From f81e4131802d8f437ef52956aa760f56f9e39dd7 Mon Sep 17 00:00:00 2001
From: Benoit Steiner <benoit.steiner.goog@gmail.com>
Date: Thu, 5 May 2016 14:15:11 -0700
Subject: Added a benchmark to measure the performance of full reductions of 16
 bit floats

---
 bench/tensors/tensor_benchmarks.h           | 2 +-
 bench/tensors/tensor_benchmarks_fp16_gpu.cu | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

(limited to 'bench')

diff --git a/bench/tensors/tensor_benchmarks.h b/bench/tensors/tensor_benchmarks.h
index 62533a608..e0631b401 100644
--- a/bench/tensors/tensor_benchmarks.h
+++ b/bench/tensors/tensor_benchmarks.h
@@ -368,7 +368,7 @@ template <typename Device, typename T> class BenchmarkSuite {
     const TensorMap<Tensor<T, 2, 0, TensorIndex>, Eigen::Aligned> B(
         b_, input_size);
     Eigen::array<TensorIndex, 0> output_size;
-    TensorMap<Tensor<float, 0, 0, TensorIndex>, Eigen::Aligned> C(
+    TensorMap<Tensor<T, 0, 0, TensorIndex>, Eigen::Aligned> C(
         c_, output_size);
 
     StartBenchmarkTiming();
diff --git a/bench/tensors/tensor_benchmarks_fp16_gpu.cu b/bench/tensors/tensor_benchmarks_fp16_gpu.cu
index 14876556e..65784d0d6 100644
--- a/bench/tensors/tensor_benchmarks_fp16_gpu.cu
+++ b/bench/tensors/tensor_benchmarks_fp16_gpu.cu
@@ -33,6 +33,7 @@ BM_FuncGPU(algebraicFunc);
 BM_FuncGPU(transcendentalFunc);
 BM_FuncGPU(rowReduction);
 BM_FuncGPU(colReduction);
+BM_FuncGPU(fullReduction);
 
 
 // Contractions
-- 
cgit v1.2.3