Name: 英偉達如何改變遊戲規則 (How Nvidia Changed the Game)
Uploaded: 2024-09-04T11:23:25.000Z
Duration: 9 min 24 s

I've heard this numerous times from the man who founded the company Jensen Wang.

我多次聽公司創始人王俊森說過這句話。

It's Nvidia And yet many think this multi-trillion dollar artificial intelligence pioneer is called Nvidia But it's fitting in a way because Nvidia now one of the world's most valuable companies emerged from an industry also often misunderstood Gaming.

這就是 Nvidia（英偉達），但很多人都認為這家市值數萬億美元的人工智能先驅公司叫 Nvidia（英偉達）。但從某種程度上來說，這也很恰當，因為 Nvidia（英偉達）現在是全球最有價值的公司之一，而它所處的行業也經常被人誤解為遊戲。

The global video game industry is in a hundred and eighty nine billion dollar industry That's larger than movies and music combined Today Nvidia alone has a market value more than ten times that Eclipsing those of rivals and that's because Nvidia now says PC graphics cards the pieces of hardware that render 3d visuals in games Were never intended to be the company's final form The story behind that is a reminder to never underestimate the scale of the video game industry or its role Influencing some of the world's biggest companies and their technologies and it's a tale that can be traced back to the time of the dinosaurs Nvidia spawned during April of 1993 in Silicon Valley two months before the premiere of Steven Spielberg's Jurassic Park What the blockbuster glossed over in terms of paleontological accuracy critics agree it made up for with exhilarating computer-generated reptiles Nvidia is absolutely central to the graphics industry now Essentially dominates it but when it first got going it was just one of a dozens of companies that had the same Right idea Spielberg used workstations from one of those businesses Silicon graphics to bring his dinosaurs to life, but in 1993 Hollywood's need for CGI was just too small for Nvidia Initially, they were just looking for a market that was big enough to generate the kind of revenue Which would then support the R&D effort and the design effort for their products and the only market back in the day that really was Capable of supporting this kind of an effort was computer gaming This is the Xbox the public witnessed the first major triumph when in 2001 Microsoft's then CEO Bill Gates decloaked his company's inaugural competitor in the console gaming battlefield.

全球電子遊戲產業價值高達 1809 億美元，比電影和音樂的總和還要大。如今，僅 Nvidia 一家公司的市值就已超過 10 倍，令競爭對手黯然失色。這個故事可以追溯到恐龍時代 1993 年 4 月，在史蒂文-斯皮爾伯格（Steven Spielberg）的《侏羅紀公園》（Jurassic Park）首映前兩個月，Nvidia 在硅谷成立。Nvidia 現在絕對是圖形行業的核心，基本上佔據了主導地位，但在它剛剛起步的時候，它只是擁有相同想法的數十家公司之一，斯皮爾伯格使用了其中一家公司的工作站硅顯卡來讓恐龍

That's fascinating that game console was based really upon PC hardware and Nvidia's big break there was that the graphics chip that was part of that initial console was one of theirs It was a vital stepping stone that aligned Nvidia hardware with PCs bought by hardcore gamers keen on upgrading their computers Components something they couldn't do with consoles I have observed that gamers might replace their graphics cards every three to five years the better the graphics card the better the gaming Experience in a lot of instances a lot of these newer games are very very graphically demanding Take a look at the difference.

令人著迷的是，那款遊戲機是基於個人電腦硬件開發的，而 Nvidia 的重大突破在於，最初的遊戲機所採用的圖形芯片就是他們的產品，這是將 Nvidia 硬件與熱衷於升級電腦的鐵桿玩家所購買的個人電腦相匹配的重要一步。我注意到，遊戲玩家可能每三到五年就更換一次顯卡，顯卡越好，遊戲體驗就越好。

This is Halo combat evolved one of the flagship games released for the Xbox in 2001 Compare it to this demo of a software development framework called Unreal Engine used by many modern games This too is being rendered using an Nvidia chip and side-by-side.

這是 2001 年 Xbox 推出的旗艦遊戲之一《光環》（Halo combat）的進化版。將其與許多現代遊戲使用的名為虛幻引擎（Unreal Engine）的軟件開發框架的演示進行比較，這也是使用 Nvidia 芯片並排渲染的。

That's what two decades of progress looks like Revenue climbed as the company continued scoring lucrative deals and shipped graphics cards to major PC manufacturers of the time But something else happened during those 20 or so years People started to ask if these things are so good at transforming code into people and vehicles.

這就是二十年進步的樣子 隨著公司不斷獲得利潤豐厚的交易並向當時的主要 PC 製造商交付顯卡，收入也在不斷攀升。

GPUs excel at something called parallel computing Taking a complex problem dividing it up into small sets of calculations and solving them in parallel very quickly So anything that is sort of numerically intensive anything that requires a lot of say Coordinate data a lot of imaging or mapping that kind of thing TPUs are much much better at than the standard option Which would be the central processor from Intel this made the technology relevant to all kinds of non gaming industries Everything from oil and gas exploration to weather mapping a lot of the time You see that there are runs on these graphics cards because they have a lot of applications even outside of gaming The demand used for chips to my cryptocurrency is becoming an important driver over the pandemic People were lining up outside of your local micro center to get the new 30 series and video GPUs because so many people were using them for crypto mining at that time and they weren't available to people who just wanted to play video games with them More recently companies acquire GPUs on mass for use with artificial intelligence Open AI's chat GPT for instance uses NVIDIA hardware, which brings us to a catalyst behind NVIDIA's recent extraordinary rise generative AI and cats More than a decade ago Engineers at Google developed a machine learning system that taught itself what a cat was and how to create an image of one by watching A lot of YouTube videos the era of deep learning and neural networks had begun When a graphics card is doing what it was designed to do, which is to generate an image it's doing that It's creating a picture of a cat.

GPU 擅長並行計算 將一個複雜的問題分割成小的計算集，並以並行的方式快速解決，所以任何數值密集型的東西，任何需要大量座標數據、大量成像或製圖的東西，TPU 都比標準的選擇（英特爾的中央處理器）要強得多。人們在當地的微型中心外排隊購買新的 30 系列 GPU 和視頻 GPU，因為當時有很多人將它們用於加密貨幣挖礦，而那些只想用它們玩視頻遊戲的人則無法獲得這些 GPU、十多年前，谷歌的工程師們開發出了一種機器學習系統，通過觀看大量 YouTube 視頻，該系統可以自學貓是什麼，以及如何創建貓的影像，深度學習

This is where the eyes are This is where the nose is the cat flicks its head to one side to look at something You've got to calculate where that nose is where those pixels that represent the nose the eyes the ears have moved in the image Think about that the opposite way if you are Showing a computer lots and lots of pictures of a cat what it has to be very good at is Recognizing points creating a pattern effectively creating the reverse of an image a Mathematical picture of what a cat looks like.

這就是眼睛的位置 這就是鼻子的位置 貓把頭扭向一邊看東西時，你必須計算出鼻子的位置 這些代表鼻子、眼睛和耳朵的像素在影像中移動的位置 換個角度想一想，如果你向電腦展示大量貓的圖片，那麼電腦必須非常擅長識別點，創建模式，有效地創建影像的反面，即一幅貓的數學影像。

So that's what graphics chips are really good at is building these mathematical Constructs of real-life things whether it's sound whether it's image whether it's other kinds of things that AI is Going to have an influence on NVIDIA became the leader it is today by advancing technology designed for games but valuable to disparate industries It knew far in advance that to succeed it needed gamers and today That's something a lot of technology companies understand and want to take advantage of there are three billion gamers in the world And those aren't just the people who are shelling out for a pricey NVIDIA or AMD graphics card or a really beautiful PC rig it's not just the domain of You know the kind of person you picture when you picture a gamer anymore Every company that is a big tech company has some sort of interest in the video game industry right now Amazon Google even Netflix are prime examples of this Netflix executives have made pretty clear in the past couple years that they see gaming as a threat for attention and entertainment time They've cited the game fortnight as a competitive threat And this is relevant to NVIDIA's story because it's symptomatic of just how greatly the games industry has evolved over the past decade and how Even streaming services meant there was no existential threat to its bottom line There are simply now more gamers more devices and more opportunities A lot of games that were previously designed for consoles and PCs are getting ported over to mobile phones so that they can reach Audiences that were previously not accessible to them We're talking triple-a games here like Capcom's popular Resident Evil series Which have similar graphics on a phone as they do on a console People who had never even considered owning a console can now play Call of Duty can now play League of Legends and all these games That are considered hardcore video games But in a more casual fun way on technology that everybody owns Shopping list nobody looked at the first iPhone in Steve Jobs's hand and thought to themselves one day that thing's gonna transform Banks and how we get groceries delivered But here we are AI is probably going to be the same that all of this rush to create this infrastructure that this platform is Fantastic, but the end use that the profound effect on the economy We just don't know what it is yet and videos benefiting from the world trying to find out Gaming's gone from being its main source of revenue to something of a rounding error We are large consumers of it because we create a lot of AI ourselves because our chips aren't possible to design without AI Our software is not possible to write without AI We also use it for an AI foundry.

是以，圖形芯片真正擅長的是構建真實事物的數學結構。無論是聲音還是影像，無論是人工智能將要影響的其他類型的事物，英偉達都是通過推進專為遊戲設計但對各行各業都有價值的技術而成為今天的領導者的。世界上有 30 億遊戲玩家，他們不只是花錢購買昂貴的英偉達（NVIDIA）或 AMD 顯卡或漂亮的電腦設備的人，也不只是你想象中的那種遊戲玩家。現在，亞馬遜、谷歌甚至 Netflix 都是最好的例子。Netflix 的高管們在過去幾年中明確表示，他們認為遊戲是對注意力和娛樂時間的威脅。現在，遊戲玩家越來越多，設備越來越多

 We make AI models for companies that would like to Have expertise in doing so and so we are a foundry for AI like TSMC is a foundry for our chips The video games industry the focus for NVIDIA's earliest innovations and partnerships is bigger and more accessible than ever And the company still plays a pivotal role within it What's going to be most exciting is seeing what access to advanced AI makes possible for game developers working in NVIDIA's world Maybe it'll even be exciting enough to tempt valve into making Half-Life 3 We can dream

NVIDIA 最早的創新和合作夥伴關係所關注的視頻遊戲行業比以往任何時候都要龐大，也更容易進入，而該公司仍在其中扮演著舉足輕重的角色。