August 18, 2010
AMD's acquisition of ATI in 2006, by chance or by design, coincided with the meteoric rise of general-purpose GPU (GPGPU) computing, especially in HPC. As AMD struggled to bring the two entities under one roof, NVIDIA, undistracted by a merger, was able to capture the GPGPU high ground and has never looked back.
But AMD is still in the hunt. To counter NVIDIA's latest Fermi GPUs, AMD is about to release its new FireStream cards, which match the Fermis in overall performance, if not capability. In a nutshell, both of the chipmakers' hardware now offer about a half a teraflop (double precision) per chip. But that's mostly where the comparison ends. NVIDIA hardware has the advantage in of ECC memory support, local cache, asynchronous transfers, and a generally more sophisticated architecture geared for general purpose computing. AMD's offerings have the advantage of better performance per watt, at least for the 150 watt FireStream 9350 product.
NVIDIA also has big advantage in its CUDA programming environment, which has become the premier software platform for GPGPU development. AMD, on the other hand, is sticking with the open standard OpenCL, which is generally considered less capable and certainly much less mature than CUDA.
A recent article penned by Lawrence Latif of The Inquirer this week does a nice job of matching up the GPGPU competition and he makes a particularly interesting point about CUDA's long-term prospects:
CUDA might not be open, or even a standard, but history tells us that such technicalities never stopped other languages from attaining widespread popularity. Being policed by IBM didn't stop Fortran from still being the numerical language, half a century after it first appeared. Even with Sun Microsystems' best efforts to create a cumbersome 'framework' and employ licensing peculiarities, Java's popularity has managed to surpass C. It has happened before and it's looking like history will repeat itself.
Much of the rest of Latif's report is based on conversations with NVIDIA's Tesla product line manager Sumit Gupta and AMD's director of stream computing Patricia Harrell. And while there's certainly a lot of he-said-she-said in the text, the whole article is quite illuminating and worth reading in its entirety.
Full story at The Inquirer
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?