GamesBeat reprint of the LG and Furiosa RNGD EXAONE numbers

Fields

id

sources/gamesbeat-lg-furiosa-exaone-4x-rngd-2025-07

URL

https://www.gamesbeat.com/lg-ai-research-taps-furiosaai-to-get-2-25x-better-llm-inference-performance-vs-gpus/

Publisher

GamesBeat

Kind

review

Published

2025-07-22

Accessed

2026-10-16

notes

Secondary reprint of the Furiosa and LG announcement. Confirms the same sentence: four RNGD cards, batch size one, EXAONE 3.5 32B, 60 tok/s at 4K context and 50 tok/s at 32K context. Also states 512 TFLOPS FP8, 180 W TDP, an eight-card 4U air-cooled RNGD server, and a vLLM-compatible Furiosa-LLM stack. Not an independent measurement.

Notes

Secondary reprint of the Furiosa and LG announcement. Confirms the same sentence: four RNGD cards, batch size one, EXAONE 3.5 32B, 60 tok/s at 4K context and 50 tok/s at 32K context. Also states 512 TFLOPS FP8, 180 W TDP, an eight-card 4U air-cooled RNGD server, and a vLLM-compatible Furiosa-LLM stack. Not an independent measurement.

Cross-references

This record references no other records.