SL Viewer Performance Comparison: Linden's Still Best, Exodus & Mikshake Best of Third Party Viewers

This is a very useful performance comparison of eight popular Second Life viewers ably and laboriously created by Inara Pey of Modem World, who deserves props for the heavy lifting. While it's not a fully rigorous benchmark test (for instance, it was only conducted on her own system), the official SL viewer comes out on top, with significantly better frame rate performance with graphics settings on High. However, when all the fancy visual features are turned up, the third party viewers (or TPVs) begin to get competitive, with perhaps Exodus (which I blogged here) and Milkshake doing best. And considering that TPVs can implement experimental code like Qarl's mesh tweaks, they're definitely worth considering for folks with high-end systems. Notably, with SL set to its maximum graphic richness (i.e. ambient lighting and dynamic shadows turned up), the world slows to a crawl, no matter what the viewer. See much more here.

And yes, there's a TPV called Milkshake. Apparently by a Swedish coder, you can get it here, and read a good write-up of it here. (It's not listed in the Linden's official directory, so user beware.) However, just like the Exodus viewer should have the Bob Marley song as its official music, I recommend this totally awesome track for Milkshake:

And do share your own experiences with TPVs in Comments!

Chart credit: Modem World.

Comments

[Pussycat Catnap]

Could have sworn SL official's number was 3.2.4.246439 and not 3.2.6?

That from the download link.

Beta lists as Second_Life_3-2-5-246958_BetaViewer_Setup.exe
(and like for Mac and Linux formats).

[Ener Hax]

great post! thanks! =)

[Pussycat Catnap]

Original blog's graph has been updated to edit version of SL viewer used to 3.2.5.

[Hitomi Tiponi]

These tests are always useful to see and provide an interesting basis for comparison, but are somewhat dependant on the equipment used (and Inara lists hers, and is getting decent figures with an old Nvidia 9800GT GPU). I did a similar, but simpler, test using my even more aged Nvidia 8600GT GPU equipped PC a fortnight ago - results at http://farm8.staticflickr.com/7143/6537350269_2e64984ef6_b.jpg, which also includes Viewer 1.

BTW Inara has since corrected the Second Life Viewer number to 3.2.5.

FYI Milkshake is produced by Cinder Galthie for herself and her friends.

[okiedokiecow]

All tests conducted on a single system? Why are you even posting this tripe to begin with? ( lets see if this doesn't get deleted.)

I guess if they all run ok for her then they must run ok for everyone. Whatever.

[Henri Beauchamp]

This tests are irrelevant and the testing method is all but rigorous. See why in my comments here: http://dwellonit.taterunino.net/2012/01/03/inaras-excellent-viewer-frame-rate-comparison/#more-5299

[M.A.DeLuca]

"Totally" irrelevant is quite an exaggeration. She published her results and documented her methodology. You didn't, Henri. In a publish or perish world, you've shown yourself the door for this debate, as it were. But the neat thing about science is that you can always lick your wounds, learn from your mistakes, and come back swinging with better methodology that addresses the criticisms you're lobbing from your comfy armchair. But for the moment, Inara's results are the best reference we've got, and it was good of her to actually share them.

[Eyeball Soup]

Actually her testing methodology is super flawed. Just reporting tons of information about how you did a test doen't make the results valid unless your audience is ignorant of the technology being tested.

To wit:

She used ancient hardware and didn't specify her graphics driver revision.

She moved her avatar which introduces an infinite number of variables from test run to test run.

Those suggestions on Shopping Cart Disco are wrong. I won't go into the technical details but post-Shining Fixes it's better to let SL tell your video card to antialias than it is to turn it off in SL and switch it on in the Nvidia control panel.

She calculated time to rez which could have been influenced by an infinite number of variables (bandwidth on the greater internet etc) that are not even remotely related to how a particular viewer performs.

She did not use the same windlight setting or the same time of day settings. She could have overcome this by hitting ctrl+shift+Y and going midday before sampling the framerate.

So yes as Henri said (and he's a whole lot smarter than me) her test results are completely invalidated by the reasons he and I gave.

NWN Staff: To cut down on the flood of misinformation that this blog seems to engender, please get a technical person on your staff that understands how hardware and software actually work.

[Pussycat Catnap]

Inara didn't claim her test was scientific or accurate - merely that it showed what was good for her, and how she found that out.

The amount of 'nerd rage' and curmudgeony 'you kids get off my lawn'-rage over mesh and/or viewer-3 is just getting absurd at this point... that folks want to burn any heretic they can find at the stake.

Like it or not, these new LL viewers just plain work great for a growing number of folks. They may not work for some, but for others they are doing just fine.

And since they're going to keep on coming, one might be better served figuring out how to get it working, or submitting bug reports when it doesn't, than trying to turn back time and bringing out the pitchforks anytime it does work for someone.

[Foxxe Wilder]

Amazing... Four out of five of those 'top viewers' I've never heard of. Plus there is no mention at ALL of Cool VL, In my opinion, the most user friendly of all mesh enabled viewers!
Admittedly there are a few issues I'd like to see in COOL like the movie on a prim stuff, But I assume such things are coming soon and naturally are not given priortity due to the fact that you can use an external browser for those links anyways...

BUT LL's being the best? Hmmmph... created by coders that are apparently ADD afflicted. They work on something, get distracted, leave it somewhere strange and forget all about it? No...

And btw... NO ONE cares what children's music you would choose for a 'nobody viewer' Music is not played upon connection (and thank the gods such is so, If I hear another piece of electronic Pac-man style 'music' I'll throw up all over the grid!

And for you geeks... you are still not equivalent to nerds. You are just a fad, and like all others, you will come and you will go.

[jjccc coronet (@JJcccART)]

just wasted my time downloading all the viwers i didnt have and they all look the same but firestorm beta is the only one with quick preferences to change the wind lite setting the rest you have to look through endless menus so they are a waste of time

[foneco zuzu]

Exodus i enjoyed, even if it limits the draw distance to 512!
I cant agree with henry that Sl runs better with AA and Aq provide that letting hardware override it!
The viewwer i use, Phoneix pre mesh latest, is the one that allows me to stand more time on sl, due to a simple fact, i can use 32 AA and 32 Aq and that makes a lot of difference for some with poor eyeisght, that starts to feel water in her eyes, after only 5min of using any mesh viewwer, and allow me to do 4h or racing with it!

[Taylor Schroeder]

Just as a heads up ... I attempted to install Exodus today, and Norton's security suite quarantined it, saying it was rarely used. Only 5 users in their user base. If you have this issue, you can unquarantine it, and you can also "report detection of a suspected clean file" at: https://submit.symantec.com/false_positive/