add async example in knet tutorial

examples/mnist_knet.jl

@@ -456,6 +456,73 @@ augmented_log = @time train_augmented(epochs=200);
 #' performance (aside from simplifying the augmentation pipeline)
 #' would be to increase the number of available threads.
 
+#' ## Improving Performance
+info("Improving Performance") #jl-only
+
+#' One of the most effective ways to make the most out of the
+#' available resources is to augment the next (couple) mini-batch
+#' while the current minibatch is being processed on the GPU.
+#' We can do this via julia's build in parallel computing
+#' capabilities
+
+#' First we need a worker process that will be responsible for
+#' augmenting our dataset each epoch. This worker also needs
+#' access to a couple of our packages
+
+# addprocs(1)
+# @everywhere using Augmentor, MLDataUtils
+
+#' Next, we replace the inner `eachbatch` loop with a more
+#' complicated version using a `RemoteChannel` to exchange and
+#' queue the augmented data.
+
+function async_train_augmented(; epochs = 500, batchsize = 100, lr = .03)
+    w = weights()
+    log = MVHistory()
+    p = Progress(epochs, desc = "Async Augmented: ") #jl-only
+    for epoch in 1:epochs
+        @sync begin
+            local_ch = Channel{Tuple}(4) # prepare up to 4 minibatches in adavnce
+            remote_ch = RemoteChannel(()->local_ch)
+            @spawn begin
+                # This block is executed on the worker process
+                batch_x_aug = zeros(Float32, size(train_x,1), size(train_x,2), 1, batchsize)
+                for (batch_x_cpu, batch_y) in eachbatch((train_x ,train_y), batchsize)
+                    # we are still using multithreading
+                    augmentbatch!(CPUThreads(), batch_x_aug, batch_x_cpu, pl)
+                    put!(remote_ch, (batch_x_aug, batch_y))
+                end
+                close(remote_ch)
+            end
+            @async begin
+                # This block is executed on the main process
+                for (batch_x_aug, batch_y) in local_ch
+                    batch_x = KnetArray{Float32}(batch_x_aug)
+                    g = costgrad(w, batch_x, batch_y)
+                    Knet.update!(w, g, lr = lr)
+                end
+            end
+        end
+
+        next!(p) #jl-only
+        if (epoch % 5) == 0
+            train = acc(w, train_x, train_y)
+            test  = acc(w, test_x,  test_y)
+            @trace log epoch train test
+            msg = "epoch " * lpad(epoch,4) * ": train accuracy " * rpad(round(train,3),5,"0") * ", test accuracy " * rpad(round(test,3),5,"0")
+            cancel(p, msg, :blue) #jl-only
+#md             println(msg)
+#jp             println(msg)
+        end
+    end
+    finish!(p) #jl-only
+    log
+end
+#md nothing # hide
+
+#' Note that for this toy example the overhead of this approach
+#' is greater than the benefit.
+
 #' ## Visualizing the Results
 info("Visualizing the Results") #jl-only