Cuda use streams and async to avoid busywaiting

This is my first stab at this and is based on this IRC converstation:

<mib2berlin> brecht: this is meaning as reminder only, I know you have other things to do > http://openvidia.sourceforge.net/index.php/Optimization_Notes#avoiding_busy_waits
<brecht> mib2berlin: thanks, bookmarked

only tested on Ubuntu 14.04 / cuda 5.0 but ill do some more testing tomorrow.

Also unsure about the placement and the lifetime of the stream and the event. But creating / deleting these seems to incur a non trivial cost.

Reviewers: brecht

Reviewed By: brecht

CC: mib2berlin, dingto

Differential Revision: https://developer.blender.org/D262

1 files changed, 10 insertions, 2 deletions

diff --git a/intern/cycles/device/device_cuda.cpp b/intern/cycles/device/device_cuda.cpp
index 0fbb48cf431..3073f078d12 100644
--- a/intern/cycles/device/device_cuda.cpp
+++ b/intern/cycles/device/device_cuda.cpp
@@ -41,6 +41,8 @@ public:
 	CUdevice cuDevice;
 	CUcontext cuContext;
 	CUmodule cuModule;
+	CUstream cuStream;
+	CUevent tileDone;
 	map<device_ptr, bool> tex_interp_map;
 	int cuDevId;
 	int cuDevArchitecture;
@@ -207,6 +209,9 @@ public:
 		if(cuda_error_(result, "cuCtxCreate"))
 			return;
 
+		cuda_assert(cuStreamCreate(&cuStream, 0))
+		cuda_assert(cuEventCreate(&tileDone, 0x1))
+
 		int major, minor;
 		cuDeviceComputeCapability(&major, &minor, cuDevId);
 		cuDevArchitecture = major*100 + minor*10;
@@ -223,6 +228,8 @@ public:
 	{
 		task_pool.stop();
 
+		cuda_assert(cuEventDestroy(tileDone))
+		cuda_assert(cuStreamDestroy(cuStream))
 		cuda_assert(cuCtxDestroy(cuContext))
 	}
 
@@ -645,9 +652,10 @@ public:
 
 		cuda_assert(cuFuncSetCacheConfig(cuPathTrace, CU_FUNC_CACHE_PREFER_L1))
 		cuda_assert(cuFuncSetBlockShape(cuPathTrace, xthreads, ythreads, 1))
-		cuda_assert(cuLaunchGrid(cuPathTrace, xblocks, yblocks))
+		cuda_assert(cuLaunchGridAsync(cuPathTrace, xblocks, yblocks, cuStream))
 
-		cuda_assert(cuCtxSynchronize())
+		cuda_assert(cuEventRecord(tileDone, cuStream ))
+		cuda_assert(cuEventSynchronize(tileDone))
 
 		cuda_pop_context();
 	}