#a100

The latest AI and developer news about #a100, with the original source and a short summary.

NVIDIA Details How Productivity, Durability, and Fungibility Maximize AI Factory ROI

SemiAnalysis data shows NVIDIA Vera Rubin NVL72 systems deliver over 30x higher throughput per megawatt than GB300 NVL72.

·1·2026.10.01 22:00

vllm-xtu-moe enables 748B MoE inference on 2×A100-40GB via CPU-GPU fused execution

The open-source vllm-xtu-moe project achieves up to ~20× faster prefill than CPU-only paths for medium contexts by streaming expert weights from CPU to GPU.

·1·2026.09.29 14:00

KV cache down 35%

Compressed the KV cache on Qwen 3.6's 1M context, reducing it from 10.7GB to 6.9GB.

·1·2026.04.19 11:00

MIG cuts p95 in half

On A100 40GB, MIG significantly lowered protected-tenant p95 compared to shared vLLM.

·2026.04.18 04:00

Llama-2 Inference Characteristics

Llama-2-70B is cheap for prompt processing, but it's at a disadvantage compared to GPT-3.5 for completion-heavy tasks.

·1·2023.07.20 09:00