[Bf-blender-cvs] [d1cb8b0] master: Use squareroot of the current epsilon error target for CG, since we square this value anyway.

[Lukas Tönne]

Commit: d1cb8b08bdcd4fbfef18960aa7ddee712d69d317
Author: Lukas Tönne
Date:   Thu Sep 4 19:57:21 2014 +0200
Branches: master
https://developer.blender.org/rBd1cb8b08bdcd4fbfef18960aa7ddee712d69d317

Use squareroot of the current epsilon error target for CG, since we
square this value anyway.

===================================================================

M	source/blender/blenkernel/intern/implicit.c

===================================================================

diff --git a/source/blender/blenkernel/intern/implicit.c b/source/blender/blenkernel/intern/implicit.c
index 6e0a9d7..0d4a7ec 100644
--- a/source/blender/blenkernel/intern/implicit.c
+++ b/source/blender/blenkernel/intern/implicit.c
@@ -839,7 +839,7 @@ static int cg_filtered(lfVector *ldV, fmatrix3x3 *lA, lfVector *lB, lfVector *z,
 {
 	// Solves for unknown X in equation AX=B
 	unsigned int conjgrad_loopcount=0, conjgrad_looplimit=100;
-	float conjgrad_epsilon=0.0001f /* , conjgrad_lasterror=0 */ /* UNUSED */;
+	float conjgrad_epsilon=0.01f;
 	
 	unsigned int numverts = lA[0].vcount;
 	lfVector *fB = create_lfvector(numverts);