Artificial Intelligence

TRUSTA Debuts Upgraded AI Scaler Toolkit at SuperComputing 2026

· GlobeWire Daily Editorial Desk · Source: www.prnewswire.com

2 min read

Key takeawaysAt SuperComputing 2026 (SC26) in Chicago, TRUSTA, the enterprise storage brand of ADATA, introduced an upgraded AI Scaler Toolkit designed to optimize enterprise AI inference. The solution integrates GPU, DRAM, and enterprise SSD resources to enhance performance and reduce reliance on expensive GPU hardware.
TRUSTA Debuts at SuperComputing 2026 with Upgraded AI Scaler Toolkit

Image: ADATA Technology Co. Ltd · Photo source · Resized and letterboxed to 16:9; no AI generation

Overview

At SuperComputing 2026 (SC26) in Chicago, TRUSTA, the enterprise storage brand of ADATA, introduced an upgraded AI Scaler Toolkit designed to optimize enterprise AI inference. The solution integrates GPU, DRAM, and enterprise SSD resources to enhance performance and reduce reliance on expensive GPU hardware.

Key Features of the AI Scaler Toolkit

The toolkit addresses common bottlenecks in LLM inference, particularly regarding the KV (Key-Value) cache, by utilizing a tiered approach:

  • vLLM KV Cache Offloading: Instead of recomputing data when GPU memory is exceeded, the toolkit offloads the KV cache to DRAM and TRUSTA PCIe Gen5 enterprise SSDs. This allows for faster responses on long documents and ensures context persistence after server reboots.
  • Strata Engine Integration: The toolkit incorporates the Strata inference engine, which utilizes Intel® AMX acceleration. Internal testing by TRUSTA indicated that this configuration can speed up prompt processing by up to 25% for multi-turn conversations.
  • Simplified Management: It offers a unified interface for allocating hardware resources, tuning parameters, and testing various models, including support for vLLM, llama.cpp, and Strata.

Infrastructure Hardware

Alongside the software toolkit, TRUSTA is showcasing new hardware aimed at AI and HPC environments:

  • PCIe Gen5 Enterprise SSDs: Available in capacities up to 15.36TB, these drives support Flexible Data Placement (FDP) and are designed for energy efficiency.
  • DDR5 R-DIMMs: Modules supporting speeds up to 6,400MT/s and capacities up to 128GB, featuring built-in ECC to maintain stability under heavy workloads.

Verification

Interested parties can review the technical performance claims through the company's official communications. Performance improvements cited in the toolkit are based on TRUSTA’s internal November 2026 testing involving specific Intel® Xeon® processors and NVIDIA RTX 6000 Ada GPUs. Actual deployment results will vary based on hardware, models, and input length.

Share this article
Read the original source: TRUSTA Debuts at SuperComputing 2026 with Upgraded AI Scaler Toolkit

AI-assisted reporting; source claims are attributed and have not been independently verified.

About the author

GlobeWire Daily Editorial Desk — AI-assisted news based on attributed published sources.

More to read