SecondLife(R) Viewer2 LL vs. Open Source SnowGlobe & Kirsten's S20(29)-(30) FPS spec comparision

Today I will be running a frames per second and various other specs. comparison test between the 4 Viewer2 versions ( please refer to the 3 previous posts beginning July 31 ) Over the next 24 hours each viewer will repeat the prior test, the control viewer will again be Emerald 1.25.36 / ALL 5 Viewers are set to a network bandwith speed of 530mbps. This series of tests will differ slightly from the previous 2. This is Windows only,(Linux is next :_) I will be using IoBit Software's, Freeware Game Booster and Smart Ram to shut down all background process's and insure maximum system Ram is available, my first step will be to check how many inventory items are still in the cache upon startup out of yesterdays 11,533
Official SecondLife Viewer2 v. 2.1.0.207030
Inventory : 10,098 auto loaded the remainder after entering the letter A / test 2 was =
Start Up Ram available before load610mb program log in 370mb program end 285mb/ test2 end 400mb
FPS : 4.5 - 7.1 relatively stable at 5.6 / test2 2.8 - 8.8 stable at 5.5
Bandwith : 30 - 410kphs peak typical average 45kphs / test2 11 - 311kphs typical average 45kphs
Packet Loss 0.0 - 0.1% / test2 0.0 - 1.8%
Ping Sim : 129 -2675msec relatively stable at 160 / test2 149 - 500msec relatively stable at 240
SnowGlobe Viewer2 v. 2.0.2.3576
Inventory : 10,975 auto loaded the remainder after entering the letter A / test 2 11.0800
Start Up Ram available before load610mb program log in 350mb program end 304mb / test2 end 300mb
FPS : 2.2 - 8.9 relatively stable at 6 test 2 1.9 - 7.2 relatively stable at 5.5
Bandwith : 12 - 417kbps test2 17 - 40kbps typical average both tests 27kbps
Packet Loss 0.0 - 1.0% / test2 0.0 - 1.7%
Ping Sim : 129 - 2675msec relatively stable at 145 / test 2 139 - 199msec relatively stable at 135
SnowGlobe Viewer2 v. 2.1.0.3587
Inventory : 4,804 after pressing the letter a then required a force reload / test2 7454
Start Up Ram available before load610mb program log in 370mb program end 296mb / test2 365mb program end
FPS : 1.6 - 6.9 * relatively stable at 2.5 - 3.0 / test2 2.8 - 8.8 relatively stable at 4.5
Bandwith : 33 - 190kbps / test2 / test2 23 - 320kbhs
Packet Loss 0.0 - 0.5% / test2 0.0 - 37%
Ping Sim : 128 - 5620ms / test2 189 - 504ms / average about 180 both tests
Kirsten's S(20) Build(30)
Inventory : 10,975 auto loaded the remainder after entering the letter a / test2 start 11352mb
Start Up Ram available before load 610mb program log in 362mb program end 360mb / test2 end 362mb
FPS : 2.2 - 10.3 / test2 3.1 - 11.7 relatively stable at 5.0 both tests
Bandwith : 25 - 465kbps / test2 17 - 227kbps
Packet Loss 0.0 - 2.3% / test2 0.0 - 4.0%
Ping Sim : 121 - 1104msec / test2 138 - 560 average at about 180 both tests
Emerald 1.25.36
Inventory : 11,110 auto loaded the remainder by it's self / test2 start fully loaded at 11,533
Start Up Ram available before load 610mb program log in 334mb program end 403mb / test2 end 327mb
FPS : 4.6 - 11.3 relatively stable at 6.0 - 6.5 / test2 4.1 - 8.9 relatively stable at 6.0 - 6.5
Bandwith : 25 - 540kbps / test2 23 - 310kbps
Packet Loss : 0.0 - 0.1% / test2 0.0 - 0.1%
Ping Sim : 104 - 383msec / test2 140 - 780msec average about 170

Ok, so what do these numbers mean ? First of all this is by no means intended to be looked upon as a controlled scientific test, it represents the real world, everyday average users experience, but in a controlled static setting, a private estate island. In a crowded mall or club I am sure the results would be different, please keep in mind the age of the hardware I am using for the test as well, a 9 year old PC and 4 or 5 year old video card ( this card was considered to be top notch when it was new, so I am guessing that helps a lot )

What these numbers say to me is that both the official LL V2 and Kirstens S20 Build(30) are for the most part in a neck and neck race, neither one is better than the other performance wise. And both of them are just slightly behind the Emerald control viewer. Thats pretty good, when you consider that the 1.23.5.xxx code base has been around for a very long time. :_)

I believe I will bypass any further Linux testing, as the majority of users are running Windows. Later today I will begin usability testing the UI (user interface ) of each individual viewer. Please feel free to post any comments.

JayR Cela :_)