Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
You should reformulate your question .
What is the difference between a DSP and a standard non dsp PROCESSOR .. Both devices are CPUS .. the difference between a DSP and a non dsp processor is that the DSP has an archirecture that allow reperititive operations to be coded an executed very efficiently .. One of those operations is multiplication and then an addition .It can take place in a specialized unit called MAC .. or multiply/accumulate . Is even possible to have several of this units in PARALLEL some DSP from TI have 4 to 6 MACs. So they execute 6 operations in PARALLEL .. thats a tremendous TROUGHPUT .. regular CPU's .. can at most crack 2 instructions per cycle .they are called superscalar CPU's .. they are very high end ..like the i960CA and the MIS6000 .. and other wild and not very comun beasts .. otherwise they execute bearly one instruction .. they are just SCALAR RISC cpu's
hi,
DSP CPUs (also most of the modern risc cpus) have Harvard Architecture .. ie it has seperate data and instruction memory.. while General Purpose processors are based on Von-Nuemann architecuture ..
I have read a paper on this some time back.. I am searching for it .. if I get it , will upload
Simple:
DSP - Digital Signal Procesor - specialy build for processing digital dignals, it's optimised for prcesings digital signals, so it usualy have buit in routines for processing signals.
DSP (Harvard Architecture) is good for dedicated task like digital filtering where sampling rate should not be violated. But not a choice for multitasking. Each instruction is executed in a single cycle. A typical instruction consists of a instruction pre fetch, multiply accumulate and post memory write operation.
General purpose Microprocessor (RISC or CISC) is good for multitasking.
I synthesized a DSP digital core where the MAC costs quarter area of this DSP whole digital part. And the MAC is also the critical path. I think it is one key difference.
just a very brief answer, this is not nearly exhaustive and the lines between differetn kinds of processors are somewhat blury
DSPs are processor targeted at signal processing applications. since signal processing usually requires lots of multiplication and accumulation, many DSPs feature hardware multiply accumulate (MAC) units.
there are many different kinds of instruction set architectures (ISAs) used for DSPs. RISC (e.g. AD), CISC (TI C54x), VLIW (TI C6x).
also the memory architectures are varying. most DSPs used to have separate data and program memories (harvard style) but meanwhile there are DSPs where memory can hold both code and data.
DSPs can serve as central processing unit (CPU) in devices, but often the central controler today is an ARm or similar microprocessor.
to make processing fast, many DSPs excute instructions in just 1 clock
cycles. Also, DSP has lots of other modules on the bus, such as
CRC, decoder, shifters, multipliers.... in order to use hardware to
accelerate the process.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.