Blogs

Changes in SEG-Y Revision 2

Changes in SEG-Y Revision 2

GPU Hardware System Engineering – A Debugging Story

GPU Hardware System Engineering – A Debugging Story

The Performance Problem

Choosing a Normalization Method

Choosing a Normalization Method

Assessing Correctness in Scientific High-Performance Computing

Assessing Correctness in Scientific High-Performance Computing

Real-Time NUMA Node Performance Analysis Using Intel Performance Counter Monitor

Real-Time NUMA Node Performance Analysis Using Intel Performance Counter Monitor

Investigating C++ ’17 and if constexpr

Investigating C++ ’17 and if constexpr

CIFAR-10 Genetic Algorithm Hyperparameter Selection using TensorFlow

CIFAR-10 Genetic Algorithm Hyperparameter Selection using TensorFlow

Timeout Detection in the Windows Display Driver Model when Running CUDA Kernels: Symptoms, Solutions, and Registry Modifications

Timeout Detection in the Windows Display Driver Model when Running CUDA Kernels:  Symptoms, Solutions, and Registry Modifications

A Simple Trick To Pass Constant Arguments Into GPU Kernels

A Simple Trick To Pass Constant Arguments Into GPU Kernels

Most CUDA developers are familiar with methods of passing constant arguments into GPU kernels.  The simplest method is directly via kernel parameters and the other option is copying to constant memory.  Under certain circumstances though, there’s another lesser-known way to get constants into your GPU kernel, that may even improve kernel performance!  

Unified Memory on Tesla P100 with CUDA 8.0

Unified Memory on Tesla P100 with CUDA 8.0

Pages

Subscribe to RSS - blogs