Blender Freeze OpenCL Render

[jensverwiebe]

And....

Code: Select all

	const u_int threadCount = 1;
	vector<boost::thread *> threads(threadCount, NULL);

Code: Select all

[LuxCore][1.590] Building visibility map of light source: __WORLD_BACKGROUND_LIGHT__
[LuxCore][1.694] Visibility map done in 0.10 secs with 1000000 samples (9.62M samples/sec)

vs.

Code: Select all

	const u_int threadCount = 1;
	vector<boost::thread *> threads(12, NULL);

Code: Select all

[LuxCore][1.445] Building visibility map of light source: __WORLD_BACKGROUND_LIGHT__
[LuxCore][1.523] Visibility map done in 0.08 secs with 1000000 samples (12.87M samples/sec)

Imho forget the threading there.

[Dade]

Fixed.

And....

Code: Select all

	const u_int threadCount = 1;
	vector<boost::thread *> threads(threadCount, NULL);

Code: Select all

[LuxCore][1.590] Building visibility map of light source: __WORLD_BACKGROUND_LIGHT__
[LuxCore][1.694] Visibility map done in 0.10 secs with 1000000 samples (9.62M samples/sec)

vs.

Code: Select all

	const u_int threadCount = 1;
	vector<boost::thread *> threads(12, NULL);

Code: Select all

[LuxCore][1.445] Building visibility map of light source: __WORLD_BACKGROUND_LIGHT__
[LuxCore][1.523] Visibility map done in 0.08 secs with 1000000 samples (12.87M samples/sec)

Imho forget the threading there.

You are comparing 1 thread Vs 1 thread. In both cases, you are running with one single thread. Only the size of the vector is changed.

Anyway, try with 100,000,000 samples and a scene with a couple of millions of triangles, you will change your mind.

[jensverwiebe]

....
Anyway, try with 100,000,000 samples and a scene with a couple of millions of triangles, you will change your mind.

I plain believe you here

Jens