Name: 亞馬遜 EC2 P6e-GB200 UltraServers 簡介：為大規模前沿人工智能提供動力 | 亞馬遜網絡服務 (Introducing Amazon EC2 P6e-GB200 UltraServers: Powering Frontier AI at Scale | Amazon Web Services)
Uploaded: 2025-08-30T13:14:48.000Z
Duration: 7 min 59 s

At AWS, it's our mission to offer the best place for AI training and inference.

在 AWS，我們的使命是為人工智能培訓和推理提供最佳場所。

And doing this means building the best compute infrastructure possible to power the most innovative applications of today and tomorrow.

而要做到這一點，就意味著要儘可能構建最佳的計算基礎設施，為當今和未來最具創新性的應用提供動力。

And that's why I'm super excited to say that NVIDIA Blackwell is now here on AWS.

是以，我非常高興地宣佈，英偉達™（NVIDIA®）Blackwell 現已登陸 AWS。

NVIDIA Blackwell marks a new chapter for accelerated computing.

NVIDIA Blackwell 標誌著加速計算的新篇章。

It delivers significant improvements in performance, cost, and energy efficiency, enabling developers to tackle the world's most challenging problems.

它在性能、成本和能效方面都有顯著提高，使開發人員能夠解決世界上最具挑戰性的問題。

AWS is now bringing the best of NVIDIA Blackwell to the cloud in two system configurations.

AWS 現已通過兩種系統配置將 NVIDIA Blackwell 的最佳性能帶入雲端。

Today we're announcing the general availability of both the Amazon EC2 P6 instance, accelerated by NVIDIA's HGX B200, and Amazon's EC2 P6e Ultra Server, accelerated by NVIDIA's GB200

今天，我們宣佈由英偉達™（NVIDIA®）HGX B200 加速的亞馬遜 EC2 P6 實例和由英偉達™（NVIDIA®）GB200 加速的亞馬遜 EC2 P6e Ultra 服務器全面上市。

NVL72, our most powerful EC2 instance ever.

NVL72，我們有史以來最強大的 EC2 實例。

Our NVIDIA HGX B200 delivers top performance for x86 scale-up systems.

我們的 NVIDIA HGX B200 可為 x86 擴展系統提供頂級性能。

And our NVIDIA GB200 NVL72 is a rack-scale, liquid-cooled supercomputer that enables 72

英偉達™（NVIDIA®）GB200 NVL72 是一款機架式液冷超級計算機，可實現 72

NVIDIA Blackwell GPUs to act as a single massive GPU.

英偉達™（NVIDIA®）Blackwell GPU 可作為單個大型 GPU 運行。

These instances will help power training and inference for the latest frontier models, providing up to 30 times greater performance than NVIDIA Hopper.

這些實例將為最新前沿模型的訓練和推理提供支持，其性能比英偉達™ Hopper 高出 30 倍。

Blackwell on AWS sets a new bar for performance in EC2, but it's more than meets the eye.

Blackwell on AWS 為 EC2 的性能設定了新的標準，但它遠不止於此。

Let's take a closer look at the unique design decisions and innovations we've made to bring

讓我們來詳細瞭解一下我們的獨特設計決策和創新，它們為我們帶來了

Modern agentic AI workloads are presenting ever-increasing demands for bandwidth in the quest for optimal performance.

在追求最佳性能的過程中，現代代理人工智能工作負載對帶寬的要求越來越高。

This requires solutions for both scaling out and scaling up.

這就需要既能縮小規模又能擴大規模的解決方案。

For Blackwell's scaled-out network, we've opted to use our in-house networking stack, the Elastic Fabric Adapter.

對於 Blackwell 的擴展網絡，我們選擇使用內部網絡協議棧 Elastic Fabric Adapter。

Developed by the AWS Nitro system, EFA leverages our latest generation of custom-designed Nitro controllers, allowing us to achieve up to 28.8 terabits per second of networking bandwidth per Ultra server, and scale to EC2 Ultra clusters with up to hundreds of thousands of GPUs.

EFA 由 AWS Nitro 系統開發，利用了我們最新一代定製設計的 Nitro 控制器，使我們能夠實現每臺 Ultra 服務器每秒高達 28.8 太比特的網絡帶寬，並可擴展至擁有多達數十萬 GPU 的 EC2 Ultra 集群。

But Nitro does more than allow us to scale these large clusters.

但是，Nitro 不僅僅能讓我們擴展這些大型集群。

It allows us to further improve both the reliability and predictability.

它使我們能夠進一步提高可靠性和可預測性。

And that's because EFA runs over Scalable Reliable Datagram, or SRD, a bespoke, topology-aware networking protocol.

這是因為 EFA 通過可擴展可靠數據報或 SRD 運行，SRD 是一種定製的拓撲感知網絡協議。

With SRD, Nitro and EFA automatically route traffic around failures on the network, avoiding slowdowns or downtime.

通過 SRD，Nitro 和 EFA 可自動繞過網絡故障進行流量路由，避免出現速度減慢或停機。

For the scaled-up network, the various GB200 compute trays are connected as a tightly coupled mesh through NVIDIA NVLink, which creates a single 72 Blackwell GPU NVLink domain with 132 terabits per second of total bandwidth, served as a single Ultra server.

在擴展網絡方面，各種 GB200 計算托盤通過英偉達™（NVIDIA®）NVLink 連接成一個緊密耦合的網狀結構，從而創建了一個單一的 72 Blackwell GPU NVLink 域，總帶寬為每秒 132 太比特，作為單一的 Ultra 服務器提供服務。

To manage these NVLink switches, we leverage our Nitro-powered headnode.

為了管理這些 NVLink 交換機，我們利用了由 Nitro 驅動的頭節點。

This reduces the number of network devices operating, with the NVLink network managed within the Nitro-trusted domain.

這減少了運行的網絡設備數量，NVLink 網絡由 Nitro 可信域管理。

In a fast-moving space, with new platforms going from idea to production faster than ever before, the ability to detect performance issues or bugs is critical for maximizing reliability, security, and performance for customers.

在這個快速發展的領域，新平臺從構思到投入生產的速度比以往任何時候都快，是以檢測性能問題或錯誤的能力對於最大限度地提高客戶的可靠性、安全性和性能至關重要。

With Nitro, we can gather a greater amount of telemetry, logging everything with respect to function of the hardware, providing us unique insight into detecting and understanding failures.

有了 Nitro，我們可以收集更多的遙測數據，記錄所有與硬件功能有關的資訊，為我們檢測和了解故障提供獨到的見解。

But the Nitro difference goes beyond detection.

但 Nitro 的與眾不同之處還不止於檢測。

It fundamentally changes the way we apply patches and optimizations.

它從根本上改變了我們應用補丁和優化的方式。

Typically, infrastructure providers need to choose between keeping their infrastructure online around the clock, forgoing these updates, or taking their capacity offline to update, sometimes resulting in kicking customers off the hardware.

通常情況下，基礎設施提供商需要在以下兩種情況中做出選擇：讓其基礎設施全天候在線，放棄這些更新；或者讓其能力脫機更新，有時會導致客戶無法使用硬件。

Nitro provides us the capability that no others in the industry have – live updates.

Nitro 為我們提供了業內其他公司所不具備的能力--實時更新。

Live updates allow us to update system firmware transparently, while systems remain in operation.

實時更新允許我們在系統運行時透明地更新系統固件。

This means that customer workloads continue uninterrupted, with ever-improving reliability, security, and performance, with no downtime required.

這意味著客戶的工作負載可以不間斷地繼續運行，可靠性、安全性和性能不斷提高，無需停機。

Lastly, in order to support the incredible compute density with GB200 NVL72 racks, we've moved towards a new first at AWS – liquid cooling.

最後，為了支持 GB200 NVL72 機架驚人的計算密度，我們在 AWS 首次採用了液體冷卻技術。

Blackwell represents the first liquid cooling hardware platform we've deployed at scale on AWS, and with it comes a unique set of goals and challenges.

Blackwell 是我們在 AWS 上大規模部署的第一個液體冷卻硬件平臺，隨之而來的是一系列獨特的目標和挑戰。

This meant meeting the ravenous demand for AI infrastructure, which meant we had to scale fast and scale globally.

這意味著要滿足對人工智能基礎設施的狂熱需求，這意味著我們必須快速擴大規模並擴展到全球。

When we looked at liquid cooling solutions, based on what's available in the market today, there were a few trade-offs.

我們在研究液體冷卻解決方案時，根據目前市場上的產品，需要權衡幾個方面。

First would have been building liquid-cooled data centers from scratch – something that tends to take a while, with long lead times on the order of years.

首先是從零開始建造液冷數據中心，這往往需要一段時間，籌備時間長達數年。

We knew those timelines wouldn't intersect with the availability of Blackwell.

我們知道這些時間線不會與布萊克韋爾的可用性相交。

The second option would have been adopting some off-the-shelf solutions, but these didn't scale.

第二種選擇是採用一些現成的解決方案，但這些解決方案無法擴展。

They would take up too much data center floor space, would still require major modifications to data centers, or increase water usage substantially.

它們會佔用過多的數據中心地面空間，仍然需要對數據中心進行重大改造，或者會大幅增加用水量。

And while some of these solutions could work for lower volumes at other providers, there simply wouldn't be enough liquid cooling capacity to support our scale.

雖然這些解決方案中的一些可以適用於其他供應商的較低產量，但根本沒有足夠的液體冷卻能力來支持我們的規模。

是以，我們做了我們在 AWS 經常做的事情。

We invented our own solution to have the best of both worlds.

我們發明了自己的解決方案，以求兩全其美。

This is the in-row heat exchanger, or IRHX as we call it.

這就是排內熱交換器，我們稱之為 IRHX。

The IRHX is an Amazon-designed liquid cooling system.

IRHX 是亞馬遜設計的液體冷卻系統。

It can be installed with no changes to the air-cooled mechanical design and minimal changes to existing infrastructure, meaning it can be deployed in existing data centers as well as new builds.

它的安裝無需改變風冷式機械設計，對現有基礎設施的改動也很小，這意味著它既可以部署在現有的數據中心，也可以部署在新建的數據中心。

The IRHX was designed to allow us to scale fast, standardizing our equipment and supply chain, and has been built to spec for our standard rack dimensions to fit within our existing data centers.

IRHX 的設計目的是讓我們能夠快速擴展，實現設備和供應鏈的標準化，並按照我們的標準機架尺寸製造，以適應我們現有的數據中心。

There are three components to the IRHX, a water distribution cabinet, a pumping unit, and fan coils.

IRHX 由三個部分組成：配水櫃、抽水裝置和風機盤管。

As the chips in the rack generate heat, cool liquid is pushed from the IRHX pumping unit to the servers.

當機架中的芯片產生熱量時，冷卻液就會從 IRHX 泵裝置推送到服務器。

This liquid is distributed to the chips through a cold plate design co-innovated between AWS and NVIDIA, cooling the chips substantially.

這些液體通過 AWS 和英偉達共同創新的冷板設計分佈到芯片上，從而大大冷卻了芯片。

This warm liquid is then returned to the IRHX, being passed through the coils where fans cool the liquid back down while expelling heat out of the back and into an aisle of contained hot air.

然後，這些溫熱的液體被送回 IRHX，通過線圈，風扇將液體冷卻下來，同時將熱量從後面排出，進入裝有熱空氣的通道。

And unlike many publicly available options, the IRHX can scale efficiently by adding or subtracting fan coil units depending on the amount of cooling needed by the row.

IRHX 與許多公開銷售的產品不同，它可以根據行所需的製冷量增減風機盤管設備，從而實現高效擴展。

As we like to say, with the IRHX, we don't need to design the data center around the rack.

正如我們常說的那樣，有了 IRHX，我們就不需要圍繞機架設計數據中心了。

With support for over 900 CUDAX libraries and AI models, NVIDIA Blackwell is now available for AWS customers to drive breakthroughs in science, industry, and the next era of AI reasoning.

NVIDIA Blackwell 支持超過 900 個 CUDAX 庫和人工智能模型，現在可用於 AWS 客戶，推動科學、工業和下一個人工智能推理時代的突破。

Blackwell is accessible directly in our P6 instances, accelerated by NVIDIA's HGX B200, and our P6e Ultra servers, accelerated by NVIDIA's GB200 NVL72, or through AWS managed offerings such as Amazon SageMaker HyperPod and NVIDIA DGX Cloud.

Blackwell可直接在我們的P6實例（由英偉達HGX B200加速）和P6e Ultra服務器（由英偉達GB200 NVL72加速）中訪問，也可通過AWS託管產品（如亞馬遜SageMaker HyperPod和英偉達DGX雲）訪問。

亞馬遜 EC2 P6e-GB200 UltraServers 簡介：為大規模前沿人工智能提供動力 | 亞馬遜網絡服務 (Introducing Amazon EC2 P6e-GB200 UltraServers: Powering Frontier AI at Scale | Amazon Web Services)