TMP Legacy mesh body

[Kia Kiyori]

3 minutes ago, halebore Aeon said:

I honestly don't think he would, he talks a big game. But doesn't give me the confidence that he knows what he is doing. I look at him like, he made a graphically pretty viewer, that is it.

I guess I've had a very different experience with Niran, so I have total confidence in his knowledge.

[Alyona Su]

49 minutes ago, Nalates Urriah said:

The current Complexity Index and ARC are calc'd using an old algorithm. It was designed for older hardware and a slower Internet. So, they no longer reflect a realistic scale.

Quoted, but response is to general readership:

Viewers seem to have their own say as well. A friend and I working on a project and the subject came up. Both of us using Firestorm, he saw both our complexity scores at 15k higher than I saw them as we stood there, side by side, at the same time. So, how can the same viewer show different score of the same two people at the same time? Rhetorical question, there.

My point is that I don't put a lot of stock in the finessing of Complexity score. Large swaths of score, like between 80k and 150k and 300k, sure. But between 60k and 80k and 125k? Seriously? There is aynahl and there is utterly splitting hairs about a near-moot subject.

I believe complexity as far as lag goes has more to do with your system abilities than it does to so with LL server abilities. Always remember that.

As for the Legacy Body causing any lag? Um, no. The number of agents (Avatars) and the over-meshing of the grid with poorly-created high-poly mesh and outrageously oversized textures create a lot more lag than the difference between Maitreya and Legacy and any other mesh body. As for customer support from the TMP or anyone else, for that matter, many of the larger-business creators rely heavily on their user groups and the crowd-sourced support. The forums are full of "support" or "help" requests and other things where the OP does not communicate very well what their problem is, or it is only one side of the coin that always has two sides.

~Shrugs~

Edited November 11, 2019 by Alyona Su

[Sammy Huntsman]

To be quite frank speaking about Nalates earlier post, I don't think many people will change to BOM. They really don't understand that these system layers have been updated, and a lot of skin companies have been creating updates, with system layers in their packages. A lot of people think, we are gonna go back to the days of 512 by 512 and that it will make their avatar not look as pretty. But maybe if they gave it a try, with the skins they have. They would understand, that it has changed a lot since then.

[Kia Kiyori]

3 minutes ago, Alyona Su said:

As for the Legacy Body causing any lag? Um, no. The number of agents (Avatars) and the over-meshing of the grid with poorly-created high-poly mesh and outrageously oversized textures create a lot more lag than the difference between Maitreya and Legacy and any other mesh body. As for customer support from the TMP or anyone else, for that matter, many of the larger-business creators rely heavily on their user groups and the crowd-sourced support. The forums are full of "support" or "help" requests and other things where the OP does not communicate very well what their problem is, or it is only one side of the coin that always has two sides.

~Shrugs~

Really?

I'd be interested in knowing exactly how many scripts are in the Legacy body because I didn't have any success in getting my script count below 3mb wearing the thing. With Maitreya, yes - easy as pie, but not Legacy.

By all means, argue that numbers aren't numbers and scripts aren't scripts but the bottom line for the end user is that those are the only numbers we have to go on for now - and if you're unable to enter a venue because your script count is too high, then what else are we supposed to go by?

[Sammy Huntsman]

1 minute ago, AnyaJurelle said:

Really?

I'd be interested in knowing exactly how many scripts are in the Legacy body because I didn't have any success in getting my script count below 3mb wearing the thing. With Maitreya, yes - easy as pie, but not Legacy.

By all means, argue that numbers aren't numbers and scripts aren't scripts but the bottom line for the end user is that those are the only numbers we have to go on for now - and if you're unable to enter a venue because your script count is too high, then what else are we supposed to go by?

But here is the thing, it could also be the attachments you wear. Some attachments do take up a lot in script count. So you gotta factor in that. Try going in with just the body, a simple outfit and hair. See if that makes a difference.

[Kia Kiyori]

Just now, halebore Aeon said:

But here is the thing, it could also be the attachments you wear. Some attachments do take up a lot in script count. So you gotta factor in that. Try going in with just the body, a simple outfit and hair. See if that makes a difference.

I don't wear attachments and don't walk around with huds on my screen either. I kill scripts in hair, clothes, jewelry and shoes and when going to an event, I make sure I'm only wearing Maitreya, with everything else on the body being 100% unscripted.

[Sammy Huntsman]

4 minutes ago, AnyaJurelle said:

I don't wear attachments and don't walk around with huds on my screen either. I kill scripts in hair, clothes, jewelry and shoes and when going to an event, I make sure I'm only wearing Maitreya, with everything else on the body being 100% unscripted.

I usually make sure that all my scripts are kept to a minimum. I don't completely get rid of them. Just make sure that the people I get it from, they have really low script counts and their complexity is fairly low.

[Kia Kiyori]

1 minute ago, halebore Aeon said:

I usually make sure that all my scripts are kept to a minimum. I don't completely get rid of them. Just make sure that the people I get it from, they have really low script counts and their complexity is fairly low.

That's great. I kill all mine (create separate folders for colours/styles) but only when going to a concert - am a smidge more relaxed about it otherwise.

[Sammy Huntsman]

Just now, AnyaJurelle said:

That's great. I kill all mine (create separate folders for colours/styles) but only when going to a concert - am a smidge more relaxed about it otherwise.

I was really bad at killing scripts and getting rid of the item entirely, especially if it was the colour I wanted. But this was back in the day, MP was big. But since I can redeliver on MP as well, I guess I can start doing that again. But I find that even then, most of my items are very low on the script count, so there is really no need to go in and delete the scripts.

 

[Orion Pastorelli]

13 minutes ago, AnyaJurelle said:

Really?

I'd be interested in knowing exactly how many scripts are in the Legacy body because I didn't have any success in getting my script count below 3mb wearing the thing. With Maitreya, yes - easy as pie, but not Legacy.

By all means, argue that numbers aren't numbers and scripts aren't scripts but the bottom line for the end user is that those are the only numbers we have to go on for now - and if you're unable to enter a venue because your script count is too high, then what else are we supposed to go by?

I think what she is trying to say is that differences between bodies is negligible. To use your example:  If an Avatar is restricted from a venue because of its high script count, then simply switching from the Legacy to the  Maitreya won't make up the difference. They'll need to examine all their attachments.

[Sammy Huntsman]

3 minutes ago, Orion Pastorelli said:

I think what she is trying to say is that differences between bodies is negligible. To use your example:  If an Avatar is restricted from a venue because of its high script count, then simply switching from the Legacy to the  Maitreya won't make up the difference. They'll need to examine all their attachments.

It's usually an attachment issue, especially when people wear ears and other stuff of that nature. All those things add up over time.

[Kia Kiyori]

2 minutes ago, Orion Pastorelli said:

I think what she is trying to say is that differences between bodies is negligible. To use your example:  If an Avatar is restricted from a venue because of its high script count, then simply switching from the Legacy to the  Maitreya won't make up the difference. They'll need to examine all their attachments.

It's not negligible. 

I can't be convinced that the sun is actually the moon. Legacy has twice the complexity of Maitreya (and heaven only knows how many more scripts if it's dependent on an external server to function at all). 

As I've already said, I don't wear attachments of any description. No. Attachments. Used. At. All.

Look, I can only go by the numbers LL gives end users to go by and I'm trying to say that according to LL's complexity and script counts, Legacy is double the weight of Maitreya.

Use any viewer you want to view bodies side-by-side; Legacy will still be twice the 'complexity' of any other body, by any viewer's count. Same with the script count.

For that reason, I simply don't wear it anymore.

 

[Kia Kiyori]

15 minutes ago, halebore Aeon said:

It's usually an attachment issue, especially when people wear ears and other stuff of that nature. All those things add up over time.

I don't wear attachments. Never have, never will.

[Nalates Urriah]

1 hour ago, halebore Aeon said:

That's the thing, I don't think everyone will not want to have to convert to SLink. She may have done an amazing job, but lately that is all I see people suggesting for a better optimized body. But the issue is, why do we give up our investments and redo our whole look from the ground up for a new body, that is optimized better?

Well, the changing to newer stuff is part of life. Giving up the existing stuff is not mandatory. You don't have to. I still have a load of pre-BOM stuff and I still wear it. Today I am wearing an outfit that isn't Redux/BOM.

The clothes for Physique Original and Redux Original are the same. The Slink bodies did not change size-wise. All my mesh clothes fit. The only place I have a challenge needing new stuff is tats and some underwear... the applier stuff. 

I gave up lots of system stuff when I went to a mesh body. I gave up some other stuff when I went to Bento mesh and more with a Bento mesh head. Going forward I expect to use my appliers less and less. The advantage of lower ACI and scripts is just way too attractive. I don't have to change over. It is a choice.

People don't have to buy Slink. It is just currently the most optimized body I know of. In time I expect to see other designers redo their bodies to eliminate onion skin layers and reduce script load. 

ARCTan is also likely to motivate people to use better optimized stuff when it rolls out.

[Kia Kiyori]

8 minutes ago, Nalates Urriah said:

The advantage of lower ACI and scripts is just way too attractive. I don't have to change over. It is a choice.

Exactly right.

[Alyona Su]

1 hour ago, AnyaJurelle said:

I can't be convinced [...]

No one is trying to convince anyone else of anything. No one is even saying anyone else is wrong. You give your perspective and opinion, I give mine, others give theirs. It's not a debate no matter how much you may try to make it such. My main point was made by @Orion Pastorelli.

Edited November 11, 2019 by Alyona Su

[steeljane42]

9 hours ago, Alyona Su said:

That's nowhere near as bad as most neck seams between most heads and bodies when from different creators. If that micro-"crack" is a dealbreaker (on anything) then so be it

As if legacy had a good neck seam. It doesn't. In fact it has worse one as far as Catwa and Lelutka heads go at least, don't like Genus so didn't bothered to test with their demos, so maybe for their heads it's better (last time I did check genus heads had pretty bad neck seam with maitreya and belleza bodies).

As far as cracks go it's just badly made mesh, nothing else. Especially considering amount of triangles and vertices. If it was on the low end, then yeah, we could justify it by "gotta sacrifice something for the sake of being light". But being the most heavy mesh body on the market... it's just pure laziness or the lack of skill. In fact those issues are visible on the first look, the matter of minutes. There is absolutely no way they didn't seen it during testing before the release, yet it's still out "as is" and how long it's been since release, half a year already? It's pretty critical bug, considering that only use the mesh body has is vanity. We'll see if they ever fix it or it will forever be in "beta" stage, like original deluxe TMP body which remained beta for 5 years.

 

Quote

As for the Legacy Body causing any lag? Um, no. The number of agents (Avatars) and the over-meshing of the grid with poorly-created high-poly mesh and outrageously oversized textures create a lot more lag than the difference between Maitreya and Legacy and any other mesh body.

Yeah, no. Let's see. Matireya: vertices - 151.467, triangles - 253.824, VRAM - 45.076. Legacy: vertices - 787.122, triangles - 823.600, vram - 113.528. Belleza (all female bodies, they have no difference as far as I know): vertices - 398.422, triangles - 521.695, vram - 59.396. Slink is the most light one by far, didn't bothered to check it this time as I don't use it, but it had like 1/3 of Maitreya's. So as far as vertices and triangles go, you can fit 4x people wearing maitreya vs 1 with legacy body. Means your viewer is more likely to choke way sooner, considering all extra mesh with 1024x1024 textures around. Attachment below. Left is maitreya (2nd part is hands, which are separate, so 2x for that), middle is legacy, right is belleza.

And to clarify: I'm certainly not the one to care about the lag or complexity in general, unless it does reach extreme values, like 1m triangles chair. Outside of crowded shopping events where I go only to grab stuff and certainly not to look at others, I don't bother with jellydolls whatsoever, it's always on "unlimited" and my own region is very far from being optimized for low-end PCs, with pretty heavy stuff almost everywhere you can go. So it's not to say "legacy causes the lag and global warming!", I really couldn't care less if it is, just pure numbers. And as far as mesh bodies go, legacy is on the very top, I think signature's gianni was on par with it long ago, but they did optimize it at some point.

 

Edited November 11, 2019 by steeljane42
typos

[Alyona Su]

Still affects actual performance minimally as compared to almost all other factors. But that's okay. You can win. Feel better?

Edited November 11, 2019 by Alyona Su

[Sammy Huntsman]

56 minutes ago, Alyona Su said:

Still affects actual performance minimally as compared to almost all other factors. But that's okay. You can win. Feel better?

I don't think that a lot of people also don't understand. That it's more your client side, so if you don't have the greatest memory, CPU or GPU. No matter what body you wear, it will cause lag. I have a beast of a computer, and I don't encounter lag problems. Unless I am in a really crowded place in SL, than I will start to feel it.

[Alyona Su]

17 minutes ago, halebore Aeon said:

I don't think that a lot of people also don't understand. That it's more your client side, so if you don't have the greatest memory, CPU or GPU. No matter what body you wear, it will cause lag. I have a beast of a computer, and I don't encounter lag problems. Unless I am in a really crowded place in SL, than I will start to feel it.

Precisely.

[Sammy Huntsman]

14 minutes ago, Alyona Su said:

Precisely.

Heck, if your internet is crap you are gonna experience lag too. So if you are playing with really bad internet, and on a toaster, of course you are gonna experience lag.

[steeljane42]

55 minutes ago, halebore Aeon said:

Heck, if your internet is crap you are gonna experience lag too

Don't mix the "two lags" with each other. Bad internet won't cause any FPS drops/bad FPS for the most part, only exception if it was the case of serious packet loss and it's over, so your viewer suddenly need to receive and work through tons of stuff at the same time, which would load piece by piece otherwise. Then again, lots of people actually have no idea what they talk about when they say "lag", it's the case since first gen of mmo games.

 

1 hour ago, halebore Aeon said:

I don't think that a lot of people also don't understand. That it's more your client side, so if you don't have the greatest memory, CPU or GPU. No matter what body you wear, it will cause lag. I have a beast of a computer, and I don't encounter lag problems.

And what also a lot of people don't understand that SL viewers use 1 core/thread for the most part. My current "SL/internet stuff" PC has 9700k@5.1GHz, GTX1070, 32GB RAM as well M2 SSD (samsung 970 evo plus) for that extra fast cache read/write (it's a few times faster than SATA ones in some tasks). Pretty sure it's in the top few % of what people use for SL. I'd change GPU, but CPU bottlenecks it in SL anyway, unless it's BD viewer with cranked up settings for photography. Do I really lag? Nah, not really. But when viewer chokes on heavy meshes then no amount of hardware will save you (or myself) from it, viewer need a rewrite so it would actually utilize current hardware, which is probably not happening knowing LL. Or they'll announce it on SL17B and post an update about it once a year, like with most recent projects. In theory cache rewrite will help a lot to make a breathing room for the CPUs, which first time I heard about 1.5 years ago... maybe 2 already. It still didn't happened.

 

2 hours ago, Alyona Su said:

Still affects actual performance minimally as compared to almost all other factors. But that's okay. You can win. Feel better?

Not really feeling better. Then again, I didn't really argue or did my best to "win" some internet points to begin with. People are free to use/wear whatever they want and it's not really my business. Besides like I already said, I really don't care about "light SL" at all, on the contrary I like pretty and shiny new things. A lot. My recent house that I got uses like 440mb of VRAM alone. Thus the note about about my region being very unfriendly towards low end PCs.

Edited November 12, 2019 by steeljane42
even more typos

[Marianne Little]

7 hours ago, halebore Aeon said:

The issue is not primarily the bodies, it's what people wear to events and totally crowded places. Like do you need to do up your avatar to the nines, when you know that the place is gonna be packed? That is the issue, you can be script and complexity conscious with any body and head. You just gotta take into consideration of what you are wearing as well. Don't be going for all those accessories you own, or that companion pet. Just come as you are, with very minimum on.

I will be "triggered" by companion pets in a popular shopping event. I mean, wtf?

[NiranV Dean]

8 hours ago, AnyaJurelle said:

Imo, the expert in the nitty gritty is NiranV Dean (Black Dragon creator); what he doesn't know, isn't worth knowing - he can shoot the nonsense down in flames without even trying.

I'm no expert, just someone who's very interested in and around SL, specifically pushing it as of lately. There are quite a few things i don't know, starting with precise instructions on how to optimize content like mesh. I can only give you the general idea and tell you what it has to be in the end and possibly some hints how to get there, how to do it in the actual program is beyond me, that is something i assume people who can mesh know much better than i do and i'd also assume telling them what the end result needs to be considered optimized is enough for them to work it out. You usually only see me appear when mentioned nonsense appears and if i know something about it i'll usually chime in.

8 hours ago, halebore Aeon said:

His viewer is nice and all, but it is pretty intensive on machines even machines running amazing graphics. His viewer has nice graphics, but not optimized to be a daily driver. I only use it to take pictures, and even then limit the amount of time I take my pictures to an hour and a half to 3 hours tops. As I know it's pretty CPU and GPU intensive.

This sadly is such nonsense and the reason i'm posting here. Black Dragon is hardly more intensive than the LL Viewer is (in fact some parts need even less considering that i improved and optimized a few things here and there over LL's Viewer), that is unless you pull the settings up beyond what is labeled as "Ultra" or activate the additional graphical features BD offers, even those are considerably low cost even for a low-end card. I have developed Black Dragon for 7 years soon and i've spend almost all of that time with an AMD FX 6200 and a GTX 670, neither of which were high end or even good and i was maintaining a good framerate all around and i've never resorted to disabling Deferred Rendering or toggling off shadows like so many people do to squeeze some fps out. Even a 10 year old Intel in a laptop could beat my FX in raw power in SL (due to SL only using most the first two cores) and the only thing the GTX 670 was lacking was simply more memory because everything SL nowadays eats multi-GB texture memory for no reason.

8 hours ago, AnyaJurelle said:

Yes it is, but it's not his viewer I'm referring to though; it's Niran's knowledge pertaining to Orion's question I was thinking of. If he was let loose at the controls of SL, he'd have it ship shape in no time flat and we'd not have a blessed thing to complain about.

If i was let loose at the controls of SL you'd see a lot of big changes (as much as a team could probably get around to doing in such short timeframe), one more controversal than the last. I wouldn't be so lenient with backwards compatibility (graphical objects from 10 years ago would now either look semi broken or wouldn't work anymore) and i would be high busy setting up guides, documentations and some actual guidelines + limitations on things like the mesh uploader, 2 million triangle avatars wouldn't even exist if i was in control, infact if someone managed to seriously get a 2 million triangle avatar running he'd get a single warning before i'd pull the banhammer. Almost-zero-tolerance-policy. But yes i'd surely start fixing as many old things as possible, while working on expanding on things that i deem overdue that have been an industry standard... namely rebindable controls would be a start, many of my tweaks would also immediately go into the Viewer (such as the real-camera based Depth of Field defaults so people don't get their entire screen completely blurred the moment they enable Depth of Field just to turn it off immediately in disgust, which is a tad shame really)

8 hours ago, halebore Aeon said:

I honestly don't think he would, he talks a big game. But doesn't give me the confidence that he knows what he is doing. I look at him like, he made a graphically pretty viewer, that is it.

I don't know what i'm doing. I just do it. As third party Viewer (and fairly small on top of that) i'm quite free in regards to what i'm allowed to do and what i can get through with. I'd be stupid if i didn't use this to try new things. Sometimes people like them, sometimes they don't but unlike Linden Labs i don't leave something as is for many years unless i either can't fix it, have much more important things to do or am convinced that this is the way it should be. So when people don't like something i either improve based on feedback (if there is any) or continue experimenting (if there isn't) until i find something that works and aligns with my envisioned goal. Sometimes its easy, sometimes it takes months, years even. Niran's Viewer was my first actual dive into programming and it was a very helpful learning experience which i used to start from scratch with a much better foundation in Black Dragon, which still is a huge learning project.

You can't win without trying and most don't even dare, so they already lost before they even began, at least i can say that i died trying.

8 hours ago, AnyaJurelle said:

I guess I've had a very different experience with Niran, so I have total confidence in his knowledge.

People's experience with me vary wildly but could be simplified and grouped into two sets of people: Those who met me anywhere not a forum or blog, like inworld, discord or in games... and those who met me in forums, blogs and other social media. Those who meet me outside of social media generally have a very nice and helpful experience with me as i take the time to adress each and every individual case and help people wherever and whenever i can, if you met me there you will generally see me as a nice and helpful person. Those who meet me on social media platforms will generally hate my guts because i am direct, unrelenting and generally don't mince matters, i will tell you my open opinion without regards to your feelings, most people can't stand this and i can relate, if you met me here you will most likely either have a neutral opinion or more likely a very bad opinion about me because i come off as complete ***** for no apparent reason.

8 hours ago, Nalates Urriah said:

Complexity...

The current Complexity Index and ARC are calc'd using an old algorithm. It was designed for older hardware and a slower Internet. So, they no longer reflect a realistic scale.

The Lindens are working on a revision to the algorithm called ARCTan. When it arrives we will see a difference how we look at our avatars and what causes lag. They expect the numbers to go up... Since larger numbers would be detrimental to Land Impact and prim count on parcel they are considering how to offset that.

You can try the Black Dragon viewer to get an alternative look at ACI using an algorithm Niran (the developer of BD) thinks is more realistic for modern hardware... and that is way debatable.

While it is very debatable how much the polygon count affects performance it is much less debatable how textures and scripts affect performance. Textures are enough of a problem that the Lab went to great lengths to develop BOM. That they even bumped the BOM output from 512x512 to 1024x1024 as an incentive to get people to switch, which I think suggests how much of an issue textures are.

Anyone that flies, sails, or drives vehicles in SL understands the problems from heavy script weight. There is work going on now to improve script performance and lighten the load on the region servers.

As to optimizing a body, I think Slink has done an amazing job with Redux. The ACI drops about 10k and the script load by a a couple of MB. AFAIK it is the only body completely redesigned for BOM.

A few things: As far as i remember the lindens designed the old ARC to include as much different hardware as possible, ranging from old to new, this is, what i think was their biggest mistake and made ARC ultimately useless.

Second: Their revision of ARC is just as useless as it is right now, i'd highly recommend not holding out for it or having any hopes, Oz has already confirmed that they are not intending to punish or "shame" bad content in any way, this would imply that most if not all current biggest offenders of broken ARC values are still going to be broken with the new ARC values which ultimately makes them useless, no matter if they are simply higher, the very fact that they "look to remedy the issue of possibly higher values" should worry you. They are not looking to offer you realistic or informative values.

To put this into perspective, the reason i even started disregarding the warnings of Oz in the code and started modifying the ARC calculation is because if a 2+ million triangle rigged mesh monster with more than twice the memory usage than my own (i think 150k polys, most of it simple flexi prim hair and a bunch of sculpts which are locked to 64x64 points) can have half my ARC, something is super off and no one can tell me that this is in any way realistic or even remotely accurate. When i spend my entire evening optimizing my avatar i can guarantee you that you can easily put my avatar a few times into the same place one of these multi-million monsters are and still using less resources.

Remember that the very reason ArcTan has even been a thing (as far as i can tell) is because i relentlessly bashed ARC and made the entire meeting talk about how ***** the current ARC is before Oz's arrival on an open source meeting which made him finally listen. Seeing that ArcTan is going to be wasted and biased again towards "a wide range" of hardware is a personal disappointment. I feel like going all these lengths and having to deal with pissing people off with my complains about everything being broken is wasted here, i don't do this for fun. I know i complain about everything but i have legitimate concerns and i'm not taking all this flak so LL just tosses it into the fire for the sake of "not shaming bad content".

Which brings me to three: I don't "think" its more realistic, i'm extremely convinced that it is more accurate (although the numbers are quite high but for simplicity and transparency remained like this for so long) than anything that will possibly ever come out of whatever LL is cooking up if they don't drop their "protect-all-content" mentality. Bad content needs shaming, bad content needs punishment, nothing is going to change if we ignore it and continue looking through pink glasses. LL NEEDS to intervene there, they absolutely MUST put limitations on this otherwise they are just going on a wild goose chase forever. I can tell you i will continue cracking skulls when ArcTan arrives until they give up and actually do something useful about it.

Four: How much of an impact polygons are obviously depends on the hardware, some hardware does certain things faster than others, ultimately we can agree that polygons aren't the sole cause for all our performance problems but they CERTAINLY are a big part of it as a lot of graphical rendering hinges on it. Polygon complexity carries its impact upwards and gets worse. While a million polygons might not directly kill your framerate (because our hardware are monster calculators and can deal will billions of these) it doesn't mean that they won't create a bottleneck somewhere. Shadows for instance, most people recommend disabling them, you know why? Polygons. Your million polygons don't impact someone without shadows as much as they do to someone with it enabled because that person basically has to render them twice (if not more). When rendering a particular item multiple times, impacts quickly multiply and get out of hand. Textures on the other hand, i'm not a graphics expert, i can only tell you from what i've seen with my own eyes but i'm pretty convinced that textures themselves have little to no impact on your framerate itself, allocating, decoding and loading them is what makes the big hit in performance, this hit however is temporary. When everything is fully loaded your framerate will stabilize because textures have been loaded into memory and just need to be rendered and possibly transformed on screen most of which should be the GPU's work anyway. Unlike the CPU the GPU is much better utilized in SL and more raw power on a GPU can be used although mostly just as "buffer" for when you turn up GPU related rendering features (depth of field for instance), which is why i say a GTX 670 was far more than enough (when it died i even used my old GTX 460 for a few days and... you'll laugh but even the 460 produced the same framerates, GPU requirements are reeeeally low in SL).

BUT. Textures cause a different problem. Texture trashing, when the alloted memory for SL is full, the Viewer will attempt to clear out textures in an attempt to make space for "more important" textures, if there are none it will result in what is commonly called "Texture Trashing", e.g your textures will repeatedly go blurry and clear. One solution is obvious: toss more memory at it. Another... and the real solution is optimizing textures. Reduce texture sizes to a minimum, a necklace doesn't need 128 1024x1024 textures for every chain a different version of the same texture. A single 64x64 or 128x128 texture is way more than enough if you reuse it for all chain parts and it looks the same but in the extreme and unlikely case of 128 1024x1024 textures would save you 512mb of texture memory minimum, taking into account that each texture is 4MB (5MB with alpha if you believe the texture console to be correct). This texture trashing is also useful for another thing other than making space for newer textures, it prevents what games often don't do, overloading your texture memory. Should this ever happen (and you can test this by enabling fullrestextures) you will see your framerate drop into completely unusable rates because (as mentioned before) the GPU starts doing whats possibly the worst (and slowest) thing to do, it deallocates textures to make space for other textures, allocating the new ones, repeatedly allocating and deallocating textures will bring any GPU to its knees, no matter which and that's something you want to prevent with all means necessary.

[Kia Kiyori]

8 hours ago, Alyona Su said:

No one is trying to convince anyone else of anything. No one is even saying anyone else is wrong. You give your perspective and opinion, I give mine, others give theirs. It's not a debate no matter how much you may try to make it such. My main point was made by @Orion Pastorelli.

Alyona, you picked half a sentenced, ignored the rest and immediately mounted a high horse.

If you'll please refrain from telling me what I do and don't see, I'll explain it again, as simply as possible.

Legacy has twice the complexity of Maitreya - going by my version of the latest FS Viewer - with heaven only knows how many more scripts than other bodes.

In a nutshell, what amounts to 20k on Maitreya will usually be found to be around 40k+ on Legacy.

And I struggled to get my script count down below 3mb wearing Legacy. It usually hovered over 3mb + so that I gave up and returned to Maitreya.

I believe you said that different viewers will give different numbers so that the numbers I was seeing don't exist/aren't valid/weren't real. 

That isn't actually a valid argument unless you can prove that the numbers given by other viewers consistently demonstrate the bottom line:  whether Legacy is half the complexity of Maitreya, or double.

Let's ignore the fact that I haven't come across someone with an identical version of FS whose numbers have disagreed with mine - no matter what numbers other viewers give, I think they'll still demonstrate that Legacy is double Maitreya's complexity.

That's all I tried to say.