MB-FICA: An ADL framework for multi-bit fault injection and coverage analysis
Bibliographic record
Abstract
Safety-critical systems (SCS) may experience soft errors due to upsets caused by externalevents such as cosmic rays, packaging radiation and thermal neutrons. Traditional errormodeling techniques often only address single bit corruptions and analysis based on populartechniques such as architectural vulnerability factor (AVF) treat each bit as independent.However recent studies have shown a dramatic increase in multi-bit upset (MBU) wherethe failure of a single bit is highly correlated with its neighboring bits. This phenomenon isdue to shrinking transistors and resulting increases in transistor density, making a particlestrike capable of corrupting multiple bits at a time. To assist designers with MBU mitigation in microprocessor register les (RF), we havedeveloped a novel framework (available at http://bhm.ece.mcgill.ca/~mb-fica) to simulate and analyze the eect of MBU and the eectiveness of fault tolerance techniques.Unlike the prior work, our approach performs fault injection in microarchitecture includingmitigation technologies and simulates the consequent behavior of the system running various benchmarks. In this framework, we consider (a) the eect of SRAM layout on MBUpatterns, (b) the data-dependent nature of transient upsets, and (c) runs benchmarks tocompletion to accurately evaluate fault coverage under dierent mitigation techniques. Fault injection is computationally expensive, especially in the context of MBU; consequently, we propose a suite of fault injection acceleration techniques that reduce theexecution time of individual trials by only simulating mitigation techniques when faults arepresent, and stopping simulation entirely when all errors have been detected or corrected.When evaluating parity, SECDED, and 2-bit 2D ECC, our results demonstrate a speedupin the fault injection performance of 14x on average, and up to nearly 60x in one case.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".