本文へ移動
AiSeekAiSeek
サインイン
安定テクノロジー報道日時 2026-09-16 23:00

NVIDIAのNVL72はMLPerf Inference v6.1において最優れた性能を実現

NVIDIAのNVL72システム性能、効率的なインフラスケーリング、および継続的なソフトウェア最適化が、AIインフェレンス経済学において重要であり、プラットフォームの汎用性によりさまざまなワークロードにおける高利用率を実現しています。

01

誰に影響するか

  1. 1NVIDIA NVL72
  2. を可能にする →事実
    2scale-up network beyond standard 8-way servers技術
02

根拠

  • NNVIDIA Blog企業開示2026-09-16 23:00
    Software optimizations in NVIDIA’s MLPerf Inference v6.1 submissions delivered up to 1.6x higher performance over v6.0.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA submitted Vera Rubin NVL72 preview results on two of the most demanding benchmarks in the MLPerf Inference v6.1 suite: DeepSeek-R1 and Qwen3-VL.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    A 288-GPU submission across four GB300 NVL72 racks achieved 99% scaling efficiency
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    GB300 NVL72 also demonstrated rack-scale efficiency on the WAN 2.2 text-to-video benchmark, reaching 0.65 720p videos per second at 5.7 seconds per video — 9x higher throughput and 7.5x lower latency than a single node.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA’s DeepSeek-R1 (DSR1) submission scaled from a single GB300 NVL72 rack (72 GPUs) to four racks (288 GPUs), achieving 99% scaling efficiency in the offline scenario.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA Vera Rubin NVL72 delivers up to 3.7x better throughput than GB300 NVL72.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA submitted Vera Rubin NVL72 preview results on two of the most demanding benchmarks in the MLPerf Inference v6.1 suite
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    scaled from a single GB300 NVL72 rack (72 GPUs) to four racks (288 GPUs)
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    Beyond the NVIDIA Grace Blackwell and Vera Rubin NVL72 platform results, NVIDIA submitted Jetson AGX Thor results using NVIDIA TensorRT Edge-LLM on the newly introduced Edge-Agentic benchmark with Qwen3.6-27B.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    Vera Rubin NVL72 delivers up to 3.7x higher throughput than GB300 NVL72 on Qwen3-VL across offline, server and interactive scenarios
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    On DeepSeek-R1, using the NVIDIA TensorRT-LLM library, throughput is up to 2.5x higher than GB300 NVL72.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA platform results from the following entries: 6.1-0073 and 6.1-0074.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    The NVL72 scale-up domain — powered by sixth-generation NVIDIA NVLink and NVLink Switch to deliver 10x higher packet rates and 3x lower latency than off-the-shelf Ethernet
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    MLPerf Inference v6.1, Closed Division.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA Vera Rubin NVL72 system debuts with leading performance
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    Vera Rubin NVL72 delivered 30x better performance than GB300 NVL72 in preview testing.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    each Vera Rubin NVL72 rack delivers significantly more tokens, serves more users and generates more revenue than a GB300 NVL72 rack, while lowering cost per token.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    NVIDIA delivers this with high-bandwidth, low-latency scale-up interconnects within each rack, high-bandwidth networking between racks and efficient request orchestration across nodes.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    Nebius also submitted Vera Rubin NVL72 preview results and demonstrated excellent performance.
    出典を見る
  • NNVIDIA Blog企業開示2026-09-16 23:00
    In v6.1, GB300 NVL72 performance on Qwen3-VL improved up to 1.6x over v6.0 results.
    出典を見る