Analog Devices Inc., ADSP-2101/2 User's Manual, 1988.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the tools generation problem. We have developed the processor models for the PowerPC 603[12] Motorola 68HC11[13] Intel 8085[14] processors. Various tools developed work for all such processors. Currently we are also developing processor models for Sun Sparc[15] ARM[16] DLX[17] and ADSP 2100 [18] series processors. Much of the information about the work is available at http: www.cse.iitk.ac.in sim nml. Various tools are also available in public domain downloadable from the same site. Acknowledgement The author would like to acknowledge Prof. S.K. Aggarwal, Prof. Deepak Gupta, V. ....

Analog Devices Inc., ADSP-2101/2 User's Manual, 1988.

....need to support certain types of arithmetic very efficiently; and ii) the need for deterministic execution times. In this section, I give an overview of the characteristics of these devices. I focus only on the modern, floating point devices; for more detailed information on specific devices, see [115, 119, 34, 5, 6, 98, 138, 137]. Figure 2.5 illustrates a simplified architecture of a typical floating point DSP core. The device contains six functional units an ALU, a multiplier, two address units, and two loadstore units (not shown here) all of which can operate simultaneously, a bank of floatingpoint data registers, ....

....(finite impulse response) filter, a key DSP benchmark. Some devices also have a parallel multiply add subtract instruction, which substantially speeds up execution of the FFT (Fast Fourier Transform) another key benchmark. In contrast to the TMS320C30 and TMS320C40, the ADSP 21020 and DSP96002 [5, 98] are load store machines: operands are explicitly loaded into data registers prior to operating on them. The programmer writes separate fields of the instruction to control the ALU and multiplier, and the load store and address units. For example, a typical parallel multiplyadd instruction in ....

Analog Devices. ADSP-21020 User's Manual, 1991.

....need to support certain types of arithmetic very efficiently; and ii) the need for deterministic execution times. In this section, I give an overview of the characteristics of these devices. I focus only on the modern, floating point devices; for more detailed information on specific devices, see [115, 119, 34, 5, 6, 98, 138, 137]. Figure 2.5 illustrates a simplified architecture of a typical floating point DSP core. The device contains six functional units an ALU, a multiplier, two address units, and two load store units (not shown here) all of which can operate simultaneously, a bank of floating point data ....

....(finite impulse response) filter, a key DSP benchmark. Some devices also have a parallel multiply add subtract instruction, which substantially speeds up execution of the FFT (Fast Fourier Transform) another key benchmark. In contrast to the TMS320C30 and TMS320C40, the ADSP 21020 and DSP96002 [5, 98] are load store machines: operands are explicitly loaded into data registers prior to operating on them. The programmer writes separate fields of the instruction to control the ALU and multiplier, and the load store and address units. For example, a typical parallel multiply add instruction in ....

Analog Devices. ADSP-21020 User's Manual, 1991.

....Throughout the paper, we will generally talk of storage resources (SRs) which comprise RFs and memories. We only talk of RFs, if it is important to distinguish between register files and memories. Example: In order to exemplify typical constraints, we consider the ADSP 210x fixed point DSP [16]. The primitive entities of instruction behavior in our model are register transfers (RTs) A RT reflects the operation ( performed, the SR where the result of the operation is stored, and the SRs where the operands of 4 the operation have to reside. In fig. 1 a partial data ....

....globally optimal code generation for a DFG will be possible within a reasonable amount of computation time. Furthermore, application of algebraic rules during code generation is another potential area for improvements. Notes 1. The numbering of types corresponds to the ADSP 210x User s Manual [16]. 2. A very good summary of first approaches using grammars and attributed grammars for specifying code selectors can be found in [25] 3. The notion tree reduction rules reflects a view of the tree covering process as the reduction of a certain input tree to a certain nonterminal, by using the ....

Analog Devices. ADSP-2101 User's Manual. Analog Devices, 1991.

....and experimental results are given in Section 7; Section 8 concludes and provides an outlook. 2 Architecture In the scope of our paper, we are considering as target architecture a 32 bit digital signal processor with load store architecture, the ADSP 2106x (super harvard architecture computer) [Ana96, Ana95c, Ana91, Ana95a, Ana95b]. Its core processor consists of the register file, three functional units, a control unit, two address generators (DAG1 and DAG2) a timer and the instruction cache (see figure 1) Data can be transported via three buses: PM , DM and I O bus, which provide connection to the program memory, 1 ....

Analog Devices. ADSP-21020 User's Manual, 1991.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC