{"slug": "please-add-amd-radeon-ai-pro-r9700-to-my-hardware", "title": "Please add \"AMD Radeon AI PRO R9700\" to \"My Hardware\"", "summary": "A user request asks that the AMD Radeon AI PRO R9700 be added to the \"My Hardware\" database, citing its RDNA4 architecture, 32 GB GDDR6 memory, and 383 FP8 TFLOPS / 383 INT8 TOPS throughput as making it a fit for 30B–70B models at Q4_K_M quantization. The request argues the addition would improve Fit Finder accuracy for AMD users and broaden hardware coverage for ROCm and DirectML setups, and includes a Python script that classifies a GPU's local-AI profile from memory capacity, FP8/INT8 throughput, and architecture family.", "body_md": "The R9700 is a good candidate for “My Hardware”: RDNA4 architecture, 32 GB GDDR6, high FP8/INT8 throughput and a memory profile that fits 30B–70B models in Q4_K_M. Adding it would help AMD users get accurate Fit Finder results and improve hardware coverage for ROCm/DirectML setups.\n\nBelow is a small diagnostic script I use to classify a GPU for local‑AI workloads.\n\nIt doesn’t benchmark anything — it simply evaluates the specs and produces a profile that tells which model sizes fit and how strong the GPU is for FP8/INT8 inference.\n\n``` python\ndef gpu_ai_profile(memory_gb, fp8_tflops, int8_tops, architecture):\n    \"\"\"\n    This function builds an 'AI profile' for a GPU based only on its specs.\n    It answers three practical questions:\n    1. Which model sizes fit in VRAM?\n    2. How strong is the GPU for FP8/INT8 inference?\n    3. What architecture family does it belong to?\n    \"\"\"\n\n    profile = {}\n\n    # VRAM capacity → determines which GGUF sizes fit\n    if memory_gb >= 32:\n        profile[\"models_fit\"] = [\"30B\", \"34B\", \"40B\", \"70B (Q4_K_M, borderline)\"]\n    else:\n        profile[\"models_fit\"] = [\"7B\", \"13B\", \"20B\"]\n\n    # FP8 throughput → good indicator for multimodal and MoE models\n    profile[\"fp8_class\"] = (\n        \"high\" if fp8_tflops >= 300 else\n        \"medium\" if fp8_tflops >= 100 else\n        \"low\"\n    )\n\n    # INT8 throughput → relevant for GGUF quantized inference\n    profile[\"int8_class\"] = (\n        \"high\" if int8_tops >= 300 else\n        \"medium\" if int8_tops >= 100 else\n        \"low\"\n    )\n\n    # Architecture tag\n    profile[\"arch\"] = architecture\n\n    return profile\n\n# Example using the R9700 specs:\nprint(gpu_ai_profile(\n    memory_gb=32,\n    fp8_tflops=383,\n    int8_tops=383,\n    architecture=\"RDNA4\"\n))\n```\n\n", "url": "https://wpnews.pro/news/please-add-amd-radeon-ai-pro-r9700-to-my-hardware", "canonical_source": "https://discuss.huggingface.co/t/please-add-amd-radeon-ai-pro-r9700-to-my-hardware/171780#post_7", "published_at": "2026-09-18 06:52:19+00:00", "updated_at": "2026-09-18 06:54:04.735806+00:00", "lang": "en", "topics": ["ai-tools", "ai-infrastructure", "large-language-models"], "entities": ["AMD", "AMD Radeon AI PRO R9700", "RDNA4", "ROCm", "DirectML", "My Hardware", "Fit Finder"], "alternates": {"html": "https://wpnews.pro/news/please-add-amd-radeon-ai-pro-r9700-to-my-hardware", "markdown": "https://wpnews.pro/news/please-add-amd-radeon-ai-pro-r9700-to-my-hardware.md", "text": "https://wpnews.pro/news/please-add-amd-radeon-ai-pro-r9700-to-my-hardware.txt", "jsonld": "https://wpnews.pro/news/please-add-amd-radeon-ai-pro-r9700-to-my-hardware.jsonld"}}