LuxRender 0.9-dev Windows build of 27-02-2012

[Lord Crc]

New builds, yay!

For a writeup, please see SATtva's post.

Important: These builds are made with Boost 1.47 and will be incompatible with older builds.

The new builds from SATtva linked above should be with Build 1.47 and should work fine in network mode. If you can test network rendering between Linux and Windows please do and report back here. Thanks!

We're trying out a new version of Boost as it solves some issues we've had, and also seems to bring with it a nice speedup on OSX and possibly Linux.

These builds contain LuxRender, LuxConsole, LuxMerger, LuxBlend25 and Pylux for Blender 2.6x, and LuxBlend for Blender 2.49.
The NoOpenCL builds does not require any OpenCL driver or capabilities, but also does not use the GPU for the "hybrid" renderer.

LuxRender_CVS_20120227_64bit_OpenCL.exe
LuxRender_CVS_20120227_64bit_NoOpenCL.exe
LuxRender_CVS_20120227_32bit_SSE2_OpenCL.exe
LuxRender_CVS_20120227_32bit_SSE2_NoOpenCL.exe
LuxRender_CVS_20120227_32bit_SSE1_NoOpenCL.exe

Built with Visual Studio 2008 SP1 + Windows 7 SDK on Win 7 64bit, using AMD APP SDK 2.5 (formerly ATI Stream SDK).

Changelog:

Code: Select all

4 hours ago   David Bucciarelli   Fixed a bug introduced in QBVH code with the last commits
8 hours ago   David Bucciarelli   Merge
8 hours ago   David Bucciarelli   First (probably) working version of SQBVH
2 days ago   David Bucciarelli   Fixed node bounding boxes clipping for SQBVH
3 days ago   David Bucciarelli   More fix for SQBVH
3 days ago   David Bucciarelli   Changed the total SAH computation for (S)QBVH
4 days ago   David Bucciarelli   Fixed the code for tringle split clipping for SQBVH
5 days ago   David Bucciarelli   Written the code for tringle split clipping for SQBVH
6 days ago   David Bucciarelli   Written the code for space split building of SQBVH
7 days ago   David Bucciarelli   Written the skeleton code for the SQBVH
8 days ago   David Bucciarelli   Added avg. primitive references per leaf to QBVH statistics
8 days ago   David Bucciarelli   Added the code for collecting statistics about the quality of the QBVH built
27 hours ago   Jens Verwiebe   Homer_s's patch for compiling Linux shared and with visibilitly-support
2 days ago   Jens Verwiebe   OSX: compile now all with hidden visibility and use symbol exporting mechanism to control export
4 days ago   Guillaume Bouchard   Merge with bxdf memory reduction
4 days ago   Guillaume Bouchard   SPPM: add a missing barrier
5 days ago   Jens Verwiebe   OSX: remove a reminder pastie
5 days ago   Jens Verwiebe   OSX: cleanup by replacing CMAKE_BUILD_TYPE with XCode env var $(CONFIGURATION) globally when generating XCode-projects
5 days ago   Jens Verwiebe   OSX: adjust substring to xcode major;minor - version only
5 days ago   Jens Verwiebe   OSX: unix makefile generator does not fill XCODE_VERSION var, so we execute xcodebuild -version  instead
5 days ago   Jens Verwiebe   OSX: messages more descriptive
5 days ago   Jens Verwiebe   OSX: remove a message set only for checking
5 days ago   Jens Verwiebe   OSX: fixed compiling with "unix makefiles" and some cleanups
6 days ago   Jens Verwiebe   OSX: forgot to set xcode env var for 1 reference
6 days ago   Asbjørn Heid   Added absorption to glossycoating, see issue #1193.
6 days ago   Jens Verwiebe   OSX: pylux.so - just not set prefix instead of renaming later
6 days ago   Asbjørn Heid   Added SMTPE color space preset, fixed colorspace preset values to match core values.
7 days ago   Jens Verwiebe   OSX: using now xcode env variables to control debug/release configurations, message changes to take this into account
8 days ago   Jens Verwiebe   OSX: remove configuration hack for now, problem lies in changed cmakebehavior
8 days ago   Jens Verwiebe   OSX: ups, set back my test dependencies to repo
8 days ago   Jens Verwiebe   OSX: hack to make build configurations failsafe on xcode < 4.3 again by limiting to cmake_build_type
8 days ago   Jean-Philippe Grimaldi   Silence a few gcc warnings.
8 days ago   Jens Verwiebe   OSX: assure buildtype is really set, make debug builds possible in xcode, todo: override the default build configuration ( default atm )
10 days ago   Jens Verwiebe   OSX: preparations for use of XCode 4.3 - atm cmake 2.87 does fail generating projects, due cmake looks in wrong location for XCode
10 days ago   Aldo Zang   Fixed issue in infinitesample Light. The sampled vector wi returned by the SampleL method was in Light coordinate system instead of the World coordinate system.
12 days ago   Asbjørn Heid   Fixed locale issue causing scenes to be incorrectly loaded via gui.
2 weeks ago   Guillaume Bouchard   SPPM: add a switch *useproba* to use probabalistic approach
2 weeks ago   Guillaume Bouchard   SPPM: remove the use of the second memory arena for storing hitpoint BSDF
2 weeks ago   Guillaume Bouchard   Transactional Memory Arena
2 weeks ago   Guillaume Bouchard   Factorise a bit the MemoryArena class
2 weeks ago   Asbjørn Heid   Added ability to set whitepoint to a given correlated color temperature (CCT).
2 weeks ago   Asbjørn Heid   Fixed erronous value in D50 whitepoint preset.
2 weeks ago   Asbjørn Heid   Small code cleanup of ColorSpaceWidget.
3 weeks ago   Jens Verwiebe   OSX: cleanup more overhead by not longer using -all_load in corelib, but link explicit binary deps into binary


[cwichura]

Is it safe to use this version to read and resume a render in progress started with the Feb 4th build?

[Abel]

Is it safe to use this version to read and resume a render in progress started with the Feb 4th build?

Nothing is certain when using development builds. On the few tests I've done this build works fine, but there seem to be issues with the new SQBVH accelerator: viewtopic.php?f=8&t=7923&start=10#p78306.

[cwichura]

SQBVH shouldn't be used unless explicitly referenced, correct? So my current scene should not be using it.

[Lord Crc]

Is it safe to use this version to read and resume a render in progress started with the Feb 4th build?

Depends on what you mean by "safe". If you care very much about your work and time, I would stay with the old build for doing things like that.

What should be ok (but might not, that's why we need testing) is to resume, however any film files you write with boost 1.47 may not be backwards compatible.

SQBVH shouldn't be used unless explicitly referenced, correct? So my current scene should not be using it.

Correct, the SQBVH is an entirely new accelerator. However introducing SQBHV may have affected some "plain" QBVH code, so if you get errors with QBVH I suggest you fall back to kdtree or the older build.

[Piita]

Great great great! Thanks everybody.

[Meelis]

Thx

[panton41]

I'm using the latest nVidia drivers (295.73 as of this writing) with a pair of GeForce 560 Ti and the hybrid mode is simply not rendering. It gives a black render with no work being done.

[Lord Crc]

I'm using the latest nVidia drivers (295.73 as of this writing) with a pair of GeForce 560 Ti and the hybrid mode is simply not rendering. It gives a black render with no work being done.

Anything in the log to give us a clue?

[b-tag]

I'm using the latest nVidia drivers (295.73 as of this writing) with a pair of GeForce 560 Ti and the hybrid mode is simply not rendering. It gives a black render with no work being done.

panton41,

If you have been previously rendering with "OpenCL work group size" = 0, now try using a different value. Changing this setting to 64 has worked for me for the recent two builds.

Credits to LordCRC btw.

b.