and can communicate with the viewer and server devs on what is needed to combat it. Or if they have, LL clearly does not listen to them. We only have one tool to manage these problems, jelly dolls, and that doesn't even work very well because it ignores texture use almost entirely. Content creators won't suddenly develop a conscience and change their bad habits all on their own. Most of them have no idea how their content is contributing to the lag because none of the tools SL provides gives them any feedback on that. Third party tools to help people manage texture use have been developed, but third party viewers have been reluctant to adopt them. (According to one TPV dev, they absolutely need a full GB worth of textures on their avatar or they can't make their avatar look good.)

So, what can LL do? For one thing they can give us better tools to manage resource use.

  • Make it a standard feature to see how much VRAM our avatars are using, and how much each attachment is using.
  • Make it so VRAM use is displayed on marketplace listings just like the Land Impact. Then incentivize people posting that information by prioritizing those listings that do in search results.
  • Give people the ability to jelly doll avatars based on VRAM use and polygon count. Turn it on by default and when people go to turn it off, give them a message explaining that the avatars they are choosing to render absolutely will hurt their framerates.
  • When someone goes to upload a mesh model with an obscene amount of textures, jack up the upload cost and explain to them why using so many textures will hurt performance. Share tips on how they can reduce the texture use.
  • When someone dresses up their avatar in over 100MB worth of textures, pop up reminders that they are dragging their performance down by doing so. Give them tips on how to reduce VRAM use as well as pointing out attachments that use an excessive amount of textures.
  • When someone is experiencing performance issues because the environment is full of badly made content, tell them why their performance is so bad. You can keep it vague, only telling them there are multiple items in the scene that are dragging their framerates down. But if they own the items, they could get more info, such as specific objects in the scene that are using excessive resources.

These are all just ideas off the top of my head, LL of course should put a lot of thought into anything like this. They haven't done anything in the nearly 20 years SL has been around, they can afford to take the time to do it right if they ever do decide to try and tackle this problem.

  • Like 4

Share this post


Link to post
Share on other sites
Posted (edited)
6 hours ago, Mollymews said:

In the same way we are comfortable with RL things like Energy Rating of household appliances and other products.

Why don't LL introduce a star rating to the marketplace. Surely such a thing would be easy to implement.

Set it up so that when listing an item on marketplace it is a requirement (cannot finalise the listing without the info added) to input the download, physics and display weights of all objects. This data is easy to find in the more info section of the build menu. Upon entering the data in the listing have a little '?' mouse over button next to the weight input section telling creators how to find this info if they don't know where it is and why it is important. Then have this information (via an algorithm LL can set themselves), on the listing being accepted, converted to a simple star rating (or even a green bar rating system for each weight - longer bar = better optimisation) which is displayed graphically (like energy ratings) under the Land Impact or price section.

One star for bad weights and up to 6 or even 10 stars for well optimised and good weights.  This would not only allow users to see what items will be impacting them the most and shop accordingly as far as optimisation goes but also mean it is in the interest of creators to ensure that they reach as many stars as possible to keep competitive. It also has the additional side effect of showing and educating both user and creator of how to make well optimised content and also why it is necessary.

As far as existing content listed on the marketplace if this is implemented; just have it so that all existing listed items have something stating it was created pre the star rating. This would mean all current listings stay as is, however have the caveat of users buying at own risk. It would also give incentive for creators to go back and edit the listing and update the weights to have the star rating show.

It could even go further (considering marketplace will have the data with the above data needing to be added) whereby they introduce a wardrobe feature to the marketplace where you can select multiple items you are buying via tick box in your cart to see how much the total avatar render cost would be. I.e., if I am looking to buy a body, hair, skin, shirt, pants and shoes in one cart transaction, allow me to tick which items I want to calculate the ARC for and it calculates the ARC value of all those items or outfit in the cart prior to pressing the buy button. Kind of like how you can check online in RL what the estimated postage would be before purchasing.

Certainly not a fix for everything but a step in the right direction of trying to educate users easily, visually and most of all prior to it being purchased and rezzed in world.

Edited by Drayke Newall
  • Thanks 1

Share this post


Link to post
Share on other sites
Posted (edited)

To be honest they should require from creators to enter the following:

  • Tri count used by each lod.
  • Number of textures and their size at any given time.
  • Number of scripts.

I like your idea of star rating. But if an item has no rating just put a big exclanation mark "the creator of this item has not supplied rating informations, buyer beware."

However I suspect showing the tricount of each lof won't help depending of the calculation method, showing the texture area might, but it also doesn't account for whether textures might be recycled across multiple items. Likewise, script count, while a useful indication, doesn't tell the full story.

Edited by Kyrah Abattoir
  • Thanks 1

Share this post


Link to post
Share on other sites
12 hours ago, Lyssa Greymoon said:

Core 0-10 <5%, core 11 100%. Maybe I’m being unrealistic and SL’s fundamental nature is incompatible with significant multi-threading, but it seems like some optimization for CPUs made in the last decade is appropriate. I can’t believe that would be more difficult than dealing with almost 20 years of bad content and creators throwing more garbage onto that pile every day.

There is the potential for a lot of multi threading in the SL client .. but it can't really be retrofitted.

As things stand right now, you can use all the threads and cores to do all the things, but then you have to get all the data back onto the main thread so it can be passed to the GPU. This will happily erase all the extra performance and then some. Net result, trying to do threading makes it slower.

This will help explain the problem ...

Image result for ccpu cores meme

 

Multi threading a thousand tiny jobs that all have to sync up 60 times a second .. exceptionally difficult. - So put all those guys in that one hole and have them share one shovel.

Multi threading a huge job, like rendering a video is easy. There is no immediate time pressure, break the job up and send it to all the cores you can.  - Everyone has a their own shovel and their own hole with no overlap.

SL would be very suited to a brand new vulkan based pipeline, but that's starting the SL client from scratch for all practical purposes. Maybe Oz will find that graphics dev.

  • Like 2
  • Thanks 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...