Running summarizing tools on a new article

Abstract: The increase of large language models (LLMs) such as Qwen, LLaMA, Mistral, and GPT variants has intensified the need for systematic benchmarking across heterogeneous hardware platforms. This paper evaluates NVIDIA A100 and RTX 3090 GPUs, AMD RX 580 and RX 7600 XT GPUs, and an Intel i5-14400F CPU on a diverse suite of open-source LLMs, including Mistral-7B, LLaMA-2 (7B, 13B), LLaMA-3.x (3B–11B), CodeLLaMA, Qwen1.5/2/2.5 (1.8B–14B), and legacy GPT models (GPT-2, GPT-Neo, GPT-J, GPT-NeoX-20B). Performance was assessed in terms of sustained TFLOPs, throughput in tokens per second (TPS), and memory/power efficiency. Results show that smaller Qwen models peaked at 276 TPS (A100), the RTX 3090 dominated 1–7B models due to higher frequency and lower overhead, and the A100 scaled efficiently on 13B–20B workloads. The RX 7600 XT, enabled by patched ROCm 6.4.2 gfx1102 support, sustained up to 35–50 TPS on 1.5B–3B models and executed models up to 14B parameters, while the RX 580 and CPU were limited to lightweight inference. These findings underscore how real-world performance depends not only on theoretical FLOPs but also on memory capacity, kernel launch overheads, and software stack maturity, offering actionable insights for selecting hardware across consumer and data-center deployments.