Method Drift›KV-cache compression
TurboRAG
KV-cache compression
superseded — cited as a baseline and beaten by newer methods
3 papers critique it · 2 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites TurboRAG as a baseline.
“block-based methods like PromptCache~Gim2023PromptCM, TurboRAG~lu2024turborag, and BlockAttention~ma2024block ignore inter-block dependencies, causing performance degradation”
— TableCache: Primary Foreign Key Guided KV Cache Precomputation for Low Latency Text-to-SQL“However, it still falls short in effectively reconnecting reused caches.”
— KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse“Reuse strategies effective for agent-side generation can be decision-non-invariant for judges, revealing a failure mode overlooked by prior work.”
— When KV Cache Reuse Fails in Multi-Agent Systems: Cross-Candidate Interaction is Crucial for LLM Judges
Beaten on benchmarks
Head-to-head results where a newer method reports beating TurboRAG. Values are copied from the source paper's tables — verify against the cited paper.
- TableCache: Primary Foreign Key Guided KV Cache Precomputation for Low Latency Text-to-SQL
TableCache beats TurboRAG · TTFT [OmniSQL-7B]
36.22 vs 81.47
- TableCache: Primary Foreign Key Guided KV Cache Precomputation for Low Latency Text-to-SQL
TableCache beats TurboRAG · TTFT [Qwen2.5-7B-coder]
42.01 vs 77.40
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · NQ [Llama-3.2-1B]
45.0 vs 43.4
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · 2WikiMQA [Llama-3.2-1B]
66.0 vs 65.5
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · TriviaQA [Llama-3.2-1B]
66.3 vs 64.8
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · HotpotQA [Llama-3.2-1B]
55.6 vs 51.8
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · Musique [Llama-3.2-1B]
19.2 vs 15.2
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · NQ [Llama-3.2-3B]
64.4 vs 62.9
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · 2WikiMQA [Llama-3.2-3B]
71.2 vs 69.5
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · TriviaQA [Llama-3.2-3B]
73.7 vs 72.9
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · HotpotQA [Llama-3.2-3B]
69.5 vs 65.6
- KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse
KVLink beats TurboRAG · Musique [Llama-3.2-3B]
35.8 vs 31.4
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.
- May 22, 2026
- Apr 28, 2026
- Predictive Multi-Tier Memory ManagementPredictive Multi-Tier Memory Management for KV Cache in Large-Scale GPU InferenceApr 19, 2026
- TableCacheTableCache: Primary Foreign Key Guided KV Cache Precomputation for Low Latency Text-to-SQLJan 13, 2026
- Jan 5, 2026
- SemShareKVSemShareKV: Efficient KVCache Sharing for Semantically Similar Prompts via Token-Level LSH MatchingSep 29, 2025