Table of Contents
Double the inference speed of large models
One More Thing
Home Technology peripherals AI Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

Sep 18, 2023 pm 03:45 PM
ai train

The inference speed of large models has doubled in just one month!

Recently, Nvidia announced the launch of a "chicken blood package" specially designed for H100, aiming to speed up the LLM inference process

Maybe now you don't have to wait for the GH200 to be delivered next year. Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed.

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

The computing power of GPU has been affecting the performance of large models. Both hardware suppliers and users hope to obtain faster computing speed

As the largest supplier of hardware behind large models, NVIDIA has been studying how to accelerate large model hardware.

Through cooperation with a number of AI companies, NVIDIA finally launched the large model inference optimization program TensorRT-LLM (tentatively referred to as TensorRT).

TensorRT can not only double the inference speed of large models, but is also very convenient to use.

Without in-depth knowledge of C and CUDA, you can quickly customize optimization strategies and run large models faster on H100.

NVIDIA scientist Jim Fan forwarded and commented that NVIDIA’s “another advantage” is the supporting software that can maximize the use of GPU performance.

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

NVIDIA injects new vitality into its products through software, just like it implements Lao Huang's saying "the more you buy, the more you save." However, this does not prevent some people from thinking that the price of the product is too high

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

In addition to the price, some netizens also questioned its operating effect:

We always I have seen how many times the performance has improved (as advertised), but when I run Llama 2 myself, I can still only process dozens of tokens per second.

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

For TensorRT, we need further testing to determine whether it is really effective. Let us first take a closer look at TensorRT

Double the inference speed of large models

TensorRT-LLM optimized H100, how fast is it for running large models?

Nvidia’s announcement provides data for two models, Llama 2 and GPT-J-6B.

On the optimized H100, the inference speed of running Llama 2 is 4.6 times that of the A100 and 1.77 times that of the unoptimized H100 in August

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

The inference speed of GPT-J-6B is 8 times that of A100 and 2 times that of the August unoptimized version.

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

TensorRT also provides an open source modular Python API that can quickly customize optimization solutions according to different LLM requirements

This API will combine the deep learning compiler with , kernel optimization, pre/post-processing and multi-node communication functions are integrated together.

There are also customized versions for common models such as GPT(2/3) and Llama, which can be "used out of the box".

Through the latest open source AI kernel in TensorRT, developers can also optimize the model itself, including the attention algorithm FlashAttention that greatly speeds up Transformer.

TensorRT is a high-performance inference engine for optimizing deep learning inference. It optimizes LLM inference speed by using technologies such as mixed-precision computing, dynamic graph optimization, and layer fusion. Specifically, TensorRT improves inference speed by reducing the amount of computation and memory bandwidth requirements by converting floating-point calculations into half-precision floating-point calculations. In addition, TensorRT also uses dynamic graph optimization technology to dynamically select the optimal network structure based on the characteristics of the input data, further improving the inference speed. In addition, TensorRT also uses layer fusion technology to merge multiple computing layers into a more efficient computing layer, reducing computing and memory access overhead and further improving inference speed. In short, TensorRT has significantly improved the speed and efficiency of LLM inference through a variety of optimization technologies

First of all, it is due to TensorRT's optimization of multi-node collaborative working.

A huge model like Llama cannot be run on a single card. It requires multiple GPUs to run together.

In the past, this work required people to manually disassemble the model to achieve it.

With TensorRT, the system can automatically split the model and run it efficiently between multiple GPUs through NVLink

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

Secondly, TensorRT also An optimized scheduling technology called Dynamic Batch Processing is used.

During the inference process, LLM is actually performed by executing model iterations multiple times

Dynamic batch processing technology will kick out the completed sequence immediately instead of waiting for the entire batch of tasks Once complete, process the next set of requests.

In actual tests, dynamic batch processing technology successfully reduced LLM's GPU request throughput by half, thereby significantly reducing operating costs

Another key point is Convert 16-bit precision floating point numbers to 8-bit precision , thereby reducing memory consumption.

Compared with FP16 in the training phase, FP8 has lower resource consumption and is more accurate than INT-8. It can improve performance without affecting the accuracy of the model

Usage Hopper Transformer engine, the system will automatically complete the conversion and compilation of FP16 to FP8, without the need to manually modify any code in the model

Currently, the early bird version of TensorRT-LLM is available for download, and the official version will be launched in a few weeks And integrated into the NeMo framework

One More Thing

Whenever a big event occurs, the figure of "Leewenhoek" is indispensable.

In Nvidia’s announcement, it mentioned cooperation with leading artificial intelligence companies such as Meta, but did not mention OpenAI

From this announcement, some netizens discovered this point and sent it to On the OpenAI forum:

Please let me see who has not been cueed by Lao Huang (manual dog head)

Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed

Are you still What kind of "surprises" do we expect Lao Huang to bring us?

The above is the detailed content of Lao Huang gives H100 a boost: Nvidia launches large model acceleration package, doubling Llama2 inference speed. For more information, please follow other related articles on the PHP Chinese website!

Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn

Hot AI Tools

Undresser.AI Undress

Undresser.AI Undress

AI-powered app for creating realistic nude photos

AI Clothes Remover

AI Clothes Remover

Online AI tool for removing clothes from photos.

Undress AI Tool

Undress AI Tool

Undress images for free

Clothoff.io

Clothoff.io

AI clothes remover

AI Hentai Generator

AI Hentai Generator

Generate AI Hentai for free.

Hot Article

R.E.P.O. Energy Crystals Explained and What They Do (Yellow Crystal)
2 weeks ago By 尊渡假赌尊渡假赌尊渡假赌
Repo: How To Revive Teammates
4 weeks ago By 尊渡假赌尊渡假赌尊渡假赌
Hello Kitty Island Adventure: How To Get Giant Seeds
4 weeks ago By 尊渡假赌尊渡假赌尊渡假赌

Hot Tools

Notepad++7.3.1

Notepad++7.3.1

Easy-to-use and free code editor

SublimeText3 Chinese version

SublimeText3 Chinese version

Chinese version, very easy to use

Zend Studio 13.0.1

Zend Studio 13.0.1

Powerful PHP integrated development environment

Dreamweaver CS6

Dreamweaver CS6

Visual web development tools

SublimeText3 Mac version

SublimeText3 Mac version

God-level code editing software (SublimeText3)

How to dynamically create an object through a string and call its methods in Python? How to dynamically create an object through a string and call its methods in Python? Apr 01, 2025 pm 11:18 PM

In Python, how to dynamically create an object through a string and call its methods? This is a common programming requirement, especially if it needs to be configured or run...

How does Uvicorn continuously listen for HTTP requests without serving_forever()? How does Uvicorn continuously listen for HTTP requests without serving_forever()? Apr 01, 2025 pm 10:51 PM

How does Uvicorn continuously listen for HTTP requests? Uvicorn is a lightweight web server based on ASGI. One of its core functions is to listen for HTTP requests and proceed...

In the ChatGPT era, how can the technical Q&A community respond to challenges? In the ChatGPT era, how can the technical Q&A community respond to challenges? Apr 01, 2025 pm 11:51 PM

The technical Q&A community in the ChatGPT era: SegmentFault’s response strategy StackOverflow...

How to gracefully handle 'pipe closed' error in Python multi-process Pipe communication? How to gracefully handle 'pipe closed' error in Python multi-process Pipe communication? Apr 01, 2025 pm 11:12 PM

Python multi-process Pipe error "Pipe is closed"? When using the Pipe method in Python's multiprocessing module for parent-child process communication, you may encounter...

How to solve the problem of missing dynamic loading content when obtaining web page data? How to solve the problem of missing dynamic loading content when obtaining web page data? Apr 01, 2025 pm 11:24 PM

Problems and solutions encountered when using the requests library to crawl web page data. When using the requests library to obtain web page data, you sometimes encounter the...

How to use Go or Rust to call Python scripts to achieve true parallel execution? How to use Go or Rust to call Python scripts to achieve true parallel execution? Apr 01, 2025 pm 11:39 PM

How to use Go or Rust to call Python scripts to achieve true parallel execution? Recently I've been using Python...

What is the reason for redirecting 404 errors after logging in with Selenium? How to solve it? What is the reason for redirecting 404 errors after logging in with Selenium? How to solve it? Apr 01, 2025 pm 10:54 PM

Solution to Redirecting 404 Errors after Simulation Login When using Selenium for Simulation Login, we often encounter some difficult problems. �...

Flask-SQLAlchemy How to serialize ORM objects to avoid the 'Object of type User is not JSON serializable' error? Flask-SQLAlchemy How to serialize ORM objects to avoid the 'Object of type User is not JSON serializable' error? Apr 01, 2025 pm 10:15 PM

A thorough analysis of the problem of Flask-SQLAlchemyORM object serialization when using PythonFlask framework combined with SQLAlchemy...

See all articles