[Bf-blender-cvs] [53f277a2e6b] master: Compositor: Allocate OIDN memory after receiving lock.

[Jeroen Bakker]

Commit: 53f277a2e6b433c7ad4fa3ad54547842e5672a8d
Author: Jeroen Bakker
Date:   Mon Apr 12 14:29:49 2021 +0200
Branches: master
https://developer.blender.org/rB53f277a2e6b433c7ad4fa3ad54547842e5672a8d

Compositor: Allocate OIDN memory after receiving lock.

Related to {T77023}. When using many Denoise nodes the memory in OIDN
are allocated up front. A mutex could stall the process until. This
change will allocate the memory after it received the lock.

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

M	source/blender/compositor/operations/COM_DenoiseOperation.cc

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

diff --git a/source/blender/compositor/operations/COM_DenoiseOperation.cc b/source/blender/compositor/operations/COM_DenoiseOperation.cc
index 587afdc2d00..ec11ad4d69a 100644
--- a/source/blender/compositor/operations/COM_DenoiseOperation.cc
+++ b/source/blender/compositor/operations/COM_DenoiseOperation.cc
@@ -101,6 +101,11 @@ void DenoiseOperation::generateDenoise(float *data,
   if (BLI_cpu_support_sse41())
 #  endif
   {
+    /* Since it's memory intensive, it's better to run only one instance of OIDN at a time.
+     * OpenImageDenoise is multithreaded internally and should use all available cores nonetheless.
+     */
+    BLI_mutex_lock(&oidn_lock);
+
     oidn::DeviceRef device = oidn::newDevice();
     device.commit();
 
@@ -145,10 +150,6 @@ void DenoiseOperation::generateDenoise(float *data,
     }
 
     filter.commit();
-    /* Since it's memory intensive, it's better to run only one instance of OIDN at a time.
-     * OpenImageDenoise is multithreaded internally and should use all available cores nonetheless.
-     */
-    BLI_mutex_lock(&oidn_lock);
     filter.execute();
     BLI_mutex_unlock(&oidn_lock);