-
Book Overview & Buying
-
Table Of Contents
-
Feedback & Rating
The C++ Programmer's Mindset
By :
CPUs are immensely complicated and capable of processing vast quantities of data, but because of their general-purpose nature, they sometimes cannot provide the throughput necessary for some problems. For instance, real-time, low-latency computations are not well-suited to running on a CPU that is also running a full operating system. In these instances, one might turn to specialized accelerator hardware or coprocessors. Graphics processing units (GPUs) offer very high throughput compute and high bandwidth memory, and are perfect for problems that require vast number-crunching capabilities. Field programmable gate arrays (FPGAs) can be configured for very low latency and low power computation, which makes them good for time-critical or embedded applications. Programming for these devices often requires a different approach from general CPU programming.
This chapter has two parts and mostly focuses on GPU programming. In the first part...