Merge pull request #59 from JuliaGPU/lcw/fixwait

Fix CUDA waiting on CUDA events

Author: vchuravy

src/backends/cuda.jl

@@ -68,7 +68,9 @@ function wait(::CPU, ev::CudaEvent, progress=nothing)
 end
 
 # Use this to synchronize between computation using the CuDefaultStream
-wait(::CUDA, ev::CudaEvent, progress=nothing) = __enqueue_wait(ev, CUDAdrv.CuDefaultStream())
+function wait(::CUDA, ev::CudaEvent, progress=nothing)
+    CUDAdrv.wait(ev.event, CUDAdrv.CuDefaultStream())
+end
 
 # There is no efficient wait for CPU->GPU synchronization, so instead we
 # do a CPU wait, and therefore block anyone from submitting more work.

test/test.jl

@@ -183,6 +183,11 @@ if has_cuda_gpu()
         wait(event5)
         @test event5 isa KernelAbstractions.Event
     end
+    @testset "CUDA wait" begin
+        event = kernel_empty(CUDA(), 1)(ndrange=1)
+        wait(CUDA(), event)
+        @test event isa KernelAbstractions.Event
+    end
 end
 
 @testset "CPU dependencies" begin