Mitigating Critical Path Decompression Latency in Compressed L1 Data Caches Via Prefetching
Bibliographic record
Abstract
Increasing the size of cache memory is a common approach for reducing miss rates and increasing performance in a CPU. Doing this, however, increases the static and dynamic energy consumption of the cache. Compression can be utilized to increase the effective capacity of cache memory without physically increasing its size. We can also use compression to reduce the physical size of the cache, and therefore reduce its energy consumption, while maintaining a reasonable effective cache capacity. Unfortunately, a decompression latency is experienced when accessing the compressed data. This affects the critical execution path of the processor and can have a significant impact on performance, especially when implemented in L1 cache. Previous work has used cache prefetching techniques to hide the latency of lower level memory accesses. Our work proposes the combination of data prefetching and compression techniques to reduce the impact of decompression latency and improve the feasibility of compression in L1 caches. We evaluate the performance of Last Outcome (LO), Stride (S), and Two-Level (2L) prefetching, as well as hybrid combinations of these methods (S/LO & 2L/S), in combination with Base-Delta-Immediate (B Δ I) compression. The results demonstrate that using B Δ I, in combination with data prefetching, provides performance improvement over BΔI compression alone in L1 data cache. We find that a 4KB Hybrid S/LO prefetcher results in an average speedup of 1.7% and improvement to the energy-delay product of the CPU by 1.5% versus B Δ I alone.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 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 source (direct Gemma or distilled Codex), 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".