[LLVMbugs] [Bug 4900] New: clang: blender 2.49b renders wrong image

bugzilla-daemon at cs.uiuc.edu bugzilla-daemon at cs.uiuc.edu
Sat Sep 5 01:40:11 PDT 2009


http://llvm.org/bugs/show_bug.cgi?id=4900

           Summary: clang: blender 2.49b renders wrong image
           Product: new-bugs
           Version: 2.6
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3449)
 --> (http://llvm.org/bugs/attachment.cgi?id=3449)
compare-blender.sh

I built blender using gcc, llvm-gcc and clang (in objgcc, obj, objclang dirs
respectively), doing something like:
export CC=llvm-gcc CXX=llvm-g++
cmake ../blender-2.49b CC=llvm-gcc CXX=llvm-g++
make -j8

Then I wrote a script to render .blend files with 2 blenders and compare
outputs (compare-blender.sh attached).
Running it on the blender 2.49 regression files from blender.org I got some
failures:

Render of ./teapot_soft.blend is not exact
Image Difference (MeanSquaredError):
           Normalized    Absolute
          ============  ==========
     Red: 0.0000081281        0.0
   Green: 0.0000084478        0.0
    Blue: 0.0000151673        0.0
   Total: 0.0000105811        0.0
Checking whether reference render is deterministic ...Yes

When the script finds a different render, it reruns the render with the
reference blender again, and checks whether the output is deterministic (for
some renders it is not).

For teapot_soft.blend the render with reference blender is deterministic, but
differs from the clang compiled blender.

When building it with llvm-gcc everything is OK for this particular render.

Attached are also the 2 images, and the output of:
gm compare -highlight-style Assign -file diff.miff file_ref0001.png
file_new0001.png


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.



More information about the llvm-bugs mailing list