The Concepts Behind CUDA Optimization
Presented by Zafrir Patt at Core C++ meetup, the slides can be found at http://bit.ly/cpp14cuda.
In this session we will start looking inside the device. Study how it works and how its components running together. We will present the CUDA terminology, especially focusing on eligible/stalled/selected warps, warp divergence, warp schedulers, warp latency, hiding latency mechanism, occupancy, global memory coalesced access, shared memory bank conflict and how shall we use all this knowledge to maximize the device utilization.
Zafrir Patt
I graduated (B.Sc) in Mathematics & Physics and M.Sc. in Applied Mathematics from the Hebrew University of Jerusalem. I was working seven and a half years at Biosense Webster as 3D software team leader, developing OpenGl graphics, 3D model of the heart and tools for ventricular arrhythmia treatment.
In the last five years I am the software architect at VOLCANO-PHILIPS, designing and implementing C++ Modules, developing generic CUDA library used for various algorithms and focusing on optimization of the CPU & GPU code.
Recorded on Nov 27, 2019 in Tel Aviv, see event details at https://www.meetup.com/CoreCpp/events.... The meetup was hosted and sponsored by Final (http://www.final.co.il), a leading algorithmic trading company located in Israel.
![Selever Vs Corrupted BF - Mid-Fight Masses But It's Anime! Parte 2 [Fandub latino]](https://images.videosashka.com/watch/FFCHm8-hcp4)
