English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Open

Presents the pdf analysis as premium feature

PDF Analysis

Insights

Presents the summarisation as premium feature

Summarisation

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

computing (HPC) [13] to mobile devices [21], making it undesirable to uniformly increase ECC strength as an answer to higher fault rates. It is preferable to protect data selectively by applying strong ECC to high-risk memory regions only, while cheaper protection can be used on lower risk regions. For this to be possible, we need to develop methodologies to automatically and dynamically quantify the vulnerability of the different portions of data stored in memory before a program executes. Prior work on measuring memory vulnerability relies on offline application profiling [10, 11, 15, 36], but online methods are needed for widespread deployment. At the same time, the growing complexity of HPC machines increasingly impedes their programmability, which has motivated the emergence of novel task-based dataflow programming models [3, 6, 9, 18, 25]. These models facilitate the programmer’s work by providing abstractions (tasks and data dependencies) and supporting software (the runtime system). With an initial focus on scheduling work transparently, runtime systems are used for many dynamic optimisations, such as prefetching or partitioning caches [17, 26]. Our key idea is to exploit the opportunity presented by runtime systems to estimate memory vulnerability online, and to allow adapting ECC protection dynamically using these estimates. We evaluate the vulnerability of the memory pages allocated by each parallel workload using the programming model’s dataflow dependencies, which are expressed in the source code and exposed to the runtime system. This allows the runtime system to predict the future data accesses to each memory location, which are a key factor in the probability of encountering an error at this location. For example, data that is overwritten by subsequent stores or that is never consumed by an application has no impact on program state. The ability to detect this kind of behaviour online is then exploited to target the most vulnerable memory pages for increased ECC protection using the VECC scheme [35], with the aim of reducing the failure rate for any storage overhead setting. Abstract— Diminishing reliability of semiconductor technologies and decreasing power budgets per component hinder designing next-generation high performance computing (HPC) systems. Both constraints strongly impact memory subsystems, as DRAM main memory accounts for up to 30 to 50 percent of a node’s overall power consumption, and is the subsystem that is most subject to faults. Improving reliability requires stronger error correcting codes (ECCs), which incur additional power and storage costs. It is critical to develop strategies to uphold memory reliability while minimising these costs, with the goal of improving the power efficiency of computing machines. We introduce a methodology to dynamically estimate the vulnerability of data, and adjust ECC protection accordingly. Our methodology relies on information readily available to runtime systems in task-based dataflow p rogramming m odels, a nd t he e xisting Virtualized Error Correcting Code (VECC) schemes to provide adaptable protection. Guiding VECC using vulnerability estimates offers a wide range of reliabilityredundancy trade-offs, as reliable as using expensive offline profiling f or g uidance a nd u p t o t o 2 5% safer than VECC without guidance. Runtime-guided VECC is more efficient than a stronger uniform ECC, reducing DIMM lifetime failure from 1.84% down to 1.26% while increasing DRAM energy consumption by only 1.03×.

Runtime-guided ECC protection using online estimation of memory vulnerability