Reply by Andor November 13, 20082008-11-13
On 13 Nov., 20:43, SYL <sya...@gmail.com> wrote:
> On Nov 13, 7:11&#4294967295;am, Andor <andor.bari...@gmail.com> wrote: > > > On 13 Nov., 08:11, ad...@ng2000.com wrote: > > > >http://www.ng2000.com/blog/2008/11/04/digital-signal-processing/ > > > Too late, in my opinion. The clock-rate increase in the SHARC family > > has been much too slow (100MHz in 1996, 450MHz in 2008) to make SHARCs > > a serious contender for large- and medium-scale signal processing > > hardwares. For small-scale applications, BLACKFINs or similar are > > better suited. > > Agreed. Despite how much love I have in Sharc, we have decided to use > TI's c674x and OMAP L1x for our next line of products, and probably > for all future development. SHARC is way more expensive, only has a > tiny cache and has no dual core in the roadmap.
Well the SHARC has been kinda of "dual core" since the SIMD generation (as of 1996 with the 21161N). I think there was a quad-SHARC processor sometime (pre 1996), but ADI decided to stay with single cores equipped with high-speed link connections. However, this is just not comparable in performance (design time, programming time, debuggin time and computation power) with multi-core processors, so I agree with you. The Cell processor shows the way for heavy-duty signal processing. Regards, Andor