BGE DEV: More BGE performance enhancements, go Ben2610!

[Ace Dragon]

Massive speed up with removing a scene with large numbers of objects.
http://lists.blender.org/pipermail/b...ay/019213.html

Another big speedup in the scenegraph, scenegraph performance will now be 4 to 10 times faster than in 2.48.
http://lists.blender.org/pipermail/b...ay/019239.html
------------------------------------------------------
Great new commits from Ben2610 for Blender 2.49, for those wanting to make large games, this and 2.50 will be the versions to start doing that with Ben improving BGE performance for the game being done by Entertainment Arts Research.

[gomer]

More speedups! Hope this gets in 2.49.

[vitorbalbio]

More one Chapter for BGE world domination!!!!
HUHUHUHAHAHAH ( Evil laugh )

Thanks Ben2610! You´re Great man!

[J09]

Go Ben go!!

[vitorbalbio]

Originally Posted by J09 Go Ben go!!

AUHAUHAuaHuahAUHAUH very good man! ^^

[AD-Edge]

Awesome, this is great.
Also good to see a breakdown of what gets the best speedups, with values to go along with different settings.

Great work Ben

[SamCameron]

simple... AMAZING, looking forward for the face culling option.

[fobsta]

Cool can't wait to test this out on my large scene. Thankk you Ben

[Mr. Smith]

BGE is jumping to next gen space!

[ideasman42]

Just tested these optimizations, they are awesome.
We were joining objects to increase performance but now you get better performance by keeping them as instances!

[B3D00]

Yeah, nice!

Originally Posted by J09 Go Ben go!!

Good one

[jessegp]

TOO Funny (make message long)

[kay_Eva]

OH

EM

JEA

keep us updated CD man!!!!!

thanks Ben, the force is strong with this one

[SamCameron]

Wow, there's somekind of face culling, I'm saying 'somekind' cause it works great in small windows, but in full screen (inside blender, not in blenderplayer) doesn't affect like in small screen.

[Doggie_B]

Originally Posted by vitorbalbio More one Chapter for BGE world domination!!!! HUHUHUHAHAHAH ( Evil laugh )

Let me join your evil join and we'll scare the little kids with our overpowered knowledge.

Muhahahahahahaha



Back to topic: also hope this will be included in 2.49

[ideasman42]

"Great new commits from Ben2610 for Blender 2.49"
Yes it will be in 2.49 - unless commits are made in a branch they will be in the next release, dont let the feature freeze RC's status confuse you!

You guys should also be getting excited about the new "Tap" sensor option which makes it easy to do jump logic with just 2sensors and an actuator, no properties, states, python which were needed before.

[AD-Edge]

Originally Posted by ideasman42 You guys should also be getting excited about the new "Tap" sensor option which makes it easy to do jump logic with just 2sensors and an actuator, no properties, states, python which were needed before.

Ah man, haven't heard about that sensor yet. Something I could have used so many times in the past, but instead had to do a timer setup with property sensors everywhere etc.
Awesome to hear.

[J09]

Originally Posted by ideasman42 You guys should also be getting excited about the new "Tap" sensor option which makes it easy to do jump logic with just 2sensors and an actuator, no properties, states, python which were needed before.

Sweet!! Thanks for all the hard work you're putting in!!

I feel like there could be another great GE project to show off all the new features.

[Sim88]

It's an insane speed up! And rasterizer is next, that's good, it's the only one which is over 2% now

[Sim88]

AMAZING NEWS!

Rasterizer improvements!
He codes so fast!

http://lists.blender.org/pipermail/b...ay/019285.html

Log Message: ----------- BGE performance, 3rd round: culling and rasterizer. This commit extend the technique of dynamic linked list to the mesh slots so as to eliminate dumb scan or map lookup. It provides massive performance improvement in the culling and in the rasterizer when the majority of objects are static. Other improvements: - Compute the opengl matrix only for objects that are visible. - Simplify hash function for GEN_HasedPtr - Scan light list instead of general object list to render shadows - Remove redundant opengl calls to set specularity, shinyness and diffuse between each mesh slots. - Cache GPU material to avoid frequent call to GPU_material_from_blender - Only set once the fixed elements of mesh slot - Use more inline function The following table shows the performance increase between 2.48, 1st round and this round of improvement. The test was done with a scene containing 40000 objects, of which 1000 are in the view frustrum approximately. The object are simple textured cube to make sure the GPU is not the bottleneck. As some of the rasterizer processing time has moved under culling, I present the sum of scenegraph(includes culling)+rasterizer time Scenegraph+rasterizer(ms) 2.48 1st round 3rd round All objects static, 323.0 86.0 7.2 all visible, 1000 in the view frustrum All objects static, 219.0 49.7 N/A(*) all invisible. All objects moving, 323.0 105.6 34.7 all visible, 1000 in the view frustrum Scene destruction 40min 40min 4s (*) : this time is not representative because the frame rate was at 60fps. In that case, the GPU holds down the GE by frame sync. By design, the overhead of the rasterizer is 0 when the the objects are invisible. This table shows a global speed up between 9x and 45x compared to 2.48a for scenegraph, culling and rasterizer overhead. The speed up goes much higher when objects are invisible. An additional 2-4x speed up is possible in the scenegraph by upgrading the Moto library to use Eigen2 BLAS library instead of C++ classes but the scenegraph is already so fast that it is not a priority right now. Next speed up in logic: many things to do there...

This table shows a global speed up between 9x and 45x compared to 2.48a for scenegraph, culling and rasterizer overhead.

This table shows a global speed up between 9x and 45x compared to 2.48a for scenegraph, culling and rasterizer overhead.

EDIT:
AND!!

Hi, You may have noticed this commit log: http://lists.blender.org/pipermail/b...ay/019285.html I forgot to mention something in the commit log: I have reactivated the combine use of VBO and display list, which should provide some speed up in the rendering. This possibility was ruled out in the BGE due to a problem with ATI card: // workaround: note how we do not use vertex arrays for making display // lists, since glVertexAttribPointerARB doesn't seem to work correct // in display lists on ATI? either a bug in the driver or in Blender .. This was some time ago and I think it's worth testing again. Could some ATI card owners test the BGE with trunk and report any problem here? Of course don't forget to activate display list ;-) /benoit

I'll test as soon as build pops up on graphicall! (Better start building myself soon)