{"id":129314,"date":"2025-11-13T12:24:28","date_gmt":"2025-11-13T04:24:28","guid":{"rendered":"https:\/\/www.ioiotimes.com\/?p=129314"},"modified":"2025-11-13T22:50:42","modified_gmt":"2025-11-13T14:50:42","slug":"nvidia%e5%9c%a8mlperf-training-v5-1%e5%9f%ba%e6%ba%96%e6%b8%ac%e8%a9%a6%e4%b8%ad%e5%85%a8%e9%9d%a2%e5%a5%aa%e5%86%a0","status":"publish","type":"post","link":"https:\/\/www.ioiotimes.com\/?p=129314","title":{"rendered":"NVIDIA\u5728MLPerf Training v5.1\u57fa\u6e96\u6e2c\u8a66\u4e2d\u5168\u9762\u596a\u51a0"},"content":{"rendered":"\n<h3 class=\"wp-block-heading\">NVIDIA Blackwell Ultra\u642d\u914dNVFP4\u6280\u8853\uff0c\u70ba\u5927\u578b\u8a9e\u8a00\u6a21\u578b\u8a13\u7df4\u5e36\u4f86\u7a81\u7834\u6027\u8e8d\u9032<\/h3>\n\n\n\n<p>\u5728<a href=\"https:\/\/www.nvidia.com\/en-us\/glossary\/ai-reasoning\/\" target=\"_blank\" rel=\"noreferrer noopener\">\u4eba\u5de5\u667a\u6167\uff08AI\uff09\u63a8\u7406<\/a>\u7684\u6642\u4ee3\uff0c\u8a13\u7df4\u66f4\u8070\u660e\u3001\u66f4\u5f37\u5927\u7684\u6a21\u578b\u5c0d\u65bc\u64f4\u5c55\u667a\u6167\u800c\u8a00\u81f3\u95dc\u91cd\u8981\u3002\u70ba\u4e86\u6eff\u8db3\u9019\u500b\u65b0\u6642\u4ee3\u7684\u5f37\u5927\u6548\u80fd\u9700\u6c42\uff0c\u9700\u8981\u5728 GPU\u3001CPU\u3001\u7db2\u8def\u4ecb\u9762\u5361\uff08NIC\uff09\u3001\u7db2\u8def\u7684\u5782\u76f4\u64f4\u5c55\uff08scale-up\uff09\u548c\u6c34\u5e73\u64f4\u5c55\uff08scale-out\uff09\u3001\u7cfb\u7d71\u67b6\u69cb\uff0c\u4ee5\u53ca\u5927\u91cf\u7684\u8edf\u9ad4\u548c\u6f14\u7b97\u6cd5\u7b49\u65b9\u9762\u53d6\u5f97\u7a81\u7834\u6027\u9032\u5c55\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1280\" height=\"680\" src=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia01.jpg\" alt=\"\" class=\"wp-image-129317\" title=\"\" srcset=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia01.jpg 1280w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia01-300x159.jpg 300w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia01-1024x544.jpg 1024w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia01-768x408.jpg 768w\" sizes=\"(max-width: 1280px) 100vw, 1280px\" \/><\/figure>\n\n\n\n<p><\/p>\n\n\n\n<p>NVIDIA \u5728 MLPerf Training v5.1 \u4e2d\u6a6b\u6383\u5168\u90e8\u4e03\u9805\u6e2c\u8a66\uff0c\u65bc<a href=\"https:\/\/www.nvidia.com\/en-us\/glossary\/large-language-models\/\" target=\"_blank\" rel=\"noreferrer noopener\">\u5927\u578b\u8a9e\u8a00\u6a21\u578b\uff08LLM\uff09<\/a>\u3001\u5716\u50cf\u751f\u6210\u3001<a href=\"https:\/\/blogs.nvidia.com\/blog\/whats-a-recommender-system\/\" target=\"_blank\" rel=\"noreferrer noopener\">\u63a8\u85a6\u7cfb\u7d71<\/a>\u3001\u96fb\u8166\u8996\u89ba\u548c<a href=\"https:\/\/blogs.nvidia.com\/blog\/what-are-graph-neural-networks\/\" target=\"_blank\" rel=\"noreferrer noopener\">\u5716\u50cf\u795e\u7d93\u7db2\u8def<\/a>\u7b49\u9818\u57df\u5747\u5275\u4e0b\u6700\u5feb\u7684\u8a13\u7df4\u901f\u5ea6\u3002MLPerf Training v5.1 \u70ba\u6301\u7e8c\u591a\u5e74\u8209\u8fa6\u7684 AI \u8a13\u7df4\u6548\u80fd\u696d\u754c\u57fa\u6e96\u6e2c\u8a66\u4e2d\u6700\u65b0\u4e00\u8f2a\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1680\" height=\"945\" src=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia02.jpg\" alt=\"\" class=\"wp-image-129319\" title=\"\" srcset=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia02.jpg 1680w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia02-300x169.jpg 300w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia02-1024x576.jpg 1024w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia02-768x432.jpg 768w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia02-1536x864.jpg 1536w\" sizes=\"(max-width: 1680px) 100vw, 1680px\" \/><\/figure>\n\n\n\n<p><\/p>\n\n\n\n<p>NVIDIA\u4ea6\u662f\u552f\u4e00\u5728\u6240\u6709\u6e2c\u8a66\u9805\u76ee\u4e2d\u90fd\u63d0\u4ea4\u7d50\u679c\u7684\u5e73\u53f0\uff0c\u5f70\u986fNVIDIA GPU\u5f37\u5927\u7684\u53ef\u7a0b\u5f0f\u5316\u7279\u6027\uff0c\u4ee5\u53ca\u5176<a href=\"https:\/\/www.nvidia.com\/zh-tw\/technologies\/cuda-x\/\" target=\"_blank\" rel=\"noreferrer noopener\">CUDA<\/a>\u8edf\u9ad4\u5806\u758a\u7684\u6210\u719f\u5ea6\u548c\u901a\u7528\u6027\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>NVIDIA Blackwell Ultra<\/strong><strong>\u518d\u5beb\u4f73\u7e3e<\/strong><strong><\/strong><\/h3>\n\n\n\n<p>\u57fa\u65bc NVIDIA Blackwell Ultra GPU \u67b6\u69cb\u7684 GB300 NVL72 \u6a5f\u67b6\u7d1a\u7cfb\u7d71\u5728\u672c\u8f2a MLPerf Training \u6e2c\u8a66\u4e2d\u9996\u6b21\u4eae\u76f8\uff0c\u5728\u6b64\u4e4b\u524d\u8a72\u7cfb\u7d71\u5df2\u5728<a href=\"https:\/\/blogs.nvidia.com.tw\/blog\/mlperf-inference-blackwell-ultra\/\" target=\"_blank\" rel=\"noreferrer noopener\">\u6700\u65b0\u4e00\u8f2aMLPerf Inference\u6e2c\u8a66<\/a>\u4e2d\u5275\u4e0b\u7d00\u9304\u3002<\/p>\n\n\n\n<p>\u8207\u4e0a\u4e00\u4ee3 Hopper \u67b6\u69cb\u76f8\u6bd4\uff0c\u57fa\u65bc Blackwell Ultra \u7684 GB300 NVL72 \u5728\u76f8\u540c\u6578\u91cf\u7684 GPU \u4e0b\uff0cLlama 3.1 405B \u9810\u8a13\u7df4\u7684\u6548\u80fd\u63d0\u5347\u4e86 4 \u500d\u4ee5\u4e0a\uff0cLlama 2 70B LoRa \u5fae\u8abf\u6548\u80fd\u589e\u52a0\u4e86\u8fd1 5 \u500d\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1680\" height=\"945\" src=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia03.jpg\" alt=\"\" class=\"wp-image-129320\" title=\"\" srcset=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia03.jpg 1680w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia03-300x169.jpg 300w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia03-1024x576.jpg 1024w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia03-768x432.jpg 768w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia03-1536x864.jpg 1536w\" sizes=\"(max-width: 1680px) 100vw, 1680px\" \/><\/figure>\n\n\n\n<p><\/p>\n\n\n\n<p>\u9019\u4e9b\u6548\u80fd\u63d0\u5347\u5f97\u76ca\u65bc Blackwell Ultra \u7684\u67b6\u69cb\u8e8d\u9032\uff0c\u5305\u62ec\u5177\u5099 15 petaflops NVFP4 AI \u904b\u7b97\u80fd\u529b\u7684\u5168\u65b0 Tensor Core\u30012 \u500d\u7684\u6ce8\u610f\u529b\u5c64\u904b\u7b97\u80fd\u529b\uff0c\u4ee5\u53ca 279GB \u7684 HBM3e \u8a18\u61b6\u9ad4\uff0c\u540c\u6642\u7d50\u5408\u5168\u65b0\u7684\u8a13\u7df4\u65b9\u6cd5\uff0c\u5145\u5206\u91cb\u653e\u8a72\u67b6\u69cb\u9f90\u5927\u7684 NVFP4 \u904b\u7b97\u6548\u80fd\u3002<\/p>\n\n\n\n<p>\u7528\u65bc\u9023\u63a5\u591a\u7d44 GB300 NVL72 \u7cfb\u7d71\uff0c\u4e14\u70ba\u696d\u754c\u9996\u6b3e\u7aef\u5230\u7aef&nbsp;800 Gb\/s&nbsp;\u5782\u76f4\u64f4\u5c55\u7db2\u8def\u5e73\u53f0\u7684 <a href=\"https:\/\/www.nvidia.com\/zh-tw\/networking\/products\/infiniband\/quantum-x800\/\" target=\"_blank\" rel=\"noreferrer noopener\">NVIDIA Quantum-X800 InfiniBand<\/a> \u5e73\u53f0\uff0c\u4e5f\u5728 MLPerf \u6e2c\u8a66\u9996\u5ea6\u4eae\u76f8\uff0c\u5176\u6a6b\u5411\u64f4\u5c55\u7db2\u8def\u983b\u5bec\u8f03\u524d\u4e00\u4ee3\u63d0\u5347 1 \u500d\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>\u91cb\u653e\u6548\u80fd\uff1a<\/strong><strong>NVFP4<\/strong><strong>\u52a0\u901f\u5927\u578b\u8a9e\u8a00\u6a21\u578b\u8a13\u7df4<\/strong><strong><\/strong><\/h3>\n\n\n\n<p>NVIDIA \u65bc\u672c\u8f2a\u53d6\u5f97\u5353\u8d8a\u6210\u679c\u7684\u95dc\u9375\uff0c\u5728\u65bc\u63a1\u7528 NVFP4 \u7cbe\u5ea6\u9032\u884c\u904b\u7b97\uff0c\u9019\u4e5f\u662f MLPerf Training \u6b77\u53f2\u4e0a\u7684\u9996\u6b21\u5275\u8209\u3002<\/p>\n\n\n\n<p>\u63d0\u5347\u904b\u7b97\u6548\u80fd\u7684\u5176\u4e2d\u4e00\u500b\u65b9\u6cd5\uff0c\u662f\u6253\u9020\u80fd\u4ee5\u8f03\u5c11\u4f4d\u5143\u8868\u793a\u8cc7\u6599\u4e26\u9032\u884c\u904b\u7b97\u7684\u67b6\u69cb\uff0c\u4e26\u4ee5\u66f4\u5feb\u7684\u901f\u5ea6\u57f7\u884c\u9019\u4e9b\u8a08\u7b97\u3002\u7136\u800c\uff0c\u7cbe\u5ea6\u964d\u4f4e\u610f\u5473\u8457\u6bcf\u6b21\u8a08\u7b97\u53ef\u7528\u7684\u8cc7\u8a0a\u91cf\u6e1b\u5c11\uff0c\u56e0\u6b64\u5728\u8a13\u7df4\u904e\u7a0b\u4e2d\u4f7f\u7528\u4f4e\u7cbe\u5ea6\u8a08\u7b97\u9700\u8981\u8b39\u614e\u8a2d\u8a08\u6c7a\u7b56\uff0c\u4ee5\u78ba\u4fdd\u7d50\u679c\u7684\u6e96\u78ba\u6027\u3002<\/p>\n\n\n\n<p>NVIDIA&nbsp;\u5718\u968a\u5728\u6bcf\u4e00\u5c64\u6280\u8853\u5806\u758a\u90fd\u9032\u884c\u4e86\u5275\u65b0\uff0c\u4ee5\u5c07 FP4 \u7cbe\u5ea6\u61c9\u7528\u65bc\u5927\u578b\u8a9e\u8a00\u6a21\u578b\u8a13\u7df4\u3002NVIDIA Blackwell GPU \u53ef\u57f7\u884c FP4 \u8a08\u7b97\uff0c\u5305\u62ec NVIDIA \u8a2d\u8a08\u7684 NVFP4 \u683c\u5f0f\u53ca\u5176\u4ed6 FP4 \u8b8a\u9ad4\uff0c\u8a08\u7b97\u901f\u5ea6\u662f FP8 \u7684\u5169\u500d\u3002Blackwell Ultra \u66f4\u5c07\u6b64\u6548\u80fd\u63d0\u5347\u81f3 3 \u500d\uff0c\u4f7f GPU \u5f97\u4ee5\u63d0\u4f9b\u66f4\u5f37\u5927\u7684 AI \u904b\u7b97\u6548\u80fd\u3002<\/p>\n\n\n\n<p>NVIDIA&nbsp;\u662f\u8fc4\u4eca\u70ba\u6b62\u552f\u4e00\u4e00\u500b\u5728\u6eff\u8db3\u57fa\u6e96\u6e2c\u8a66\u56b4\u683c\u7cbe\u5ea6\u8981\u6c42\u7684\u540c\u6642\uff0c\u4f7f\u7528&nbsp;FP4&nbsp;\u7cbe\u5ea6\u9032\u884c\u8a08\u7b97\u4e26\u63d0\u4ea4&nbsp;MLPerf&nbsp;\u8a13\u7df4\u7d50\u679c\u7684\u5e73\u53f0\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>NVIDIA Blackwell<\/strong><strong>\u7a81\u7834\u64f4\u5c55\u6548\u80fd<\/strong><strong><\/strong><\/h3>\n\n\n\n<p>NVIDIA \u6191\u85c9\u903e 5,000 \u9846 Blackwell GPU \u7684\u9ad8\u6548\u5354\u540c\u904b\u4f5c\uff0c\u5728 Llama 3.1 405B \u6e2c\u8a66\u4e2d\u5275\u4e0b\u4e86\u50c5\u9700 10 \u5206\u9418\u8a13\u7df4\u6642\u9593\u7684\u5168\u65b0\u7d00\u9304\u3002\u9019\u9805\u6210\u7e3e\u8f03\u4e0a\u4e00\u8f2a\u57fa\u65bc Blackwell \u7684\u6700\u4f73\u6210\u7e3e\u9ad8\u51fa 2.7 \u500d\uff0c\u6b78\u529f\u65bc GPU \u6578\u91cf\u589e\u52a0\u81f3\u539f\u5148 2 \u500d\u4ee5\u4e0a\u7684\u9ad8\u6548\u64f4\u5c55\uff0c\u4ee5\u53ca\u63a1\u7528 NVFP4 \u7cbe\u5ea6\u5927\u5e45\u63d0\u5347\u6bcf\u9846 Blackwell GPU \u7684\u6709\u6548\u6548\u80fd\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1680\" height=\"945\" src=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia04.jpg\" alt=\"\" class=\"wp-image-129322\" title=\"\" srcset=\"https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia04.jpg 1680w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia04-300x169.jpg 300w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia04-1024x576.jpg 1024w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia04-768x432.jpg 768w, https:\/\/www.ioiotimes.com\/wordpress\/wp-content\/uploads\/2025\/11\/20251113-nvidia04-1536x864.jpg 1536w\" sizes=\"(max-width: 1680px) 100vw, 1680px\" \/><\/figure>\n\n\n\n<p><\/p>\n\n\n\n<p>\u70ba\u8aaa\u660e\u6bcf\u9846 GPU \u7684\u6548\u80fd\u63d0\u5347\u5e45\u5ea6\uff0cNVIDIA \u672c\u8f2a\u63d0\u4ea4\u63a1\u7528 2,560 \u9846 Blackwell GPU \u7684\u6e2c\u8a66\u7d50\u679c\uff0c\u8a13\u7df4\u8017\u6642\u7e2e\u77ed\u81f3 18.79 \u5206\u9418\uff0c\u8f03\u4e0a\u4e00\u8f2a\u63a1\u7528 2,496 \u9846 GPU \u7684\u63d0\u4ea4\u65b9\u6848\u5feb\u4e0a 45%\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>\u5168\u65b0\u57fa\u6e96\u6e2c\u8a66\uff0c\u5168\u65b0\u6548\u80fd\u7d00\u9304<\/strong><strong><\/strong><\/h3>\n\n\n\n<p>NVIDIA \u5728\u672c\u8f2a\u65b0\u589e\u7684\u5169\u9805\u57fa\u6e96\u6e2c\u8a66\u4e2d\u540c\u6a23\u5275\u4e0b\u6548\u80fd\u7d00\u9304\uff0c\u5305\u62ec Llama 3.1 8B \u8207 FLUX.1\u3002<\/p>\n\n\n\n<p>Llama 3.1 8B&nbsp;\u662f\u4e00\u6b3e\u7cbe\u5de7\u537b\u529f\u80fd\u5f37\u5927\u7684\u5927\u578b\u8a9e\u8a00\u6a21\u578b\uff0c\u53d6\u4ee3\u4e86\u9577\u671f\u904b\u884c\u7684 BERT-large \u6a21\u578b\uff0c\u70ba\u57fa\u6e96\u6e2c\u8a66\u5957\u4ef6\u589e\u6dfb\u66f4\u73fe\u4ee3\u5316\u3001\u66f4\u5c0f\u7684\u5927\u578b\u8a9e\u8a00\u6a21\u578b\u3002NVIDIA \u4f7f\u7528\u591a\u9054 512 \u9846 Blackwell Ultra GPU \u63d0\u4ea4\u4e86\u6e2c\u8a66\u7d50\u679c\uff0c\u5275\u4e0b\u50c5\u9700&nbsp;5.2 \u5206\u9418\u5373\u53ef\u5b8c\u6210\u8a13\u7df4\u7684\u7d00\u9304\u3002<\/p>\n\n\n\n<p>\u6b64\u5916\uff0cFLUX.1 \u662f\u4e00\u6b3e\u5148\u9032\u7684\u5716\u50cf\u751f\u6210\u6a21\u578b\uff0c\u53d6\u4ee3\u4e86 Stable Diffusion v2\uff0c\u4e14\u53ea\u6709 NVIDIA \u5e73\u53f0\u63d0\u4ea4\u4e86\u8a72\u57fa\u6e96\u6e2c\u8a66\u7684\u7d50\u679c\u3002NVIDIA \u4f7f\u7528 1,152 \u500b Blackwell GPU \u63d0\u4ea4\u7684\u6e2c\u8a66\u7d50\u679c\uff0c\u5275\u4e0b\u5c07\u8a13\u7df4\u6642\u9593\u7e2e\u77ed\u81f3 12.5 \u5206\u9418\u7684\u65b0\u7d00\u9304\u3002<\/p>\n\n\n\n<p>NVIDIA&nbsp;\u6301\u7e8c\u4fdd\u6301\u73fe\u6709\u5716\u50cf\u795e\u7d93\u7db2\u8def\u3001\u76ee\u6a19\u5075\u6e2c\u548c\u63a8\u85a6\u7cfb\u7d71\u6e2c\u8a66\u7684\u7d00\u9304\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>\u5ee3\u6cdb\u4e14\u6df1\u5ea6\u7684\u5925\u4f34\u751f\u614b\u7cfb<\/strong><strong><\/strong><\/h3>\n\n\n\n<p>NVIDIA \u751f\u614b\u7cfb\u5728\u672c\u8f2a\u6e2c\u8a66\u4e2d\u5c55\u73fe\u7a4d\u6975\u53c3\u8207\u5ea6\uff0c\u5171\u8a08 15 \u500b\u6a5f\u69cb\u63d0\u4ea4\u4e86\u4ee4\u4eba\u77da\u76ee\u7684\u6210\u679c\uff0c\u5176\u4e2d\u5305\u62ec\u83ef\u78a9\u3001\u6234\u723e\u79d1\u6280\u96c6\u5718\u3001\u6280\u92fc\u79d1\u6280\u3001HPE\u3001Krai\u3001Lambda\u3001\u806f\u60f3\u96c6\u5718\u3001Nebius\u3001\u96f2\u9054\u79d1\u6280\u3001Supermicro\u3001\u4f5b\u7f85\u91cc\u9054\u5927\u5b78\u3001Verda\uff08\u539fDataCrunch\uff09\u548c\u7def\u7a4e\u3002<\/p>\n\n\n\n<p>NVIDIA \u4ee5\u4e00\u5e74\u70ba\u9031\u671f\u6301\u7e8c\u5275\u65b0\uff0c\u5728\u9810\u8a13\u7df4\u3001\u5f8c\u8a13\u7df4\u53ca\u63a8\u8ad6\u968e\u6bb5\u63a8\u52d5\u986f\u8457\u4e14\u5feb\u901f\u7684\u6548\u80fd\u63d0\u5347\uff0c\u70ba\u9081\u5411\u5168\u65b0\u5c64\u7d1a\u7684\u667a\u6167\u92ea\u8def\uff0c\u52a0\u901f AI \u666e\u53ca\u5316\u3002<\/p>\n\n\n\n<p><\/p>\n\n\n\n<p><sub>\u5982\u6b32\u4e86\u89e3\u66f4\u591aNVIDIA\u7684\u6548\u80fd\u8cc7\u6599\uff0c\u8acb\u53c3\u95b1 <a href=\"https:\/\/developer.nvidia.com\/deep-learning-performance-training-inference?sortBy=developer_learning_library%2Fsort%2Ffeatured_in.deep_learning_performance%3Adesc%2Ctitle%3Aasc\" target=\"_blank\" rel=\"noreferrer noopener\">NVIDIA\u8cc7\u6599\u4e2d\u5fc3\u6df1\u5ea6\u5b78\u7fd2\u7522\u54c1\u6548\u80fd\u4e2d\u5fc3<\/a>\u548c <a href=\"https:\/\/aibenchmarking.ngc.nvidia.com\/\" target=\"_blank\" rel=\"noreferrer noopener\">Performance Explorer<\/a> \u9801\u9762\u3002<\/sub><\/p>\n\n\n\n<p><\/p>\n\n\n\n<p>\u3000\u3000\u3000\u3000<\/p>\n\n\n\n<h4 class=\"wp-block-heading has-text-align-right has-very-light-gray-to-cyan-bluish-gray-gradient-background has-background\">\ud83d\udfe6\u4ec0\u9ebc!?\u804a\u5929\u6253\u5c41\u4e5f\u53ef\u4ee5\u62bd\u734e\u52f5!! <a href=\"https:\/\/reurl.cc\/5blVWq\" data-type=\"link\" data-id=\"https:\/\/reurl.cc\/5blVWq\" target=\"_blank\" rel=\"noreferrer noopener\">11\u6708\u641e\u754f\u6709\u79ae\u6d3b\u52d5<\/a>\u73fe\u6b63\u9032\u884c\u4e2d!! <br>\ud83d\udfe6\u73fe\u5728\u5c31\u52a0\u5165\u00a0ioioTIMES \u81c9\u66f8\u7c89\u7d72\u5718\u00a0\u66f4\u591a\u4e92\u52d5\u3001\u66f4\u591a\u597d\u5eb7\u650f\u62b5\u52a0!!<br>\ud83d\udfe6<strong>\u6211\u5011\u6709<a href=\"https:\/\/today.line.me\/tw\/v2\/publisher\/103117\" target=\"_blank\" rel=\"noreferrer noopener nofollow\">LINE TODAY<\/a>\u983b\u9053\u4e86\uff0c\u5feb\u4f86\u8ffd\u8e2a\u6211\u5011\u5427!!&#8211;\u6700\u65b0\u79d1\u6280\u65b0\u805e \u76e1\u5728\u4f60<\/strong>\u624b<\/h4>\n","protected":false},"excerpt":{"rendered":"<p>NVIDIA Blackwell U<\/p>\n","protected":false},"author":3,"featured_media":129320,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"rank_math_lock_modified_date":false,"footnotes":""},"categories":[13],"tags":[13024,580,15994,15995,71],"class_list":["post-129314","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news","tag-blackwell-ultra","tag-focus","tag-mlperf-training-v5-1","tag-nvfp4","tag-nvidia"],"_links":{"self":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts\/129314"}],"collection":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=129314"}],"version-history":[{"count":6,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts\/129314\/revisions"}],"predecessor-version":[{"id":129399,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/posts\/129314\/revisions\/129399"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=\/wp\/v2\/media\/129320"}],"wp:attachment":[{"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=129314"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=129314"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.ioiotimes.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=129314"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}