{
  "docId": "019de517-a9ac-776f-9473-0e375a72fdfd",
  "docSlug": "a7286dbbd5b1c9b3685bbbc04ec9c159",
  "documentTitle": "NVIDIA | Investor Presentation Deck | 65 slides",
  "authorId": "nvidia",
  "authorName": "NVIDIA",
  "documentKindSlug": "conference-presentation",
  "documentKindLabel": "Conference presentation",
  "sourceTypeSlug": "investor_relations",
  "sourceTypeLabel": "Investor relations",
  "presentationDate": "2023-12-01 00:00:00",
  "orientation": "landscape",
  "aspectRatio": 1.7777778,
  "pageNumber": 14,
  "pageCount": 65,
  "prevPage": 13,
  "nextPage": 15,
  "slideType": "propose_solution",
  "function": "propose_solution",
  "density": "overcrowded",
  "nDataPoints": 8,
  "notes": null,
  "elementsJson": null,
  "metadataConfidence": 0.9,
  "imagePath": null,
  "slideHref": "/slides/019de517-a9ac-776f-9473-0e375a72fdfd/14",
  "deckHref": "/decks/019de517-a9ac-776f-9473-0e375a72fdfd",
  "deckJsonHref": "/decks/019de517-a9ac-776f-9473-0e375a72fdfd.json",
  "deckAnchorHref": "/decks/019de517-a9ac-776f-9473-0e375a72fdfd#slide-14",
  "components": [
    {
      "bbox": {
        "h": 0.39,
        "w": 0.21,
        "x": 0.76,
        "y": 0.33
      },
      "kind": "chart",
      "text": "4.6X Higher Llama2 Inference Performance\n5x\n4x\n3x\n2x\n1x\n0x\nA100\nH100 August\nH100 TensorRT-LLM",
      "attrs": null,
      "subkind": "bar-vertical",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "8f6200f9-b6f8-4979-9bbd-24f04b85b7b6",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.39,
        "w": 0.21,
        "x": 0.55,
        "y": 0.33
      },
      "kind": "chart",
      "text": "8X Increase in GPT-J 6B Inference Performance\n8x\n7x\n6x\n5x\n4x\n3x\n2x\n1x\n0x\nA100\nH100 August\nH100 TensorRT-LLM",
      "attrs": null,
      "subkind": "bar-vertical",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "ef8ebcbc-b712-4cf3-893e-2b4f25f3f85d",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.42,
        "w": 0.41,
        "x": 0.07,
        "y": 0.34
      },
      "kind": "list",
      "text": "NVIDIA developed TensorRT-LLM, an open-source software library that enables customers to more than double the inference performance of their GPUs\nTensorRT-LLM on H100 GPUs provides up to an 8X performance speedup compared to prior generation A100 GPUs running GPT-J 6B without the software\n5.3X reduction in TCO and 5.6X reduction in energy costs\nWith TensorRT-LLM for Windows, LLMs and generative AI applications can run up to 4x faster locally on PCs and Workstations powered by NVIDIA GeForce RTX and NVIDIA RTX GPUs\nTensorRT-LLM for data centers now publicly available; TensorRT-LLM for Windows in beta",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "06ee810a-1a49-49af-83a2-edbbe72e0c9e",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.03,
        "w": 0.42,
        "x": 0.55,
        "y": 0.76
      },
      "kind": "source-note",
      "text": "Text summarization, variable input/output length, CNN/DailyMail dataset | A100 FP 16 PyTorch eager mode/H100 FP8 | H100 FP8, TensorRT-LLM, in-flight batching",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "0cb03768-ceed-49e8-a2a5-2007782889b0",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.06,
        "w": 0.35,
        "x": 0.61,
        "y": 0.22
      },
      "kind": "title",
      "text": "TensorRT-LLM Supercharges Hopper Performance\nSoftware optimizations double leading performance",
      "attrs": null,
      "subkind": "action-title",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f59646e5-95db-44cd-9bb7-3eb882f7b7cd",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.13,
        "w": 0.41,
        "x": 0.07,
        "y": 0.08
      },
      "kind": "title",
      "text": "New TensorRT-LLM Software More Than Doubles Inference Performance",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "4c1d4c1c-ed4f-40ce-b24a-6d11e717997b",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "2x2 matrix",
      "slug": "matrix-2x2",
      "agent": null,
      "layer": "slide",
      "matchId": "f5bd04fb-6e2d-4f6a-9055-30836a5d082c",
      "evidence": "chart/bar-vertical: 4.6X Higher Llama2 Inference Performance",
      "confidence": 0.7
    },
    {
      "name": "Waterfall chart",
      "slug": "waterfall-chart",
      "agent": null,
      "layer": "slide",
      "matchId": "19d0dff9-f3a0-4727-8f06-e9fc6e9e8699",
      "evidence": "chart/bar-vertical: 8X Increase in GPT-J 6B Inference Performance",
      "confidence": 0.9
    }
  ],
  "frameworks": [],
  "arcBeats": [],
  "loops": [
    {
      "to": 22,
      "from": 13,
      "name": "Why Now",
      "slug": "15-why-now",
      "bestFor": "Sales pitches, fundraising, requesting immediate budget approval",
      "matchId": "859696e7-5459-4daa-b41d-935f0f2f2014",
      "evidence": "The document highlights NVIDIA's recent announcements and partnerships in the AI space",
      "position": 1,
      "objective": "Explain why NVIDIA is well-positioned to drive growth in the AI industry",
      "structure": "The Context (Trends) -> The Trigger Event -> The Window of Opportunity",
      "confidence": 0.6,
      "description": "Create temporal urgency by proving that the window of opportunity is opening or closing"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}