AMD has never been used synchronous multi-threaded (SMT) is also referred to as the Intel Hyper-Threading Technology. Although such a technology in the P4 era when it is of no practical use, but now, the increasingly popular multi-threaded environment for the re-HT rejuvenated. Application of the current environment, Hyper-Threading Technology allows a number of specific application significantly speed up to 10-15%. In addition to Intel’s Nehalem, Atom, such as the introduction of Hyper-Threading, regardless of IBM’s Power series, Sun’s T1/T2/Rock series application processor architectures have a similar multi-threaded SMT synchronization technology, with a small number of transistors with to large multi-threaded performance.
AMD engineers who candidly to the media a few days ago, does not support multi-threaded single-core Opteron processor technology, Intel appears to the low-end performance than Xeon. Allegedly, AMD has admitted that an internal high-level, there is no early introduction of such technology is the choice of a technical mistake.
However, AMD vice president of server and workstation business general manager Patrick Patla an interview a few days ago, said there is no clear single-core multi-threading technology in the future, but continue to reiterate the Opteron has announced the road map: „If you look at our road map and our multi-threaded processor performance of the market will know, we believe that each thread has a complete core is the best choice. in 2010, we will launch 12-core processor, 16 core in 2011.
We believe that within the next few years we will be able to improve the support of 48 or 64-thread environment, let us take a look at the 2012-2013 annual meeting what you bring. “
Since 2011, only 16 nuclear, then the 2012-2013 jump directly to 48 or even 64 nuclear does not seem so normal. In addition, Patrick Patla sentence in speaking in front of „nuclear“ and to the back and turned into a „thread“, it seems to imply that AMD may be time to adopt single-core multi-threading technology.