CUDA vs. OpenCL NVIDIA Pascal GPU Computing
Tests by Michael Larabel.


GeForce GTX 1080: 

	Processor: Intel Xeon E3-1280 v5 @ 4.00GHz (8 Cores), Motherboard: MSI C236A WORKSTATION (MS-7998) v1.0, Chipset: Intel Sky Lake, Memory: 16384MB, Disk: Samsung SSD 950 PRO 256GB, Graphics: Device 8187MB (1603/5005MHz), Audio: Realtek ALC1150, Network: Intel Connection

	OS: Ubuntu 16.04, Kernel: 4.4.0-22-generic (x86_64), Desktop: Unity 7.4.0, Display Driver: NVIDIA 367.18, OpenGL: 4.5.0, Vulkan: 1.0.8, Compiler: GCC 5.3.1 20160413 + CUDA 8.0, File-System: ext4, Screen Resolution: 3840x2160


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: Triad
GB/s > Higher Is Better
CUDA ... 14.81 |===============================================================
OpenCL . 11.93 |===================================================


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: FFT SP
GFLOPS > Higher Is Better
CUDA ... 462.20 |==============================================================
OpenCL . 324.56 |============================================


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: MD5 Hash
GHash/s > Higher Is Better
CUDA ... 11.97 |===============================================================
OpenCL . 11.84 |==============================================================


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: Max SP Flops
GFLOPS > Higher Is Better
CUDA ... 9366.80 |=============================================================
OpenCL . 9322.01 |=============================================================


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: Bus Speed Download
GB/s > Higher Is Better
CUDA ... 12.51 |===============================================================
OpenCL . 12.53 |===============================================================


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: Bus Speed Readback
GB/s > Higher Is Better
CUDA ... 13.22 |===============================================================
OpenCL . 13.22 |===============================================================


SHOC Scalable HeterOgeneous Computing 2015-11-10
Benchmark: Texture Read Bandwidth
GB/s > Higher Is Better
CUDA ... 525.64 |==============================================================
OpenCL . 518.08 |=============================================================