2010
DOI: 10.1587/transinf.e93.d.2635
|View full text |Cite
|
Sign up to set email alerts
|

DDR3 SDRAM with a Complete Predictor

Abstract: SUMMARYIn the arsenal of resources for improving computer memory system performance, predictors have gained an increasing role in the past few years. They enable hiding the latencies when accessing cache or main memory. In our previous work we proposed a DDR SDRAM controller with predictors that not only close the opened DRAM row but also predict the next row to be opened. In this paper we explore the possibilities of trying the same techniques on the latest type of DRAM memory, DDR3 SDRAM, with further improv… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
7
0

Year Published

2014
2014
2014
2014

Publication Types

Select...
1

Relationship

1
0

Authors

Journals

citations
Cited by 1 publication
(7 citation statements)
references
References 4 publications
0
7
0
Order By: Relevance
“…Namely, the best variant of the zero live time predictor from [1], signed as 2b16, has two bits for every 16 adjacent DRAM rows in the system, used as a saturated counter, with values from 0 to 3. For the adopted 2 GB DDR3 SDRAM structure from [1] with total of 16 banks in the system and 8 k rows per bank, the total number of rows in the system is 128 k. For that number of rows we need 16 kb or 2 kB of SRAM memory. Another important thing considering the implementation are the time intervals needed to execute the two basic operations of the predictors: Lookup and Update.…”
Section: Behaviormentioning
confidence: 99%
See 4 more Smart Citations
“…Namely, the best variant of the zero live time predictor from [1], signed as 2b16, has two bits for every 16 adjacent DRAM rows in the system, used as a saturated counter, with values from 0 to 3. For the adopted 2 GB DDR3 SDRAM structure from [1] with total of 16 banks in the system and 8 k rows per bank, the total number of rows in the system is 128 k. For that number of rows we need 16 kb or 2 kB of SRAM memory. Another important thing considering the implementation are the time intervals needed to execute the two basic operations of the predictors: Lookup and Update.…”
Section: Behaviormentioning
confidence: 99%
“…Although this predictor seems rather complex, with lots of different hardware elements, the results show that the implementation of the best dead time predictor from [1], signed as sep2, demands only 17,718 equivalent gates. The number of 4-input LUTs used for this predictor is 1,404.…”
Section: Hardware Cost Estimationmentioning
confidence: 99%
See 3 more Smart Citations