AMD Announces New GPGPU Programming Tools

AMD's GPU solutions have come a long way since the company acquired ATI. The combined companies have competed very well against Nvidia for the past several years, at least at the consumer level. When it comes to HPC/GPGPU tools, however, Nvidia has had the market all to itself. Granted, the GPGPU market hasn't exactly exploded, but Nvidia has sunk a great deal of effort into developing PhysX and CUDA.

AMD has announced a new suite of programming tools it plans to use to woo developers in the burgeoning field. "AMD is working closely with the developer community to make it easier to bring the benefits of heterogeneous computing to consumers, enabling next-generation system features like vivid video, supercomputer-like performance and enhanced battery life,” said Manju Hegde, corporate vice president, AMD Fusion Experience Program. “Our advanced developer tools and solutions enable a new era of parallel programming that’s based on industry standards and focused on delivering innovative user experiences that span a variety of computing form factors."

One of the new tools, gDEBugger, is an "advanced OpenCL and OpenGL debugger, profiler, and memory analyzer." AMD acquired the software when it purchased graphics startup Graphic Remedy in October 2010. Other programs, which aren't expected until Q3 2011, include Parallel Path Analyzer (useful for optimizing a program to run across both CPU and GPU), Global Memory for Acceleration (simplifies memory management) and the Task Manager API. This last is described as "a framework for managing compute tasks in a heterogeneous multi-core environment. OpenCL kernels can be automatically scheduled to execute on an available and task-appropriate device, providing dynamic load balancing, optimizing use of available compute resources and removing the burden of explicit schedule handling."

gDEBugger, hard at work

AMD's interest in GPGPU computing has grown significantly in the past six months, due in no small part to the fact that the company's Fusion APU's are finally shipping. With Llano's launch date approaching, AMD's interest in heterogenuous computing is about to transition from theoretical to practical and necessary. As such, the company's willingness to spend R&D funds to promote APU-style programming is trending upwards.

AMD had little choice but to take the route it has. Nvidia's overall performance since it launched the G80 in 2006 has been quite strong. AMD, in contrast, has spent much of the past five years on life support. Even now that the company has returned to profitability, its margins have not been high enough to allow it to chase GPGPU computing to the same degree.

The question now is whether or not Nvidia's years of work will give the company an early adopter advantage. On the one hand, the number of CUDA applications and/or games that take advantage of hardware PhysX has been relatively small. On the other hand, Nvidia has spent a significant amount of time/money on Tesla and HPC GPU applications. AMD's focus will be very different--the two Fusion products shipping this year are Brazos and Llano. Team Red has its own Stream initiative, but again, we expect it'll take quite some time to ramp it.