Jump to content

Catznip 12.1 (+Animesh)


Coffee Pancake
 Share

You are about to reply to a thread that has been inactive for 1834 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

 

8 hours ago, Qie Niangao said:

One thing about Catznip: When enabling "Allow Multiple Viewers" it fusses "Running multiple Second Life viewers is not supported. It can lead to texture cache collisions, corruption and degraded visuals and performance." It does it anyway though. I run multiple viewers often, although I've taken to running a different viewer for each account: while I'm on Catznip, one alt might run the Linden viewer, or another runs Firestorm. I certainly have violated that "different viewer" rule... and, historically, I've had cache corruption enough times that I've wondered what the anti-cache-clearing zealots are smoking, but I never particularly associated those corruptions with running multiple instances of the same viewer (as opposed to network disruptions). But maybe.

 

A SL viewer is constantly reading and writing files to the cache. If you have multiple viewers in the same cache it's only a matter of time before they trip over each others feet and demand your OS do the impossible . . . which is by todays standards is performing a very complicated card shuffling act involving multiple hands, decks and ghost cards.

Cache corruption happens less than it did because our operating systems got better. The viewer, If anything has gotten progressively worse as devs, upon finding they need to cache something have gone to cache code, peeked behind the curtain only to discover cthulhu himself. In the face of blinding existential terror the only course of action has been toss new cache code in the door and run away screaming, each new submission only serving to strengthen his master so that one day he may venture forth, all glory to his master, fewhj iuhwefdjhsd  kjkgefys ysdfh jgsd skjhe hsd g.

So yes if you can rule out hardware problems, multiple viewers fishing in the same cache is 100% the cause of cache problems.

A new cache is being developed using only soothing pastel tones and small chickens, it may be some time.

 

  • Like 1
  • Thanks 2
Link to comment
Share on other sites

8 minutes ago, CoffeeDujour said:

If you have multiple viewers in the same cache ...

One might wonder, then, why viewers' default cache locations are in the avatar-shared directory (under ...AppData\Local on Windows) rather than the avatar-specific directory (under ...AppData\Roaming), considering that we can't login more than once under the same account. I guess very theoretically some alt's journey through texture space could pre-heat the cache for the next alt's login session, but that's gotta save at best milliseconds per millenium. Must be some other reason.

Link to comment
Share on other sites

On 3/6/2019 at 8:13 PM, NiranV Dean said:

I had a good chuckle at that.

Coming from the Viewer that has the most suspicious backstory out of them all.

Then i realized, wait isn't Kitty technically on your Team as well? Does this mean you are distrusting your own members? But then wouldn't that be okay again since she's technically part of your team and thus suspicious like you? Is the Emerald story going to repeat soon? Will RLVa turn out to be a cryptominer? Does this make RLV a cryptominer too and why is a cryptominer based on another? So many questions.

As you might know our repository is publically accessible at all time - not just a drop of a ton of commits after release or after asking for it several times - so specific people making other viewers and usually yanking unreleased code into their own can continue doing so.

Link to comment
Share on other sites

13 minutes ago, Qie Niangao said:

One might wonder, then, why viewers' default cache locations are in the avatar-shared directory (under ...AppData\Local on Windows) rather than the avatar-specific directory (under ...AppData\Roaming), considering that we can't login more than once under the same account. I guess very theoretically some alt's journey through texture space could pre-heat the cache for the next alt's login session, but that's gotta save at best milliseconds per millenium. Must be some other reason.

File IO is one of the most expensive things SL does, the second is probably talking to your GPU. It bet running multiple viewers would be a lot smoother if you just didn't have a cache at all. For the latter, you can run multiple nvidia graphics cards (don't have to be matched) each to their own display and then choose which one certain opengl processes run on .. personally found it handy for keeping SL on the side and using my main gpu for blender. Just don't try to make that work on Linux, because Linux says no.

Edited by CoffeeDujour
  • Like 1
  • Thanks 1
Link to comment
Share on other sites

Setting a custom cache location should help mitigate cache corruption when running 2 avatars.
The first logged in avatar will use the custom cache location.
The second logged in avatar will use the default cache location in read only mode.

You can also set separate cache locations for each avatar by using the viewer parameters: http://wiki.secondlife.com/wiki/Viewer_parameters

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

3 hours ago, Whirly Fizzle said:

That warning actually comes from the LL viewer code.

The viewer cache is a special little snowflake & has some odd problems.  LL actually have a project in the works that totally rewrites the viewer cache system, which should improve matters.
On systems that have a very slow hard drive but a fast connection, you can actually get better performance just not using the cache at all.
Also never underestimate how badly real time scanning of the viewers cache by antivirus software can slow down the times it takes to load stuff out of cache. Always whitelist the viewer cache folder in your antivirus software.
For systems with a fast hard drive/SSD, especially if you have a slow connection, you should absolutely use the maximum cache size available.

The only way Firestorm makes money is from having adverts on the website & on the wiki. A lot of people use adblockers these days too, we don't make much from the adverts tbh.
The ad money barely covers the cost of the server fees & Jess has to often pay out of pocket.
When funds get short to cover running costs, we do a donation drive, but this doesn't happen very often. Each time we have done a donation drive though, the response from our users has been amazing & we have had to shut down the donations within a couple of days as we have more then enough money.
Firestorm also pays for a KDU license which is incredibly expensive.  I'm not sure I'm allowed to reveal how much this costs per year but it's well into a four figure US$ sum.

"The Phoenix Firestorm Project Inc." is a registered not for profit organization. No-one on the team makes any money from the viewer, all ad money & donations from users go towards running costs.

Yes, exactly this.

Some TPVs receive Patreon donations, but the amount they earn per month is tiny for the amount of work it takes to develop & run a third party viewer.

Thank you for this. :) I know enough to be dangerous, so I apologize if I bit too hard there. LOL

That explanation of how Cache works is also really helpful. I've always presumed, through deductive reasoning, that viewer performance and user experience is relying on multiple different parts of the total equation: not only how fast is your computer brain, but how fast is the storage, how fast the circuits between the storage and the RAM, how fast the RAM is and that's all before we even get out of the first half box before getting to the graphics part! Hahaha

Then there is all that network stuff, which is another bazillion little parts that all have to work together.

I saw an old movie (that I really loved) about Bruce Willis going into space to blow up some asteroid and then the line I remember the most was when they are on the launch pad waiting to blast off and one guy says something like "Our lives relying on 500 million moving parts working perfectly together... all built by the lowest bidder." Hahaha

I guess I've always thought of the SL experience as something like that: so many "moving parts" all relying on each other to work perfectly. As is always the case, some parts work better than others and when those other parts aren't working right IT'S LL'S FAULT!

Edited by Alyona Su
  • Like 2
Link to comment
Share on other sites

I made a few sessions with Catznip since they all talk about the performance and I made the following observations:

Out of the box Catznip is alot faster than FS. But I have tweaked my FS alot for performance - compared to this tweaked FS: Catznip has higher FPS in areas where I get +60 or +90. Not important for me. At heavy locations where I get -30 or -20 Catznip was slower by 10 fps.

Ok, time to tweak Catznip. Not easy. In FS I type a word and get all advanced settings that contain that word. In Catznip - nada - I have to remember the exact setting. Ok so I do a partial tweaking. :) I can get Catznip to about the same performance as FS at heavy locations.

It looks a bit different on the screen, so the rendering must be a bit different. I have one hud that reacts on the windlight setting by getting darker or brighter (in Catznip only but not important enough to complain)

That's observations based on my system of course so can be completely different for someone else.
For the features: I don't miss what I never had - but I miss what I have used all the time. So I will not switch viewers without a significant reason. But I don't use viewer specific things that would block me from switching. Like the FS Preprocessor or Viewer AO.

Link to comment
Share on other sites

@Nova Convair that's great to hear (that you've got your viewers tweaked for maximum performance)! I think many of us do that. But since the majority of people do not, the real question is how do they (all viewers, including LL) compare out of the box? (Rhetorical question, of course). :D

Edited by Alyona Su
Link to comment
Share on other sites

5 hours ago, Nova Convair said:

I made a few sessions with Catznip since they all talk about the performance and I made the following observations:

Out of the box Catznip is alot faster than FS. But I have tweaked my FS alot for performance - compared to this tweaked FS: Catznip has higher FPS in areas where I get +60 or +90. Not important for me. At heavy locations where I get -30 or -20 Catznip was slower by 10 fps.

Ok, time to tweak Catznip. Not easy. In FS I type a word and get all advanced settings that contain that word. In Catznip - nada - I have to remember the exact setting. Ok so I do a partial tweaking. :) I can get Catznip to about the same performance as FS at heavy locations.

There was an interesting section on the Catznip release notes about performance: http://catznip.com/index.php/Catznip_R12_1_Release_Notes

Quote

Actual performance improvements often depend on your specific computer: all being well, performance should be at least that of R12. If your computer is CPU bound and has GPU to spare then you should see an improvement. If you don’t currently have shadows enabled because it was a little too slow to have enabled all the time, try enabling them with the new release, for some the change is enough to allow “shadows by default”.

So if your system is CPU bound, it's probably going to give you a nice performance boost.

 

5 hours ago, Nova Convair said:

It looks a bit different on the screen, so the rendering must be a bit different. I have one hud that reacts on the windlight setting by getting darker or brighter (in Catznip only but not important enough to complain)

This is actually an old bug & it will also reproduce on Firestorm too, but it depends on your graphics settings whether you will see it. I suspect you may have ALM enabled in one viewer & not the other?

BUG-225631 - HUDs being effected by windlight

This bug should actually be fixed with EEP.
BUG-225446 - [FIXED] [EEP] HUDs are affected by Atmosphere & Lighting changes.

 

  • Thanks 1
Link to comment
Share on other sites

2 hours ago, Whirly Fizzle said:

This is actually an old bug & it will also reproduce on Firestorm too, but it depends on your graphics settings whether you will see it. I suspect you may have ALM enabled in one viewer & not the other?

Both viewers with ALM ON and shadows ON tested. I never switch off ALM.

  • Thanks 1
Link to comment
Share on other sites

Even with the best possible practices and control over everything and everyone in a rendered scene, getting a repeatable benchmark in SL with one viewer is notoriously difficult all by itself, I would't want to even try comparing viewers. Our advice is to put the FPS counter away and use the viewer that feels better, whichever one that may be.

And regardless of which viewer you use - relog more often, relogs are  great.

As a side note, we make a point of not exposing every debug setting or adding options with options in the preferences . . keep it simple and all that. We don't support going off into the actual debug settings and changing things. Many debug settings are exactly that, some are junk, some are poorly named, some have non obvious and entirely negative side effects and unless you have the source code open while you poke about . . well, here be dragons.

If you think we have missed a trick and find yourself changing something constantly, please use the feedback button in the viewer to let us know.

There may well be a search in an upcoming release. 

  • Like 3
Link to comment
Share on other sites

  • 2 weeks later...
13 hours ago, CoffeeDujour said:

3XOpSfm.png

Preferences > chat > logging tab

OH! So THAT is what that preference does. I never knew 1) that there was a difference in those log file names and 2) What that preference did.

Thank you! :D

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 1834 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...