MiniMax M2.7 AWQ-4bit on 2x Spark vs 2x RTX 6000 96GB - performance and energy efficiency

Hello, This model/quant is my daily driver and I wanted to have some reference benchs for comparing my setup with a 3x more expensive and 4x time power hungry setup. Results first, methodology after, link at the end with all results Model: cyankiwi/MiniMax-M2.7-AWQ-4bit Results (c1) (tried to upload the table as text, didn't work as expected) So to my surprise, the Spark cluster isn't that far behind. On average the 2x RTX 6000 is 2.7x faster on prompt processing and 4.88x faster on token genera

More from LocalLlama

Bruh

LocalLlama·3

Qwen3.6-27B-NVFP4 - images

LocalLlama·5

Hybrid on-device inference on Android: llama.cpp + LiteRT + NPU/GPU routing

LocalLlama·6

Qwen3.6-27B at 72 tok/s on RTX 3090 on Windows using native vLLM (no WSL, no Docker), portable launcher and installer

LocalLlama·6

A Dark-Money Campaign Is Paying Influencers to Frame Chinese AI as a Threat

LocalLlama·6

"LLM is created so engineer don't have to write a report", anyway found out ONLYOFFICE can connect to OpenAI compatible, using Qwen 3.6 to do elaboration.

LocalLlama·5