Skip to content

README.md

Latest commit

 

History

History
213 lines (162 loc) · 7.16 KB

README.md

File metadata and controls

213 lines (162 loc) · 7.16 KB

Multimodal Inference in Dynamo

Dynamo supports multimodal inference across multiple LLM backends, enabling models to process images, video, and audio alongside text. This section provides comprehensive documentation for deploying multimodal models.

Important

Security Requirement: Multimodal processing must be explicitly enabled at startup. See the relevant documentation for each backend for the necessary flags.

This prevents unintended processing of multimodal data from untrusted sources.

Backend Documentation

:maxdepth: 1

vLLM Multimodal <multimodal_vllm.md>
TensorRT-LLM Multimodal <multimodal_trtllm.md>
SGLang Multimodal <multimodal_sglang.md>

Support Matrix

Backend Capabilities

Stack E/PD E/P/D EP/D EPD Image Video Audio
vLLM ✅ ✅ ✅ ✅ ✅ ✅ 🧪
TRT-LLM ❌ 🚧* ✅ ✅ ✅ ❌ ❌
SGLang ✅ ✅ ❌ ❌ ✅ ❌ ❌

* E/P/D supported in TRT-LLM with pre-computed embeddings only; image URL support is WIP (PR #4668)

Pattern Key:

  • EPD - All-in-one worker (Simple Aggregated)
  • E/PD - Separate encode, combined prefill+decode
  • E/P/D - All stages separate
  • EP/D - Combined encode+prefill, separate decode

Status: ✅ Supported | 🚧 WIP | 🧪 Experimental | ❌ Not supported

Input Format Support

Format vLLM TRT-LLM SGLang
HTTP/HTTPS URL ✅ ✅ ✅
Data URL (Base64) ✅ ❌ ❌
Pre-computed Embeddings (.pt) ❌ ✅ ❌

Architecture Patterns

Dynamo supports several deployment patterns for multimodal inference based on two dimensions:

  1. Encoding: Is media encoding handled inline (within prefill) or by a separate Encode Worker?

    • Inline: Simpler setup, encoding happens in the prefill worker
    • Separate (EPD): Dedicated encode worker transfers embeddings via NIXL (RDMA), enabling independent scaling

  2. Prefill/Decode: Are prefill and decode in the same worker or separate?

    • Aggregated: Single worker handles both prefill and decode
    • Disaggregated: Separate workers for prefill and decode, with KV cache transfer between them

These combine into four deployment patterns:

EPD - Simple Aggregated

All processing happens within a single worker - the simplest setup.

HTTP Frontend (Rust)
    ↓
Worker (Python)
    ↓ image load + encode + prefill + decode
Response
Component Purpose
Frontend (Rust) HTTP entry point, tokenization, image URL preprocessing
Worker Complete inference pipeline (encode + prefill + decode)

When to use: Quick setup, smaller models, development/testing.

E/PD - Encode Separate

Encoding happens in a separate worker; prefill and decode share the same engine.

HTTP Frontend (Rust)
    ↓
Processor (Python)
    ↓ tokenizes, extracts media URL
Encode Worker (Python)
    ↓ downloads media, generates embeddings, NIXL transfer
PD Worker (Python)
    ↓ receives embeddings via NIXL, prefill + decode
Response
Component Purpose
Frontend (Rust) HTTP entry point
Processor (Python) Tokenization, extracts media URLs
Encode Worker Media encoding, embeddings generation
PD Worker Prefill + Decode with embeddings

When to use: Offload vision encoding to separate GPU, scale encode workers independently.

E/P/D - Full Disaggregation

Full disaggregation with separate workers for encoding, prefill, and decode. There are two variants of this workflow:

  • Prefill-first, used by vLLM
  • Decode-first, used by SGLang

Prefill-first:

HTTP Frontend (Rust)
    ↓
Processor (Python)
    ↓ tokenizes, extracts media URL
Encode Worker (Python)
    ↓ downloads media, generates embeddings, NIXL transfer
Prefill Worker (Python)
    ↓ receives embeddings via NIXL, prefill only, KV cache transfer
Decode Worker (Python)
    ↓ decode only, token generation
Response

OR

Decode-first:

HTTP Frontend (Rust)
    ↓
Processor (Python)
    ↓ tokenizes, extracts media URL
Encode Worker (Python)
    ↓ downloads media, generates embeddings, NIXL transfer
Decode Worker (Python)
    ↓ Bootstraps prefill worker
Prefill Worker (Python)
    ↓ receives embeddings via NIXL, prefill only, KV cache transfer
Decode Worker (Python)
    ↓ decode only, token generation
Response
Component Purpose
Frontend (Rust) HTTP entry point
Processor (Python) Tokenization, extracts media URLs
Encode Worker Media encoding, embeddings generation
Prefill Worker Prefill only, transfers KV cache
Decode Worker Decode only, token generation

When to use: Maximum optimization, multi-node deployment, independent scaling of each phase.

EP/D - Traditional Disaggregated

Encoding is combined with prefill, with decode separate.

HTTP Frontend (Rust)
    ↓
Processor (Python)
    ↓ tokenizes, extracts media URL
Encode+Prefill Worker (Python)
    ↓ downloads media, encodes inline, prefill, KV cache transfer
Decode Worker (Python)
    ↓ decode only, token generation
Response
Component Purpose
Frontend (Rust) HTTP entry point
Processor (Python) Tokenization, extracts media URLs (vLLM only)
Encode+Prefill Worker Combined encoding and prefill
Decode Worker Decode only, token generation

Note: TRT-LLM's EP/D mode skips the Python Processor - the Rust frontend handles tokenization and routes directly to the Prefill worker. For multimodal requests, the Python prefill worker still re-tokenizes/builds inputs; Rust token_ids are ignored.

When to use: Models without pre-computed embedding support (Llama 4), or TRT-LLM disaggregated deployment.

Example Workflows

You can find example workflows and reference implementations for deploying multimodal models in: