About us
Categories
- Agent (1)
- AMD (2)
- ascend (1)
- Behind the Build (1)
- Benchmark (12)
- Best practices (3)
- lmcache (8)
- New features (9)
- News (17)
- NVIDIA (2)
- Performance (7)
- Tech Explained (2)
- Tutorial (10)
Tags
Follow us on: X, LinkedIn
Initiated and Officially Supported by Tensormesh
Category: Performance
-
A practitioner’s guide to KV-cache tiering on ROCm — what works, what doesn’t, and the regime where it actually matters. Key Summary We benchmarked multi-turn agentic workloads using 739 anonymized Claude Code conversation traces from kv-cache-tester against MiniMax-M2.5 (230 GB FP8 MoE) on 2× AMD MI300X with vLLM 0.19.0 + LMCache (built from source for…
-
Modern LLM serving workloads are defined by strict latency requirements, high concurrency, and rapidly growing context lengths. Applications such as multi-turn chat, AI agents, and retrieval-augmented generation continuously build on prior context, leading to substantial reuse of previously computed states. In production, systems must minimize time-to-first-token (TTFT) while maintaining stable decoding throughput under heavy concurrent…
-
The standard approach to reducing LLM inference costs is prefix caching, which reuses previously computed token states to avoid redundant computation. In practice, however, this approach misses significant caching opportunities in real-world agentic workloads! Caching in Agentic Workflows In agentic workloads, shared content (e.g., retrieved contexts and documents) frequently appears across requests at varied positions,…
-
Baolong Mao (Tencent), Chunxiao Zheng (Tencent), Weishu Deng (Tensormesh), Darren Peng (Tensormesh), Samuel Shen (Tensormesh) What is P2P and what does it promise? In this blog post, we will go over: Most production vLLM deployments run multiple identical instances behind a load balancer. Each instance builds its own KV cache only from the traffic it…
-
The challenge: Scaling enterprise AI Enterprises today are racing to integrate large language models (LLMs) into their products and workflows, but doing it at scale brings challenges in performance, cost, and accuracy. Organizations need models to be based on their specific data, while making sure that this information remains private. Cohere, one of the leading…
-
We’re thrilled to announce that the Nvidia Dynamo project has integrated LMCache as its KV caching layer solution. This is a big milestone: Dynamo gets a battle-tested caching solution, and LMCache becomes part of a production-scale ecosystem used by many developers worldwide. Why KV Caching Matters KV caching is a foundational optimization for modern LLM…
-
TL;DR LMCache, the state-of-the-art KV cache layer library developed by TensorMesh and the project’s open-source community, delivers breakthrough performance improvements to modern enterprise LLM inference frameworks, including the vLLM Production Stack, KServe, and NVIDIA’s Dynamo. With fast and scalable caching of long-context KV cache, LMCache helps reduce inference costs and ensures SLOs for both latency…
