Skip to Content
皮衣与硅片
第二部:向死而生(1993—1998)

第七章:五百万美元的救命稻草(1996—1997)

Chapter Seven: The Five-Million-Dollar Lifeline (1996–1997)

1996年初秋的东京羽田区大鸟居,铅灰色的云层低垂在世嘉企业(Sega Enterprises)总部大楼的上方。

In early autumn 1996, in the Otorii district of Tokyo's Haneda area, lead-gray clouds hung low over the headquarters of Sega Enterprises.

东京:1996 年初秋,黄仁勋单刀赴会世嘉总部的“自杀式谈判”(公开资料图)
东京:1996 年初秋,黄仁勋单刀赴会世嘉总部的“自杀式谈判”(公开资料图)

当时的家用游戏机战场,正处于一场决定十年国运的惨烈大决战之中。索尼凭借支持多边形三维渲染的PlayStation横扫全球,任天堂刚刚祭出64位主机Nintendo 64负隅顽抗,而世嘉的“土星”(Sega Saturn)因为双CPU架构过于繁琐、三维性能落后,正在北美和日本市场节节败退。为了挽救危局,世嘉把全部翻盘希望都押在了秘密研发的下一代主机(代号“Katana”/“Blackbelt”,即后来的Dreamcast)上。

The home-console battlefield of the day was in the midst of a ferocious showdown that would decide ten years of destiny. Sony was sweeping the world with its polygon-rendering PlayStation; Nintendo had just unsheathed its 64-bit Nintendo 64 in a last-ditch defense; and Sega's Saturn, hobbled by a cumbersome dual-CPU architecture and lagging 3D performance, was retreating defeat by defeat across North America and Japan. To salvage the crisis, Sega staked every hope of a comeback on its secretly developed next-generation console (codename “Katana”/“Blackbelt,” later known as the Dreamcast).

日本街机厅:世嘉、索尼与任天堂正在打一场决定十年国运的主机大战(公开资料图)
日本街机厅:世嘉、索尼与任天堂正在打一场决定十年国运的主机大战(公开资料图)

操盘这场生死决战的,是世嘉美国区总裁兼集团执行副社长(后升任社长)——入交昭广(Shoichiro Irimajiri)。

Directing this battle of life and death was the president of Sega of America and executive vice president of the group (later to rise to president of Sega Enterprises): Shoichiro Irimajiri.

在等级森严、讲究论资排辈的日本商界,入交昭广是一个极其罕见的异类。这位毕业于东京大学航空工程系的传奇高管,年轻时曾是本田技研(Honda)最耀眼的天才发动机设计师,设计过横扫F1赛场的涡轮增压引擎与经典的本田CBX六缸摩托车,四十岁出头便成为本田历史上最年轻的董事,更在执掌本田美国制造工厂期间展现出了非凡的跨文化魄力。因心脏超负荷离开本田后,他于1993年被世嘉挖来力挽狂澜。这是一个既懂最硬核工程物理极限,又带有武士道式豪侠之气的商业领袖。

In the rigidly hierarchical, seniority-bound world of Japanese business, Irimajiri was a rarity of the rarest kind. A legendary executive who had graduated in aeronautical engineering from the University of Tokyo, he had been Honda's most dazzling young genius of engine design — creator of turbocharged engines that swept the Formula One circuits and of the classic six-cylinder Honda CBX motorcycle — had become the youngest director in Honda's history barely past forty, and had shown extraordinary cross-cultural daring while running Honda's American manufacturing plants. After an overworked heart forced him out of Honda, Sega recruited him in 1993 to turn the tide. Here was a business leader who understood the hardest physical limits of engineering and carried, withal, the chivalrous spirit of the samurai.

这天上午,三十三岁的黄仁勋穿着一身笔挺却略显紧绷的深色西装,独自走进了入交昭广的办公室。

That morning, thirty-three-year-old Huang, in a dark suit crisp but slightly too tight, walked alone into Irimajiri's office.

在此之前,黄仁勋已经在东京酒店的房间里对着镜子演练了无数遍措辞。按照正常的商业法务逻辑,他接下来要说的每一句话都是在自寻死路:英伟达拿了世嘉的前期研发经费,耗时一年多开发的NV2芯片走错了技术路线;如果违约中止开发,世嘉不仅一分钱尾款都不会付,甚至完全有理由把英伟达告上法庭索赔巨额违约金,直接让这家只剩三十六个人的硅谷小作坊当场破产清盘。

Before this, Huang had rehearsed his wording before the mirror of his Tokyo hotel room countless times. By the ordinary logic of business and law, every sentence he was about to speak was a bid for self-destruction: NVIDIA had taken Sega's early R&D money, and the NV2 chip it had spent more than a year developing had taken the wrong technical road; if it breached the contract and halted development, Sega would pay not a cent of the final installment — indeed it would have every right to haul NVIDIA into court for massive breach penalties, driving this Silicon Valley workshop of thirty-six people straight into bankruptcy and liquidation on the spot.

落座之后,面对年长自己二十三岁、不怒自威的入交昭广,黄仁勋没有兜任何圈子,也没有拿任何漂亮的演示文稿去粉饰太平。他深吸一口气,把残酷的真相赤裸裸地摊在了桌面上:

Once seated, facing Irimajiri — twenty-three years his senior, awesome without anger — Huang beat around no bushes and produced no glossy presentation to gild the lily. He drew a deep breath and laid the brutal truth naked on the table:

“入交先生,经过我们反复测试与评估,我必须向您坦白——我们为您设计的NV2二次曲面架构,是错的。”

“Mr. Irimajiri, after repeated testing and evaluation, I must be honest with you — the NV2 quadratic-surface architecture we designed for you is wrong.”

办公桌对面的日本老人目光微微一凛,身体向后靠进椅背,静静地看着这个额角渗汗却目光直视自己的华裔年轻人,示意他继续说下去。

Across the desk, the old Japanese executive's eyes hardened almost imperceptibly; he leaned back into his chair and watched in silence this young Chinese-American — brow beaded with sweat, gaze locked straight on his own — and motioned for him to go on.

“微软的Direct3D已经统一了开发生态,未来的三维世界属于三角形多边形,”黄仁勋的声音有些干涩,却异常清晰,“如果我们硬着头皮把NV2做完交付给世嘉,你们的新主机在技术上将从第一天起就落后于时代,世嘉会输掉这场战争;而我们如果继续在这个错误的架构上耗下去,也永远做不出真正有竞争力的产品。所以,我请求世嘉立刻终止与英伟达的合同,去寻找其他采用多边形架构的合作伙伴。”

“Microsoft's Direct3D has already unified the development ecosystem; the three-dimensional world of the future belongs to triangular polygons,” Huang said, his voice somewhat dry yet exceptionally clear. “If we grit our teeth and finish NV2 and deliver it to Sega, your new console will be behind the times technically from its very first day, and Sega will lose this war; and if we keep grinding away at this wrong architecture, we will never make a genuinely competitive product either. So I ask Sega to terminate its contract with NVIDIA at once, and to seek other partners who work in polygon architecture.”

作为前顶级发动机工程师的入交昭广,瞬间听懂了这份坦白背后的分量。在商业世界里,绝大多数供应商在明知方向错误时,都会选择隐瞒缺陷、把垃圾产品硬塞给甲方以骗取尾款。眼前这个年轻人明明自己已经快淹死了,却主动剪断了连在世嘉身上的绳索,避免把甲方一起拖下深渊。

Irimajiri, once a top-flight engine engineer himself, grasped instantly the weight behind this confession. In the business world, the overwhelming majority of suppliers, knowing the direction is wrong, choose to bury the defect and cram their garbage product down the client's throat to swindle the final payment. This young man, already half-drowned himself, had voluntarily cut the rope tied to Sega so as not to drag his client down into the abyss with him.

然而,黄仁勋接下来的第二句话,却让见多识广的入交昭广彻底愣住了。

But Huang's next sentence left even the vastly experienced Irimajiri frozen where he stood.

“但是,入交先生,”黄仁勋喉结滚动了一下,身体前倾,几乎是一字一顿地说道,“我还需要您把合同里剩下的五百万美元全额付给我们。因为如果没有这五百万美元,英伟达下个月就会破产倒闭。”

“But, Mr. Irimajiri,” said Huang, his Adam's apple bobbing, his body leaning forward, speaking nearly syllable by syllable, “I must also ask you to pay us the remaining five million dollars of the contract in full. Because without that five million dollars, NVIDIA will go bankrupt and close its doors next month.”

房间里陷入了死一般的寂静,只有墙上挂钟的秒针在滴答作响。

The room fell into a deathlike silence, broken only by the ticking of the second hand on the wall clock.

这简直是商业谈判史上最荒诞不经的逻辑:我搞砸了你的项目,我要求解除合同不给你交货,但你还得把剩下的五百万美元巨款照付给我。 哪怕放在硅谷最疯狂的沙丘路,这也是会被保安直接架出去的疯话。

It was the most absurd logic in the history of business negotiation: I wrecked your project, I am asking to dissolve the contract and deliver you nothing, and yet you must still pay me the remaining five million dollars in full. Even on the wildest stretch of Sand Hill Road, such talk would get a man escorted out by security.

入交昭广盯着黄仁勋看了足足十几秒。在那令人窒息的对视中,黄仁勋迅速抛出了他唯一的筹码——不是哀求怜悯,而是一笔置之死地而后生的投资账:“如果您把这五百万美元当作货款,那它确实打了水漂;但如果您把它变成世嘉对英伟达的股权投资,只要英伟达靠这笔钱活下来、造出下一代芯片,世嘉手里的股份就还有希望连本带利赚回来。相反,如果英伟达明天就破产清盘,世嘉之前投进来的几百万美元研发费,将彻彻底底变成零。”

Irimajiri stared at Huang for a good dozen seconds. Into that suffocating exchange of gazes, Huang swiftly threw out his only chip — not a plea for mercy, but an investment reckoning of the kind that wins life back from death: “If you treat this five million dollars as payment for goods, then it has indeed vanished down the drain. But if you turn it into an equity investment by Sega in NVIDIA, then so long as NVIDIA survives on this money and builds its next chip, the shares in Sega's hands still have a hope of returning principal and profit together. And if, on the contrary, NVIDIA goes bankrupt and liquidates tomorrow, the several million dollars Sega has already poured in as R&D funding will be reduced to absolute zero.”

入交昭广沉默着走回办公桌后,没有当场拍桌子震怒,只是淡淡地说了一句:“让我考虑一下。”

Irimajiri walked back behind his desk in silence. There was no slam of the table, no burst of fury — only a flat sentence: “Let me think about it.”

接下来的几天,是黄仁勋人生中最漫长的凌迟。远在加州桑尼维尔的三十六名员工每天都在等着他的电话,财务账面上的现金正在以小时为单位蒸发。而在东京的世嘉总部内部,法务部与财务部的高管们得知黄仁勋的请求后简直炸开了锅——所有人一致反对,认为把五百万美元汇给一家已经宣告技术失败、随时会咽气的美国小公司,无异于往东京湾里扔钞票,一旦亏损,入交昭广本人甚至要向世嘉董事会承担严重的渎职责任。

The days that followed were the slowest, longest torture of Huang's life. Far away in Sunnyvale, California, his thirty-six employees waited each day for his call while the cash on the financial statements evaporated by the hour. And inside Sega's Tokyo headquarters, learning of Huang's request, the executives of the legal and finance departments nearly blew the roof off — unanimous opposition: wiring five million dollars to a small American company that had publicly declared its own technology a failure and could stop breathing at any moment was like throwing banknotes into Tokyo Bay, and if the money were lost, Irimajiri himself would have to answer to the Sega board for grave dereliction of duty.

但入交昭广看到了财务报表之外的东西。

But Irimajiri saw something beyond the financial statements.

在这个三十三岁华裔年轻人的眼睛里,他看到了当年自己在本田实验室里彻夜调校引擎时不服输的影子,更看到了一种在绝境中依然敢于直面残酷真相的罕见诚实。“他犯了大错,但他没有撒谎,而且他依然充满斗志,”入交昭广后来回忆起那个决定命运的瞬间时说道,“我想赌这个年轻人一把。”

In the eyes of this thirty-three-year-old Chinese-American he saw a reflection of his own younger self, refusing to concede defeat through nights of tuning engines in the Honda laboratories — and something rarer still: an honesty that dared to face brutal truth even in the depths of desperation. “He made a huge mistake, but he did not lie, and he was still full of fight,” Irimajiri said later, recalling that fateful moment. “I wanted to place a bet on this young man.”

几天后,入交昭广力排众议,动用自己的最高管理权限拍板:解除NV2芯片定制合同(世嘉随后转而采用了日本NEC与VideoLogic合作的PowerVR芯片),同时将剩余的五百万美元改签为对英伟达的战略股权注资!

A few days later, Irimajiri overrode the objections of everyone around him and exercised the highest authority of his office to make the final call: the NV2 custom-chip contract would be dissolved (Sega afterward turned to the PowerVR chip of NEC and VideoLogic in Japan), and the remaining five million dollars would be re-signed as a strategic equity injection into NVIDIA!

当这笔五百万美元的跨国电汇真正落入英伟达几近干涸的银行账户时,桑尼维尔办公室里的黄仁勋握着电话听筒,后背的衬衫早已被冷汗彻底浸透。

When that five-million-dollar transpacific wire transfer truly landed in NVIDIA's all but evaporated bank account, Huang, in the Sunnyvale office, was gripping the telephone receiver, the back of his shirt already soaked through with cold sweat.

这五百万美元,为英伟达买来了至关重要的六个月寿命。(命运在几年后给了入交昭广最丰厚的回报:2000年前后,当入交昭广因Dreamcast主机最终不敌索尼PS2而引咎辞去世嘉社长职务时,世嘉财务部清点资产,赫然发现当年入交昭广“施舍”给英伟达的那五百万美元原始股,在英伟达上市狂飙后已经暴涨到了一千五百万美元以上,成了亏损累累的世嘉账面上最救命的一笔现金收益。黄仁勋终身铭记这份救命之恩,不仅在二十多年后公开撰文致谢,每次去东京也必定拜访这位昔日的恩人。)

That five million dollars bought NVIDIA its vitally important six months of life. (Fate would repay Irimajiri most handsomely a few years on: around 2000, when he resigned the Sega presidency taking responsibility for the Dreamcast console's ultimate defeat by Sony's PlayStation 2, Sega's finance department, taking stock of its assets, discovered to its astonishment that the five million dollars of original shares Irimajiri had “bestowed” on NVIDIA had soared to fifteen million dollars or more after NVIDIA's IPO went on its wild ascent — the single most lifesaving cash gain on the ledger of a loss-ridden Sega. Huang carried that debt of salvation with him for life: not only did he publicly write his thanks more than twenty years later, he made a point of calling on his old benefactor on every visit to Tokyo.)

然而,在1996年底的寒冬里,五百万美元绝不意味着安全上岸——它只是一颗装进左轮手枪枪膛里的最后一颗子弹。

Yet in the deep winter at the close of 1996, five million dollars meant nothing like safe passage ashore — it was only the last bullet loaded into the chamber of a revolver.

黄仁勋把留下的三十六名核心工程师全部召集到会议室,白板上写着冷冰冰的倒计时:六个月,流片一次,不成功便成仁。

Huang summoned all thirty-six remaining core engineers into the conference room, where the whiteboard bore an icy countdown: six months, one tape-out; succeed or die.

在当时的半导体行业,设计一颗全新的复杂图形芯片(代号NV3),常规流程是先画出电路图,送到晶圆厂流片(Tape-out)造出第一批工程样品,拿回来在电路板上测试找Bug,修改设计后再做第二次、第三次流片,整个周期通常需要一年半到两年。

In the semiconductor industry of the day, the standard process for designing an all-new, complex graphics chip (codename NV3) was to draw the circuit schematics first, send them to a fab for tape-out to produce the first batch of engineering samples, bring them back to test on circuit boards and hunt for bugs, revise the design, then run a second and a third tape-out — a cycle that normally took a year and a half to two years.

但英伟达账上的钱,连第二次流片的掩膜版(Mask)费用都付不起了,更撑不到一年半之后——此时3dfx的Voodoo显卡正在市场上疯狂收割钞票,ATI和S3也在虎视眈眈。如果NV3第一批芯片从晶圆厂拿回来点不亮,或者存在致命硬件缺陷,英伟达将连修改的机会都没有,当场暴毙。

But the money in NVIDIA's account could no longer cover even the mask-set fees for a second tape-out, let alone sustain them a year and a half — at this very moment 3dfx's Voodoo graphics cards were madly harvesting cash in the market, with ATI and S3 circling hungrily alongside. If the first batch of NV3 chips came back from the fab and refused to light up, or carried a fatal hardware defect, NVIDIA would not even get the chance to revise them; it would die on the spot.

“我们没有时间做物理试错了,”黄仁勋的眼神冷得像刀锋,“必须在芯片真正送进晶圆厂之前,在虚拟世界里把所有的Bug全部杀光!”

“We have no time left for trial and error in physical silicon,” said Huang, his eyes cold as a blade's edge. “Before the chip ever enters the fab, every last bug must be slaughtered in a virtual world!”

为了实现这个看似不可能的目标,黄仁勋做出了一个令所有人心脏骤停的疯狂举动:他从仅有的几百万美元救命钱里,一咬牙掏出巨资,买下了一套当时全硅谷最昂贵、最前沿的IKOS系统硬件仿真器(Hardware Emulator)。

To reach that seemingly impossible goal, Huang made a mad move that stopped every heart in the company: with a grimace he tore a huge sum out of the few million dollars of life-saving money and bought the IKOS system hardware emulator — the most expensive, most cutting-edge equipment in all of Silicon Valley at the time.

那是一台体型庞大如冰柜、造价高达上百万美元的专用超级计算设备,连顶级芯片巨头都舍不得轻易采购。它可以在没有真实硅片的情况下,通过成千上万颗可编程逻辑门阵列,在软件层面完整模拟出整颗NV3芯片的每一个时钟周期与晶体管跳动。

It was a dedicated supercomputing machine built like a freezer and costing upward of a million dollars, one that even the top chip titans balked at purchasing. Without a single piece of real silicon, it could use tens of thousands of programmable logic gates to simulate, at the software level, every clock cycle and every transistor's twitch of the entire NV3 chip.

不仅如此,黄仁勋还彻底颠覆了整个半导体行业的研发流水线。

And that was not all: Huang went on to overturn, root and branch, the R&D assembly line of the entire semiconductor industry.

过去,所有的芯片公司都是“串联作业”:硬件团队花一年把芯片造出来,扔给软件驱动团队;软件团队再花半年写驱动程序去适配微软的系统。黄仁勋一脚踹翻了这堵墙——依托那台昂贵的IKOS仿真器,在NV3连一片真实硅晶圆的影子都还没有的时候,他就逼着软件工程师们直接在虚拟仿真器上同步编写Windows 95与Direct3D驱动程序,甚至把热门游戏的测试代码直接灌进仿真器里一帧一帧地跑!

Until then, every chip company had worked “in series”: the hardware team spent a year building the chip and tossed it over the wall to the software-driver team; the software team then spent another six months writing drivers to fit Microsoft's systems. Huang kicked that wall flat. Leaning on that costly IKOS emulator, at a time when NV3 had yet to see even the shadow of a real silicon wafer, he forced the software engineers to write Windows 95 and Direct3D drivers in parallel, directly on the virtual emulator — and even poured the test code of hit games into the machine to run frame by frame!

虽然仿真器的运行速度比真实芯片慢上成千上万倍——真实游戏里一秒钟闪过的三十帧画面,在仿真器上可能要跑上整整一个通宵——但这意味着,硬件架构师与软件程序员每天都在同一条战壕里贴身肉搏。任何一个微小的渲染管线死锁或寄存器冲突,在画图纸的当天晚上就会被软件团队揪出来当场改掉。

True, the emulator ran thousands upon thousands of times slower than real silicon — the thirty frames a real game flashes past in one second might take the emulator an entire night to churn through — but it meant the hardware architects and the software programmers fought hand to hand in the same trench every single day. Any tiny deadlock in the rendering pipeline or clash of registers would be dragged out by the software team the very evening the schematic was drawn, and fixed on the spot.

那是桑尼维尔弗里蒙特大道上最疯狂的一百八十个日夜。

Those were the maddest one hundred and eighty days and nights Fremont Avenue in Sunnyvale would ever see.

三十六个人的团队像上了发条的敢死队,吃住几乎全在办公室里。就在这生死存亡的关头,黄仁勋还干了一件展现其顶级猎头嗅觉的大事:1997年初,趁着昔日贵族硅谷图形公司(SGI)内部动荡,他用极具煽动性的愿景,把SGI顶尖的三维图形首席科学家大卫·柯克(David Kirk)挖到了这艘还在漏水的破船上。柯克的加盟,瞬间将英伟达的多边形光栅化与浮点运算实力拉升到了世界之巅。

The thirty-six-person team ran like a suicide squad with wound-up springs, eating and sleeping almost entirely in the office. At this hinge of survival, Huang also pulled off a move that displayed his top-tier headhunter's instincts: in early 1997, exploiting the internal turmoil of the old aristocrat Silicon Graphics (SGI), he used a vision of incendiary power to poach SGI's chief scientist of 3D graphics, David Kirk, onto this leaking wreck of a ship. With Kirk aboard, NVIDIA's polygon rasterization and floating-point muscle was lifted to the very summit of the world in an instant.

在这座高压锅的中心,三十四岁的黄仁勋每天清晨六点准时出现在办公室,直到午夜才离开。那句日后被印在无数商业教材上、甚至在英伟达市值突破万亿美元后依然在每次全员大会开场时被他反复念诵的冷酷箴言,正是从1997年那些濒死挣扎的清晨里淬炼出来的:

At the center of this pressure cooker, thirty-four-year-old Huang appeared at the office punctually at six each morning and did not leave until midnight. That cold maxim — later printed in countless business textbooks, the one he still recites at the opening of every all-hands meeting even now that NVIDIA's market value has passed a trillion dollars — was tempered in those near-death mornings of 1997:

“各位,我们距离倒闭,只有三十天(Our company is thirty days from going out of business)。”

“Everyone, we are thirty days from going out of business.”

在那个阶段,这绝不是什么企业家用来鞭策员工的修辞鸡汤,而是字面意义上的银行对账单现实。

At that stage this was no rhetorical chicken soup for goading employees, the kind entrepreneurs love; it was, in the most literal sense, the reality of the bank statement.

1997年春,耗尽最后一滴心血的NV3设计图终于定稿,正式发往太平洋彼岸的台湾——交到了新竹科学园区由张忠谋执掌的台积电(TSMC)与意法半导体的0.35微米晶圆产线上。黄仁勋与马拉科夫斯基、普里姆把这颗承载着全公司性命的芯片正式命名为:RIVA 128(Real-time Interactive Video and Animation accelerator,实时交互视频与动画加速器,128代表其惊人的128位显存位宽)。

In the spring of 1997, the NV3 design — drained of its last drop of heart's blood — was at last finalized and formally dispatched across the Pacific to Taiwan, onto the 0.35-micron wafer production lines of TSMC under Morris Chang and of SGS-Thomson in the Hsinchu Science Park. Huang, Malachowsky, and Priem formally christened the chip that carried the whole company's life: RIVA 128 (the Real-time Interactive Video and Animation accelerator — the 128 standing for its astonishing 128-bit memory bus).

在等待第一批工程样片寄回加州的那几个星期里,连一向泰山崩于前而色不变的黄仁勋,也陷入了失眠的煎熬。所有人都清楚,快递箱拆开的那一秒,决定的不仅是三十六个家庭的饭碗,更是过去四年所有屈辱与豪赌的终极审判。

In the weeks of waiting for the first engineering samples to arrive back in California, even Huang — the man who never changed color though a mountain collapsed before him — sank into the torment of insomnia. Everyone understood that the second the shipping carton was opened would decide not only the rice bowls of thirty-six families but the final judgment on four years of humiliation and wild gambles.

一个阳光刺眼的早晨,装有首批RIVA 128封装样片的防静电黑盒终于送抵桑尼维尔实验室。

On a morning of glaring sunlight, the antistatic black box holding the first packaged RIVA 128 samples finally reached the Sunnyvale laboratory.

全公司上下几十号人密密麻麻地围在测试台四周,连大气都不敢喘一口。硬件工程师颤抖着手将那块还带着余温的绿色工程显卡插进测试主板插槽,接通电源,按下了开机键——一秒,两秒,三秒……显示器漆黑的屏幕骤然亮起!没有任何花屏,没有任何死机,借助在IKOS仿真器上早就千锤百炼成熟的驱动程序,Windows桌面瞬间弹了出来,紧接着,微软Direct3D基准测试程序里那架精细流畅的三维战斗机,以压倒性的超高帧率在屏幕上呼啸盘旋!

The company's dozens of people packed densely around the test bench, not one daring to breathe. With trembling hands a hardware engineer slotted that still-warm green engineering graphics card into the test motherboard, connected power, and pressed the switch — one second, two seconds, three seconds... and the monitor's pitch-black screen blazed to life! No corrupted pixels, no freeze; on the strength of drivers long hammered to maturity on the IKOS emulator, the Windows desktop popped up at once, and then the finely detailed, silk-smooth 3D fighter jet from Microsoft's Direct3D benchmark wheeled and howled across the screen at an overwhelming, superhuman frame rate!

搭载 RIVA 128 芯片的板卡(ELSA 制造):一次流片点亮的生死奇迹(资料图)
搭载 RIVA 128 芯片的板卡(ELSA 制造):一次流片点亮的生死奇迹(资料图)

实验室里先是死寂了一秒,随即爆发出掀翻屋顶的狂吼、哭喊与击掌声。有连续熬了几个月通宵的年轻工程师直接瘫坐在地板上捂脸痛哭。

For one second the lab was deathly still — then it erupted into roars, sobs, and clapping that threatened to lift the roof off. A young engineer who had pulled all-nighters for months on end simply collapsed onto the floor, covering his face and weeping.

黄仁勋死死盯着屏幕上飞速跳动的帧率数字,紧绷了整整两年的下颌线终于微微松动。他知道,那把悬在脖颈上方毫厘之处的断头铡刀,终于被他们徒手震碎了。

Huang stared fixedly at the frame-rate digits leaping across the screen, and the jawline clenched for two full years finally loosened, just a little. He knew that the executioner's blade hanging a hair's breadth above their necks had, at last, been shattered by their own bare hands.

第七章:五百万美元的救命稻草(1996—1997) · 第 1/44 段