Proceedings of the 48th Design Automation Conference 2011
DOI: 10.1145/2024724.2024940
|View full text |Cite
|
Sign up to set email alerts
|

Matching cache access behavior and bit error pattern for high performance low Vcc L1 cache

Abstract: Cache is a roadblock towards low supply voltage (Vcc). It is mainly because low Vcc incurs process variation-induced bit errors in large SRAM in cache. Existing approaches for low Vcc cache suffer from low performance due to reduced effective capacity, long latency to correct errors, and increased misses due to accesses to faulty words. In our work, we propose a word-level sub-block disable-based method which increases the utilization of available cache capacity. Our key idea is to minimize accesses to faulty … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
5
0

Year Published

2012
2012
2022
2022

Publication Types

Select...
6
3

Relationship

0
9

Authors

Journals

citations
Cited by 11 publications
(5 citation statements)
references
References 19 publications
0
5
0
Order By: Relevance
“…Techniques based on replacing faulty cache entries [Wilkerson et al 2008;Chishti et al 2009;Ansari et al 2009;Sasan et al 2009] introduce significant overheads due to bypassing and signal rerouting. Likewise, techniques based on simply disabling faulty storage [Wilkerson et al 2008;Abella et al 2009;Choi et al 2011;Khan et al 2013] may provide noticeable performance variation for a given program depending on the faults' location because the distribution of faulty bits is random. Such techniques are shown effective from an average performance perspective and provide functional correctness, but fail to provide any guarantees on available cache size, number of sets, number of ways, etc., thereby making them useless in real-time scenarios where WCET estimation requires full knowledge of the hardware features below to provide strong timing guarantees [Wilhelm et al 2008].…”
Section: State-of-the-art On Low-energy Cache Designsmentioning
confidence: 99%
“…Techniques based on replacing faulty cache entries [Wilkerson et al 2008;Chishti et al 2009;Ansari et al 2009;Sasan et al 2009] introduce significant overheads due to bypassing and signal rerouting. Likewise, techniques based on simply disabling faulty storage [Wilkerson et al 2008;Abella et al 2009;Choi et al 2011;Khan et al 2013] may provide noticeable performance variation for a given program depending on the faults' location because the distribution of faulty bits is random. Such techniques are shown effective from an average performance perspective and provide functional correctness, but fail to provide any guarantees on available cache size, number of sets, number of ways, etc., thereby making them useless in real-time scenarios where WCET estimation requires full knowledge of the hardware features below to provide strong timing guarantees [Wilhelm et al 2008].…”
Section: State-of-the-art On Low-energy Cache Designsmentioning
confidence: 99%
“…Likewise, techniques based on disabling faulty cache entries [21], [1], [7] fail to provide strong timing guarantees required for the worst-case execution time (WCET) estimation, as needed for critical applications in our target market [20]. A failure to perform an operation correctly and within a given time may have catastrophic consequences in these environments.…”
Section: Related Workmentioning
confidence: 99%
“…First, there have been many studies on SRAM cache architectures for low Vccmin [17][18][19][20][21][22]. In low-Vcc SRAMs, process variation induces bit errors increase.…”
Section: Related Workmentioning
confidence: 99%