While AMD is certainly not proclaiming the death of the standalone processor, the company believes it is at the end stages of the "one size fits all" computing model.
Eighteen months after first announcing its "accelerated computing" initiative aimed building out an ecosystem of third party-enabled products that compliment and enhance its own processor platform; Advanced Micro Devices has about two dozen partners developing new products.
Although only few commercial products have emerged as yet from the accelerated computing initiative announced by AMD in June 2006, the company believes it has moved the effort past the first phase of a three-stage strategy to make specialized processing engines more commercially viable, according to Doug O'Flaherty, division manager for acceleration strategies at AMD.
"While we are certainly not proclaiming the death of the standalone processor, AMD believes we are at the end stages of the 'one size fits all' computing model," O'Flaherty said. "The ability to tailor parts of your platform with specialized silicon or software to gain specific application performance benefits will be a reality of the next generation of silicon."
The first phase of AMD's effort has been a bit of a "chicken and egg" effort as the company has attempted to promote its approach and worked with industry bodies to define standards that will enable tighter integration of processors such as AMD's Opteron with co-processors, graphics processor and other logic arrays that can improve overall performance and efficiency.
Announced "acceleration" products include Activ Financial System's use of the open-socket Torrenza platform that combines AMD's Opteron processor with a field programmable gate array that provides hardware acceleration specifically for running critical financial software applications. Hewlett-Packard has said some "select" upcoming quad-core Opteron-based servers will provide support that will enable end-users to deploy customized acceleration solutions.
AMD believes it is entering the second phase of its accelerated computing effort and that there will be much broader support and commercialization of customized and optimized processor chip-sets over the next 12 to 18 months. In the third phase, rapid adoption would make the third-party enabled platforms significant contributors to AMD's bottom line.
Nathan Brookwood, an analyst with Insight 64, said he has remained "pretty consistently skeptical" about AMD's ability to make specialized accelerated processing platforms into significant mass market products, but believes the approach is a reasonable strategy for markets where the specialized improvements are critical to success, such as high-performance computing, academia and security.
"The effort has began to find some homes in niches where the people inside are really smart and have the time to invest effort in special purpose hardware," Brookwood said. "The kinds of things you can do with these extensions are very specialized and deal with signal processing or making specific algorithms go faster, but for most people it's not going to merit the effort and will fall to the bottom of the list of things they could potentially do."
But AMD believes as demands for peak performance at higher efficiencies will necessitate the move to accelerated computing platforms, and that its early efforts to work with equipment designers and innovative chip developers will pay off in the years ahead.
"We believe that freedom of choice is important, and that's what accelerated computing is all about," O'Flaherty said. "We are offering choice is how you can go about solving technical problems. We've got the early adopters and we are continuing to work to identify markets that have very specific and real needs that can be more adequately addressed by this approach."
2014 Next-Gen WAN SurveyWhile 68% say demand for WAN bandwidth will increase, just 15% are in the process of bringing new services or more capacity online now. For 26%, cost is the problem. Enter vendors from Aryaka to Cisco to Pertino, all looking to use cloud to transform how IT delivers wide-area connectivity.
The UC Infrastructure TrapWorries about subpar networks tanking unified communications programs could be valid: Thirty-one percent of respondents have rolled capabilities out to less than 10% of users vs. 21% delivering UC to 76% or more. Is low uptake a result of strained infrastructures delivering poor performance?