←Story detail

Trending story

Shan Donglai, AI Storage Product Director at MacroSAN Technology, Shares KV Offloading Practices

1 reports1 story updatesUpdated Oct 30 16:03

Understand this story

AI overview

On October 30, 2026, at the 2026 Data Storage Industry Conference, Shan Donglai, AI Storage Product Director at MacroSAN Technology, delivered a keynote speech titled "Application Practices of External Storage in KV Offloading." At the same event, MacroSAN Technology showcased the Alpha700 AI-dedicated storage, which alleviates the GPU HBM memory bottleneck through two tiered KV offloading modes: L3.5 low latency and L4 high throughput. In real-world testing on a large manufacturing enterprise's agent platform, the L3.5 solution raised the global KV Token hit rate from 38% to 79%, reduced TTFT P50 from 1200 milliseconds to 520 milliseconds, and cut Agent response time by more than 40%. MacroSAN Technology also participated as a core co-drafting unit in the formulation of two AI storage group standards, including "Capability Requirements for Artificial Intelligence Storage Systems."

AI-generated from public reports

Latest development · Report

MacroSAN Showcases at the 2026 Data Storage Industry Conference: Alpha700 Accelerates Large Model Inference with KV Cache Offloading

Developments

1 confirmed developments

  1. Shan Donglai, AI Storage Product Director at MacroSAN Technology, Shares KV Offloading Practices

    Source:China.com Technology

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Oct 30

  1. China.com Technology

    MacroSAN Showcases at the 2026 Data Storage Industry Conference: Alpha700 Accelerates Large Model Inference with KV Cache Offloading

    At the 2026 Data Storage Industry Conference, MacroSAN showcased Alpha700 AI-dedicated storage, which alleviates the GPU HBM memory bottleneck through two tiered KV offloading modes: L3.5 low-latency and L4 high-throughput. In real-world testing on the agent platform of a large manufacturing enterprise, the L3.5 solution raised the global KV token hit rate from 38% to 79%, reduced TTFT P50 from 1,200 milliseconds to 520 milliseconds, and cut Agent response time by more than 40%. MacroSAN also participated as a core co-drafting unit in the formulation of two AI storage group standards, including "Artificial Intelligence Storage System Capability Requirements."

Related stories

  1. Related storiesMusk Reveals Tesla and SpaceX Formed Special Committees to Evaluate Merger
  2. Related storiesDeepSeek open-sources the inference engine DeepInfer on GitHub
  3. Related storiesCohere completes first close of Series E, targeting up to C$7 billion
  4. Related storiesAnthropic Publicly Submits S-1 Registration Statement for Proposed IPO on Nasdaq