Unverified50% confidenceFactExact time
UP主使用FP8精度部署千问3.8 27B在48GB显卡上，跑4个并发request占用约46GB
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/18/2026
First Seen
Valid until: 11/16/2026
Sources
Related Claims
UnverifiedA 35B model requires approximately 70GB of VRAM at FP16 precision, far exceeding the RTX 5090's 32GB capacity, so single-GPU deployment requires quantization.75% similarVerified以LLaMA-3 70B（80层、64头、每头128维）为例，单个token的KV Cache约占2.6MB（FP16精度），32K上下文的KV Cache总量约达83GB72% similarUnverifiedA dual 24GB GPU tensor-parallel configuration increases context support from 134,049 to 262,144 tokens at f16 precision.72% similarUnverifiedFP8 quantization can compress the weights of a 7B model from approximately 14GB (FP16) to approximately 7GB.72% similarUnverifiedThe NVIDIA A100 has a memory bandwidth of 2 TB/s; loading a 280 GB FP32 model takes approximately 140 ms, meaning weight loading alone limits generation to a maximum of approximately 7 tokens per second.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/768404API
curl https://kongchang.com/api/v1/knowledge/claims/768404MCP
get_claim(id=768404)