Subversion Repository Public Repository


This repository has no backups
This repository's network speed is throttled to 100KB/sec

Entry Size Revision Updated
branches 222 Wed 08 Jan, 2014 18:55:11 +0000
trunk 557 Tue 10 Nov, 2015 12:00:20 +0000

Commits for Divide-Framework/

Revision Author Commited Message
557 IonutCava picture IonutCava Tue 10 Nov, 2015 12:00:20 +0000

- Gather all rendering calls from glVertexArray and glGenericVertexData into a single function in glResources: submitRenderCommand.
— This allows both classes to interpret GenericDrawCommands in the same way
- Use unsigned ints for all hasehs instead of size_t to avoid negative number hashes (useful for debugging)

556 IonutCava picture IonutCava Mon 09 Nov, 2015 23:10:09 +0000

- Fixed faulty VB partitioning code
- Cleanup in state management system:
— Remove some useless clearing of buffers (e.g. in glimBatch code)
— Changing a VAO will invalidate the element buffer cache
-— ToDo: Investigate all states that need to invalidate cache on VAO change
- Don’t add RenderingComponent to the SceneRoot node

Known bug:
- If we are rendering only the skybox, it is not rendered properly until a different node comes into view. Needs investigation. Not state/shader or texture issue!

555 IonutCava picture IonutCava Sun 08 Nov, 2015 22:39:36 +0000

- Crash fix related to persistently mapped buffer’s lock managers
- Anaglyph shader fix

554 IonutCava picture IonutCava Sun 08 Nov, 2015 21:48:55 +0000

- Forgot one debug message print

553 IonutCava picture IonutCava Sun 08 Nov, 2015 21:48:10 +0000

- Improve VBO allocation system
- Disable mipmap generation for blur FBO in CascadedShadowMaps
- Code cleanup

552 IonutCava picture IonutCava Fri 06 Nov, 2015 17:07:22 +0000

- Use UNIT0 and UNIT1 during depth pass for transparency texture to reduce texture binds between calls
- Allow different VAOs with different attributes per render stage per glVertexArray instance
- Initial work on multiple glVertexArray instances per single VBO (attempt to get VBOs to the 4MB mark) - Currently crashes the build
- Fix a bug on exit if background tasks depended on scene elements. Wait for the main threadpool to clear before destroying the scene.
- Use a different node buffer per rendering stage to avoid waiting too much on fence syncs

551 IonutCava picture IonutCava Thu 05 Nov, 2015 17:08:03 +0000

- Streamline RenderPass/RenderBin/RenderPassCuller logic:
— Separate culling from updating nodes
— Improve sorting and batching
- Cleanups and sanity checks
- Replace some std::bind calls with lambdas

550 IonutCava picture IonutCava Thu 05 Nov, 2015 00:23:23 +0000

- Do not bind unneeded textures during render pass, but sort properly if they are used

549 IonutCava picture IonutCava Wed 04 Nov, 2015 22:19:15 +0000

- Axis gizmo on/off toggle fixes
- Removed the manual call to makeTextureResident in RenderingComponent. Used registerTextureDependency instead
- Small cleanup in Material class

548 IonutCava picture IonutCava Wed 04 Nov, 2015 17:23:15 +0000

- More work on improving cascaded shadow maps