[KongchangAI]
Unverified50% confidenceFactExact time

两张RTX 3090通过NVLink互联时合计48GB显存可以全精度FP16运行27B模型，若没有NVLink则只能依赖PCIe通信遭遇带宽瓶颈

1
Sources
50%
Confidence
Long-term
Relevance
7/11/2026
First Seen

Sources

Why Is Qwen3 27B So Slow on a 4090? A Hands-On Guide to Local Inference Performance Optimization
redditr/ollama7/10/2026

Related Claims

UnverifiedGB300 搭载 NVLink 第五代互联技术，单机8卡的卡间带宽达到1.8TB/s73% similarUnverifiedThe RTX 4090 has 24GB of VRAM and can smoothly run most 13B–34B quantized models72% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/486090
API
curl https://kongchang.com/api/v1/knowledge/claims/486090
MCP
get_claim(id=486090)