๐คฏ This entire 64GB AMD AI workstation cost its builder LESS than a single RTX 5090 and Qwen3.8-27B is hitting 111 tok/s.
This is the kind of rig that makes me want to build it because RTX 5090s are so expensive.
Setup ๐
๐ 2ร Radeon AI PRO R9700 32GB
๐พ 64GB total dedicated VRAM
๐ง Ryzen 7500F
๐พ 64GB DDR5
๐ PCIe 5.0 x8 per GPU
โก vLLM Radiance + TP2 + MTP
Running Qwen3.8-27B Quark AWQ MXFP4 ...
๐ 111.4 tok/s median decode
โก 4,224โ4,410 tok/s prefill
โฑ๏ธ 81ms TTFT
๐ 131K server context
Native FP8 still managed 87.6 tps
๐ฎ RTX 5090 โ 32GB VRAM
๐ 2ร R9700 โ 64GB VRAM
The builder says the entire machine cost ~โฌ4,000, more than โฌ1,000 less than a single RTX 5090 in his market.
So for less money, he ended up with 2ร the VRAM + 111 tok/s on Qwen3.8-27B. ๐ฅ
Reddit Link in ALT