ACM Home Page
Please provide us with feedback. Feedback
Exploiting forwarding to improve data bandwidth of instruction-set extensions
Full text PdfPdf (716 KB)
Source Annual ACM IEEE Design Automation Conference archive
Proceedings of the 43rd annual conference on Design automation table of contents
San Francisco, CA, USA
SESSION: Session 4: processor and communication centric SOC design table of contents
Pages: 43 - 48  
Year of Publication: 2006
ISBN:1-59593-381-6
Authors
Ramkumar Jayaseelan  National University of Singapore
Haibin Liu  National University of Singapore
Tulika Mitra  National University of Singapore
Sponsors
SIGDA: ACM Special Interest Group on Design Automation
ACM: Association for Computing Machinery
Publisher
ACM  New York, NY, USA
Bibliometrics
Downloads (6 Weeks): 5,   Downloads (12 Months): 68,   Citation Count: 2
Additional Information:

abstract   references   cited by   index terms   collaborative colleagues  

Tools and Actions: Review this Article  
Save this Article to a Binder    Display Formats: BibTex  EndNote ACM Ref   
DOI Bookmark: Use this link to bookmark this Article: http://doi.acm.org/10.1145/1146909.1146924
What is a DOI?

ABSTRACT

Application-specific instruction-set extensions (custom instructions) help embedded processors achieve higher performance. Most custom instructions offering significant performance benefit require multiple input operands. Unfortunately, RISC-style embedded processors are designed to support at most two input operands per instruction. This data bandwidth problem is due to the limited number of read ports in the register file per instruction as well as the fixed-length instruction encoding. We propose to overcome this restriction by exploiting the data forwarding feature present in processor pipelines. With minimal modifications to the pipeline and the instruction encoding along with cooperation from the compiler, we can supply up to two additional input operands per custom instruction. Experimental results indicate that our approach achieves 87--100% of the ideal performance limit for standard benchmark programs. Additionally, our scheme saves 25% energy on an average by avoiding unnecessary accesses to the register file.


REFERENCES

Note: OCR errors may be found in this Reference List extracted from the full text article. ACM has opted to expose the complete List rather than only correct and linked references.

1
 
2
 
3
4
 
5
 
6
Altera Corp. Nios processor reference handbook.
 
7
 
8
 
9
Xilinx Inc. Microblaze soft processor core.
 
10
R. Jayaseelan, H. Liu, and T. Mitra. Exploiting forwarding to improve the data bandwidth of instruction-set extensions. Technical Report TRB5/06, School of Computing, National University of Singapore, 2006.
11
 
12
13
 
14
P. Shivakumar and N. P. Jouppi. CACTI 3.0: An integrated cache timing, power and area model. Technical Report 2001/2, Compaq Computer Corporation, 2001.
15
16


Collaborative Colleagues:
Ramkumar Jayaseelan: colleagues
Haibin Liu: colleagues
Tulika Mitra: colleagues