文 | 最话FunTalk，作者 | 王芳洁
第一颗芯片回片那天，新芯航途（XHEART）的一位同事发现，自己桌上的日历上写着“今天有好事发生。”他赶紧把这句话拍下来，发到了公司群里。
果然，不到24小时内，这颗芯片被点亮了。
整个事情当中，最引人注目的当然是那颗芯片，但还有一个人不该被忽略，当他看到好事，首先想到的是公司，然后福至心灵。
这真是一家幸运的公司，无论在事上，还是在人身上。
成立不过三年时间里，XHEART就已经推出了第一代大模型芯片X7系列，并实现了规模化量产上车。
10月29日， 这家年轻的公司召开发布会，宣布推出新一代大模型芯片——
X7 Ultra，一款拥有700 TOPS大模型算力的车规级芯片。并且通过架构与工具链协同创新，X7 Ultra支持可部署模型参数量实现10倍提升。
和X7系列一样，X7 Ultra仍然是一款由大模型定义的芯片，芯片与模型的适配被前置到了研发链路的最前端。
并且，XHEART也仍然坚持基于真实量产智驾大模型进行评测。因为在XHEART CEO李宗霖看来，通常行业用实验室里的基础测试评估芯片性能，但实验室结果与量产模型的实际体验往往相去甚远。
根据真实量产模型的评测，X7 Ultra量化部署后模型输出与训练后FP32浮点模型的相似度达到了99.99%。
四个 9的概念并不令人陌生，人们经常在黄金领域看到：纯度 99.99% 的黄金一旦被炼造出来，就会被称为足金。某种程度上，打造一款芯片也很像炼金术，蚀刻硅片，然后得到改变人类世界的科技文明。
在X7 Ultra背后，的确有一群炼芯术士，他们有的在XHEART内部，有的在产业链上下游，他们都相信被大模型定义的芯片，能够充分释放模型能力，支持中国智驾实现L4级别自动驾驶。
“今天有好事发生”，是他们的咒语。
01 几年后，市场需要什么样的芯片？
2023年，OPPO关闭了芯片设计公司哲库之后，时任哲库COO的李宗霖决定带着核心团队一起创业。他是中国芯片行业非常资深的专家，在哲库之前，李宗霖是联发科无线通信事业部总经理，主导过 Helio 处理器系列和天玑芯片系列的开发。
但哪怕在芯片行业做了快二十年，李宗霖仍然无法回避芯片行业的一大痛点：模型迭代速度很快，但芯片的研发周期以年计，那么等芯片做出来，还能满足模型的需要吗？
而在同一时间，身为Momenta的创始人，曹旭东也有自己的疑问。按照他的判断，2028 年将是 L4 级别自动驾驶的爆发点，届时模型规模将非常庞大，需要至少10B、大概率 100B 参数的模型，而现有通用芯片无法支撑——车上的功耗、车规的可靠性和成本，是一道"不可能三角"。
如果强行用通用芯片堆算力，相当于在车上放一台巨大的 GPU 服务器；向功耗和成本妥协，模型规模又做不上去。该怎么办？
机缘巧合下，两个抱有困惑的人碰到了一起。他们深聊了几次，发现这两个问题都有同一个解法——
由大模型来定义芯片。
很快，曹旭东与李宗霖决定联袂，一起做芯片。他们互相需要。
李宗霖团队需要和 Momenta 合作，让模型在最早期就介入到研发当中，先理解模型如何运行，再倒推硬件架构与软件工具链。
Momenta 同样需要李宗霖团队，因为要支撑 L4 级别的目标，它需要一款能支持模型规模的高性能芯片，而这款芯片得能上车量产，所以还得同时满足能效和成本问题，打破不可能三角。
同年，XHEART成立了, 从成立的第一天起，XHEART便确立了“大模型定义芯片”的技术路线：从真实应用中的模型需求出发，倒推芯片架构，重新组织计算、存储和数据流转。团队从芯片定义阶段就与算法团队共同研究模型架构和关键算子，为未来几年的模型演进提前做好准备。
按照这条技术路线做出来的芯片，不会被时间打败。
以X7 Ultra为例，其超大核 NPU 把模型的数据交互从传统架构的百 GB 每秒级，聚合到片内 TB 每秒级；
针对 Transformer 模型中 Linear、Softmax、Transpose 等关键算子，X7 Ultra专门设计了硬件加速引擎，且原生支持浮点高精度向量计算；
通过底层软硬协同，X7 Ultra 的计算利用率达到 98.7%；
通过AI工具链重构，内存带宽消耗相较行业旗舰模型下降 80%，能效提升 100%。
如此性能的芯片被做出来后，就不光能支持现有模型了，它还会反向推动模型的迭代。
据曹旭东介绍，Momenta即将推出的R7 Ultra世界模型，基于X7 Ultra芯片，模型参数的规模扩大到了原来的5-10倍，安全的性能表现到现在也有5倍到10倍以上的提升。
同时，推演万物的能力开始涌现。
02 从点亮到上车
作为兄弟公司，XHEART和Momenta独立运作，股权上没有直接隶属关系，但因为大模型定义芯片技术路线，两家公司的很多工作从一开始就深度协同。
2023 年国庆假期，Momenta的模型团队和XHEART的芯片团队没有放假，他们一起围着持续演进的模型架构，一项一项分析计算时间花在哪里，再倒推硬件应该为未来做什么准备。
注意这个时间，那时XHEART几乎还是一个“新生儿”。不难看出，传统芯片行业的工作流被颠覆了，李宗霖和SoC 设计负责人讨论最多的一件事，就是怎么把对大模型的判断真正做进芯片架构里。
时间很快来到2025年，有好事发生的那一天，第一颗芯片回到了公司，几乎所有的研发人员都围了上来。
有人在实验室的白板上记下了这个时刻：15:30，芯片到达。随后，20多个部门都先后在白板上签下了自己负责的工作，完成一个就写一个，标注好时间，精确到分钟。
不到 24 个小时后，第一颗 X7 被点亮了。
这几乎是一个创造性的周期。芯片点亮是一个环节非常多的过程，按行业经验，大型芯片点亮顺利的话要几周或一两个月，不顺利的话，拖上半年也是常有的事。
点亮后的上车测试，进行得也异常顺利——但"顺利"的前提是，团队先给自己加了一道难题。
按照行业通用做法，芯片的上车测试通常会选择在封闭道路上进行。但李宗霖觉得不行，他希望交付的是一颗真的能让真实车辆在道路上稳定可靠工作的芯片。
所以，XHEART决定用更严格、更接近量产的标准来重新定义"回片上车"这个节点。
它要验证的不是芯片单项性能，而是由芯片、底层软件和智驾模型组成的完整系统，能不能在真实的道路上跑出接近量产的表现——细致到每一个路口的转弯，模型规划出来的行驶轨迹都必须与量产标准保持一致。
这个标准，对一颗刚回来的芯片是非常有挑战的。经过团队的日夜攻坚，整个系统从芯片到底层软件再到算法，磨合得越来越稳定。
首台测试车在短短几天后就成功上路；17 天后，X7 整车闭环完成，测试车可以在区域的城区道路上连续稳定行驶数小时，全程无接管；无论是白天夜晚，还是那段时间经常出现的暴雨天气，系统都经受住了考验；第 21 天，搭载 X7 的测试车驶入真实车流，向客户开放体验。
按照行业规律，通用芯片从拿到手到"跑通、跑稳、跑对、跑好"，顺利的话要三到六个月，不顺利要六到十二个月，甚至更久；而 X7 完成这一切，用了不到一个月。
为什么XHEART可以这么快？发布会上，李宗霖解释了原因——
因为量产模型和实车数据提前进入了硅前验证，从芯片的设计，从模型性能的验证，在流片之前都已经做完了。
03 只买单程票的旅程
在上汽集团总裁贾健旭的记忆里，也有一个瞬间，是一辆车汇入了车流。那辆车上搭载了 Momenta 智驾系统，曹旭东陪他一起试驾。
在车企当中，上汽是 Momenta 最坚定的同路人。早在 2021 年开始，上汽集团就多次投资 Momenta，是后者最大的机构投资者。智己、上汽通用、上汽奥迪、上汽大众均与 Momenta 达成了战略合作，在多款量产车型上搭载其智驾系统。
这种深度合作中，上汽投入的不仅是资金和订单，还有信任——而当那辆测试车灵活地避开所有路障、稳稳将贾健旭送达目的地时，这种信任得到了交付。
所以，那一路，贾健旭有些感动。他和曹旭东聊了很多。突然想到这家做软件的公司，有没有做硬件的计划，以实现双轮驱动？
曹颇为神秘地说，已经在路上了。
当时正值X7芯片的研发周期，显然，曹旭东相信事以密成。
今年上半年，作为首款搭载 X7 芯片的量产车型，上汽大众 ID.ERA 9X 也上路了。陆续地，更多量产车汇入真实车流。
到今天为止，X7 系列芯片已经有了 10 款量产车型，获得超过 40 款车型的订单。
这 40 款车型背后，包括上汽在内的所有厂商，都向XHEART这样一家年轻的芯片公司投入了信任成本。和当年上汽选择 Momenta 时一样，XHEART当时也没有可以证明的成功案例，它同样要在路上证明自己。
这是一趟只买单程票的旅程。
在XHEART，这不是一个比喻，而是一个真实的故事：在某款重磅车型上市前，公司一位同事准备去客户现场，出发前 HR 问什么时候回来，他说："我也不知道，就先买单程票吧。"
在这家公司，归期是交付客户信任的那一天。
但客户的信任怎么会有被完全交付的那一天？值得注意的是，汽车厂商与智驾解决方案供应商之间的关系非常紧密，协同的深度也远超一般零部件采购。
正如李宗霖所言，把芯片做出来之后，他们还会进一步提供丰富的解决方案，协助客户降低开发难度，让项目更快地走向量产。
所以，XHEART手上拿的，也是一张单程票。好吧，现在，这是一个比喻。
这个比喻不仅适用于它和客户之间，也适用于它和供应商之间。车规芯片是一个特殊行业，它处于汽车产业链中游，在 3 年这么短的时间内实现量产，不是靠XHEART一家公司就能完成的，它是产业链上下游共同做功的结果。
是的，XHEART还得成为上游的“同路人”。
首先，需要确认的事实是，在 X7 之前，XHEART的产品序列是一张白纸，它的兄弟公司 Momenta 成立时间也不长，并且是一家软件公司。但台积电成为了XHEART的供应商。
众所周知，当下想要拿到台积电的产能是非常难的，创业公司更是难上加难。3 年前，在和曹旭东深聊之后，台积电副总经理陈平决定接下XHEART的单子，尽管当时他觉得对方说的 L4 级别自动驾驶像是科幻片。
陈平坦言，如果换个人来讲，他是会怀疑的，但眼前的曹旭东让他选择了相信，同时李宗霖也加码了这份信任，"在我们行业里，李宗霖一直被公认是最可靠、最靠谱的人"。
3 年后的今天，XHEART量产已经 8 个月，累计出货超过 20 万颗。
发布会前，一位媒体人给曹旭东发来了两段短片：一辆搭载 X7 芯片和 Momenta 智驾系统的车开在复杂车流里，复杂路况下倒车避让，多把掉头，车表现得比老司机还老司机。
这其实就是曹旭东当年和陈平聊的东西。今天，那个几乎无声、快速自主旋转的方向盘，兑现了当初的相信。
在商业合作当中，信任关系不是一种闭环结构，而是会随着一次次兑现loop起来。
例如当智己采用Momenta智驾系统获得令消费者满意的结果后，上汽的更多车型也开始采购Momenta，当上汽大众ID.ERA 9X量产上市后，MG 07、荣威家越07也开始搭载X7芯片。
10月29日，李宗霖还披露了X9芯片的进展，其大模型算力达到3000TOPS,内存带宽达 700GB/s的芯片。
这些参数就足以说明，在下一代模型问世之前，XHEART先做好了准备。这家公司的技术战略是“量产一代、开发一代、预研一代。”
它的供应商也做好了准备。
陈平在发布会上留下伏笔：目前最先进的车规工艺是 3 纳米的 N3AA，而 L3、L4 对算力、功耗和面积的约束，决定了"非用 3 纳米不可"。
台积电的承诺是："我们的工艺已经准备好了，你们把最好的创造力、最好的设计送过来，我们一起为世界创造最美好的东西。"
他们的确都想让好事发生，曹旭东、李宗霖、陈平、贾健旭……，以及每个炼芯术士。
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
"FunTalk," by Wang Fong Jie.
On the first chip return, a colleague of XHEART found that the calendar on his desk stated, "Good things happen today." He quickly filmed it and sent it to the company.
It's been lit in less than 24 hours.
The most striking thing about this whole thing is, of course, the chip, but there's another person who should not be ignored when he sees the good things, first the company, then the soul.
It's a lucky company, in matters and in people.
In only three years of its existence, XHEART has launched the first generation of LLM chips X7 series and has made it possible to scale up and load cars.
On October 29th, this young company held a publish conference and announced the launch of a new LLM chip.
X7 Ultra, a car code chip with 700 TOPSLLM calculus. The X7 Ultra supports a 10-fold upgrading of the deployable model parameter count through a synergistic innovation of architecture and tool chains.
Like the X7 series, X7 Ultra remains a chip defined by LLM, and the adaptation of chips and models is pre-positioned to the front end of the R & D chain.
Moreover, XHEART still insists on making an evaluation based on real-life smart driving LLM. Because, according to XHEART CEO Lee Jong-ho, the industry usually uses basic tests in laboratories to assess the chip performance, but laboratory results are often far from the actual experience of a quantitative model.
The X7 Ultraquantization post-deployment model output was 99.99 per cent similar to the trained FP32 floating point model according to the real-volume model.
The concept of four nines is not new, and it is often seen in the gold sector: 99.99 per cent of pure gold, once refined, is called gold. To a certain extent, the creation of a chip is also like alchemy, carving silicones and then getting the technological civilization that changes the human world.
Behind the X7 Ultra is a group of core magicians, either inside the XHEART or up and down the industrial chain, who believe that the chip, as defined by LLM, can fully release model capabilities and support China ' s intellectual driving to achieve L4-level self-pilot.
"Good things happen today" is their spell.
What kind of chip does the market need in a few years?
In 2023, after OPPO closed down the philosophies of the chip design company, Li Zongzheng of the then Zhu Coo decided to start a business with a core team. He was a very senior expert in the Chinese chip industry, and before Chulku, Li was the Managing Director of the Radio Communication Department of UNDGC, leading the development of the Helio Processor Series and the Skychip Chip Series.
But even after nearly 20 years of work in the chip industry, Li Zongjun can't escape the core. One of the major pains in the sector is that the models are highly iterative, but the research and development cycles of the chips are annual, so can they meet the needs of the models?
At the same time, as the founder of Momenta, Cao Xudong had his own doubts. According to his judgement, 2028 will be the L4-level self-driving flash point, when the model will be very large, requiring a model of at least 10B, with an approximate rate of 100B parameters, and the existing generic chip cannot support it — the power of the vehicle, the reliability of its rules and the cost of it — is a “no triangle”.
If a generic chip stack is imposed, it would be equivalent to putting a large GPU server in the car; the model size would be impossible to reconcile with the cost of effort. What do we do?
By chance, two confused people met together. They talked a few times, and they found that both problems had the same solution.
LLM defines the chip.
Soon, Cao Xudong and Li Zongjun decided to join together to make chips. They need each other.
The Li Zongjun team needs to work with Momenta to get the model involved at the earliest stage of development, first to understand how the model works and then to reverse the hardware architecture and the software tool chain.
Momenta also needs the Li Jong-hoon team, because to support the L4 level, it needs a high-performance chip that supports the size of the model, and that chip has to be loaded and produced, so it has to meet both energy efficiency and cost and break the impossible triangle.
In the same year, XHEART was established, and from the first day of its existence, XHEART established a technical route for “LLM defined chip”: Rewind the chip structure, reorganise computation, storage and data flow from the model needs in real applications. The team worked with the algorithm team from the chip definition stage to study model structures and key algorithms in advance of the evolution of models in the coming years.
The chips that follow this technical route will not be defeated by time.
In the case of X7 Ultra, the supernuclear NPU interlocks the model's data from the 100 GB per second of the traditional structure and aggregates it into a piece of TB per second;
For key algorithms such as Linear, Softmax and Transpose in the Transformer model, X7 Ultra was specially designed for hardware acceleration engines, with primary support for high-precision vector calculations;
The calculation of X7 Ultra reached 98.7 per cent through a bottom-down soft and hard-synergy;
Re-engineered through the AI tool chain, memory bandwidth consumption decreased by 80 per cent compared to industry flagship models, and energy efficiency increased by 100 per cent.
When a chip of this kind is produced, it will not only support the existing model, but will also reverse the iterative model.
According to Cho Xudong, the forthcoming R7 Ultra World Model of Momenta, based on the X7 Ultra chip, expanded the size of the model parameters to 5-10 times, and the safety performance is now five to 10 times higher.
At the same time, the ability to develop everything began to emerge.
02, from the light to the car.
As a brother company, XHEART and Momenta operate independently and have no direct reporting interest, but as LLM defines the chip technology route, much of the work of the two companies has been coordinated from the very beginning.
On the national holiday of 2023, Momenta ' s model team and XHEART ' s chip team did not take leave, and together they surrounded the evolving model structure, one analysis of where time was spent and then what should be prepared for the future.
Look at this time, when XHEART was almost a “newborn child”. It's not hard to see how the workflow in the traditional chip industry has been subverted, and one of the things that Li Jong-soon and SoC designers have discussed most is how to really put the judgement on LLM into the chip architecture.
Soon after 2025, the day when the good thing happened, the first chip returned to the company, and almost all the researchers and developers were surrounded.
The moment was recorded on the whiteboard of the lab: 15:30, the chip arrived. Subsequently, more than 20 departments signed their own work on the whiteboard, and completed one, marking the time and the minutes.
Less than 24 hours later the first X7 was lighted.
It's almost a creative cycle. Chip-lighting is a very many-step process, and in industry experience, it takes weeks or months or months for a large chip to light well, and if it does not work, it is common for the first half of the year.
It's a very good time to get in the car after the light. But "success" presupposes that the team adds a problem to itself.
In line with industry-wide practice, the testing of chips on board is usually carried out on closed roads. But Li Zongjun felt no good, and he wanted to deliver a chip that would really allow real vehicles to stabilize their work on the road.
So XHEART decided to redefine the node of "back-to-back" with more stringent, closer to the volume of production.
It's not a chip's individual performance, it's a complete system of chips, bottom software and smart driving models that can run close to mass production on a real road -- – To the point of turn of every intersection, the trajectories planned by the model must be consistent with the production standards.
This standard is very challenging for a new chip. Through the team's day and night effort, the whole system has become more and more stable, from chips to bottom software to algorithms.
The first test vehicle was successfully launched in just a few days; 17 days later, the X7 vehicle was closed and closed, and the test vehicle was able to travel steadily on the region's urban roads for several hours without taking over; the system was tested, both during the day and night, and during that time, in heavy rain; and on 21 days, the X7 test vehicle entered the real traffic and was open to clients.
By industry, it takes three to six months, six to twelve months, or even longer, for a universal chip to get its hands on "run, run, run, run, run well," and it takes less than a month for X7 to do all this.
Why is XHEART so fast? At the publish meeting, Li Jong-soon explained why.
As mass production models and real-car data are pre-insilated, the design of the chip, and the validation of model performance, have been completed before the stream.
03, one-way trip.
In the memory of Ja Kenwu, the president of the upper steam conglomerate, there was also a moment when a car remitted into the traffic. The car was loaded with Momenta's smart driving system, and Cao Xudong was with him.
Among the cars, the ups and downs are the most determined fellow of Momenta. As early as 2021, the Upper Gas Group invested several times in Momenta, the largest institutional investor in the latter. In addition, Chile, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America, the United States of America and the United States of America.
In this in-depth cooperation, not only funds and orders, but also trust — a trust that was delivered when the test vehicle was able to avoid all roadblocks with flexibility and to secure the delivery of Jayan/Jan/Wu to his destination.
So, that's the way Ja Ken Wing was touched. He talked to Cao Xudong a lot. Did this software company have any plans to do hardware to achieve a two-wheel drive?
Cao is mysteriously on his way.
It was during the research and development cycle of the X7 chip that it was clear that Cao Xudong believed in secrecy.
In the first half of this year, as the first mass-producing model to carry the X7 chip, the crowd ID.ERA 9X is on its way. Over time, more large-volume cars are being sent into real traffic.
As of today, the X7 series of chips already have 10 model cars and orders in excess of 40 models.
All the manufacturers behind the 40 cars, including the gas, invested in trust costs for a young chip company like XHEART. As was the case with Momenta, when it was the last time it was a steam option, XHEART had no proven success, and it had to prove itself on the road.
This is a one-way trip.
In XHEART, this is not a metaphor, but a true story: before a heavy-pound car was listed, a colleague from the company was going to go to the client's site, and before leaving HR asked when he would come back, he said, "I don't know, buy a one-way ticket."
In this company, the deadline is the day of delivery of the customer ' s trust.
But how can the trust of the client be fully delivered? It is worth noting that there is a strong relationship between automobile manufacturers and suppliers of smart-driving solutions, and that the level of synergy is much greater than the general procurement of spare parts.
As Li Zongzhen said, once the chips are made, they will provide further rich solutions to help clients reduce development difficulties and move the project more quickly towards production.
So, XHEART's got a one-way ticket. Well, now, that's a metaphor.
This analogy applies not only between it and the customer, but also between it and the supplier. Carchips are a special industry that is in the middle of the car industry chain and that achieves a volume of production in such a short time of three years, not by a company in XHEART, but by working together up and down the chain.
Yes, XHEART has to be a “sympathetic” upstream.
First of all, it needs to be established that before X7 the product sequence of XHEART was a piece of paper, and that its brother company, Momenta, was also established for a short time and was a software company. But it became a supplier of XHEART.
It is well known that it is very difficult to get the capacity to build electricity at the moment, and even more so for start-ups. Three years ago, after a deep conversation with Cao Xudong, Chen Ping, Deputy Director-General of Power, decided to take the XHEART list, although at the time he thought that the L4 level autopilot was like a science fiction film.
Chen Ping said he'd be suspicious if he were someone else, but he chose to believe, and Lee added that trust, "It has always been recognized in our profession that Lee is the most reliable and reliable person."
Three years later, the XHEART production has been in place for eight months, with more than 200,000 cumulatively delivered.
Before the publish meeting, a mediaman sent two clips to Cao Xudong: a car carrying X7 chips and Momenta ' s smart driving system was in a complex traffic, with complex road conditions turning back, turning back, and it was more of a driver than an old driver.
That's what Cao Xudong was talking about with Chen Ping. Today, that almost silent, rapidly autonomous wheel of rotation fulfils its original belief.
In the context of commercial cooperation, trust relationships are not a closed-ring structure, but are expected to come together as a loop.
For example, after the introduction of the Momenta Smart Driving system to achieve consumer satisfaction, more car models of uplifting have started to purchase Momenta, and when the public ID.ERA 9X volume was on the market, the MG 07 and the Vengwei family also started carrying X7 chips.
On 29 October, Li also disclosed the progress of the X9 chip, with its LLM calculator up to 3,000 TOPS and its memory bandwidth of 700 GB/s.
These parameters are sufficient to show that XHEART was prepared before the next generation model. The company's technical strategy is “to produce, develop and prepare a generation”.
Its suppliers are also ready.
Chen Ping left a pen at the publish meeting: the state-of-the-art car procedure is N3AA for 3 nanometres, while L3 and L4 have determined "no use of 3 nanometers" for the power, power and area constraints.
The promise of power generation is, "Our process is ready, you bring the best of creativity, the best of design, and together we create the best of things for the world."
Indeed they all want good things to happen: Cao Xudong, Li Zongzheng, Chen Ping, Jia Ken Wing... and every charlatan.
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.