Thoughts on GPUTech 2016

on Apr 29, 16 • by Martin Bakal • with No Comments

Marty Bakal, dynamic analysis product manager, tells us what he saw at GPUTech 2016 recently...

Home » High Performance Computing » Thoughts on GPUTech 2016

As a first time participant at the GPU Technology Conference a few weeks ago, I was impressed by the size of the conference and the number of interactive things that attendees could see and do. They had complete simulations where people could walk around on Mars, see robots built using GPUs, and much more. I also attended a number of interesting talks on the future of GPUs with the soon to be released CUDA 8.0 and NVIDIA’s new Pascal hardware, a significant evolution from their current Maxwell architecture that we’ve all been looking forward to.

Where we play

I also had the opportunity to talk to both existing and new TotalView customers as well as potentially new CodeDynamics users. Many of the developers I talked to were excited about the possibility of debugging their GPU code with their CPU code all in the same debug session, making their lives much easier by not having to switch from one context to the other all the time. TotalView gives you this functionality along with a whole host of additional capabilities for multithreaded and multiprocess debugging, replay, and memory debugging.

Attending conferences like this and meeting the community, especially one as enthusiastic and close-knit as this, always helps us improve our products as we look into the future. Overall, I feel the future is very bright for hybrid (GPU and CPU) debugging and we’ll be there to help along the way.

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *

Scroll to top