Yangyu Chen (@yangyuc) header
Yangyu Chen (@yangyuc) profile picture

Ph.D. Student. Open-source enthusiast. Computer Architecture and software co-design. Intern at $INTC. Opinions are my own. GH: https://teapaw.org/t.co/QtjKsKmkNE

Joined May 2014
Yangyu Chen retweeted
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA. I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation. Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO. Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours. Jensen
61,308
121,768
24,520
1,175,116
59,236,131
It's the first time Astra leaned weak accept to my paper
1
5
335
Yangyu Chen retweeted
I have a particularly nasty case of cancer. When I was hospitalized again at my lowest point last year, I told Jensen. He immediately reached out to contacts at Stanford, who were able to cut through red tape, and get me into a clinical trial for which I've turned out to be a perfect fit. I think everyone can have their own views about recent events. But any claims that Jensen did not really care about his staff is simply inaccurate. I, personally, owe him more than I can express. I think the best tribute we can pay to Jensen's legacy is to drive NVIDIA's continued success and push the whole industry forward, by showing we've learned the lessons from years of working here. We will compare ourselves against Speed of Light. The mission will be the boss. And we will not be scared to take smart, bold risks and fail fast with ruthless intellectual honesty. But if there is one management story I remember the most, it's a particular Jensen grilling from my time in autonomous vehicles that stayed with me. Luckily I was not in the hot seat for this one. Team A didn't want to use Team B's framework, because they thought it was too slow. Jensen was surprised - it's just applying a bunch of optimized CUDA kernels in sequence, how could it be slow? So he asked the executive from team B to draw the entire data flow sequence from their framework on the whiteboard and highlight every memcpy. Sure enough, there were a bunch of unnecessary memcpy's surrounding each pipeline stage. Everyone in the room could see why it wasn't getting to SoL and a path to improve things. No arguing about fiefdoms or appeals to authority - the data got us all on the same page. The lesson was not (just) that unnecessary memory copying is bad. It's that executives should not be afraid to probe into details if something doesn't add up, and leaders need to be technical enough and honest enough to engage in real time and not turn every meeting into "I'll get back to you in a week." If that means counting memcpy's with a billionaire CEO, so be it. My best executive meetings at NVIDIA have been when I'm fully prepared to present every detail of my team's work, and my worst failures have come when I realized I couldn't go deep enough to really solve the problem. But the thing that gives me confidence is that Jensen was never the only one to live this philosophy. Anyone who has presented to an SVP at NVIDIA knows they are at risk for getting grilled on fundamental constraints and technical details. I have been part of amazing conversations with both recent and old timer leaders where we were able to get to the core of a problem in an hour rather than a month, because we had the right people in the room and were not afraid to go deeper. The company is filled with supremely talented people who have been trained for 3 decades to work at the highest level, and you can see it in what we produce every day. I'm so proud to be an NVIDIAn, and I can't wait to show off what we ship in the coming months and years.
417
1,983
273
21,565
2,601,290
Sad to hear that. While AMD is catching up with the CUDA ecosystem since ROCm 7. And the ecosystem becomes much cheaper in the agentic era.
This quoted post is unavailable.
3
522
Right after the release of GestaltX [1], which can be used to unlock some iPhone features from regulation limited by the purchased area. I have an option to resolve the travel abroad issue with an eSIM-only iPhone in China. I bought an iPhone Air (eSIM only) and will give it to my mom to replace her old iPhone 14. The main concern with the pure eSIM iPhone in China before was that installing an eSIM from any Chinese wireless carrier needs your eUICC to have a China CI instead of a GSMA CI, which only appears in the China Model of iPhone (it also supports GSMA CI). But the regulation limits Chinese iPhones to have only 2 eSIMs installed. Thus, if you need 2 phone numbers and want to travel abroad without roaming, you will need to delete 1 eSIM. And the Chinese wireless carrier needs you to go to their shop with your ID and iPhone to get the eSIM QR Code. They need to scan your Network Access Label, EID, and IMEI, and the Network Access Label will not be shown on an iPhone purchased anywhere other than China. But the 2 eSIM restriction is only during installation, without restrictions in further use. Why I finally purchased an iPhone Air for my mom is that: I found that some travel eSIM service providers like CMLink provides option to add a new plan on existing eSIM ICCID with affordable price. Thus, even if this bug is resolved in the future and Chinese iPhone models won't get this crack after a software update, we can still use such an already installed eSIM to get travel plans. And even with this crack enabled, I can still add a China Mobile eSIM. I tried that in person today. I think it will be an option if you are Chinese and travel abroad frequently. Just install eSIMs from all such travel carriers now and pick one to add a travel plan when you need to use it in the future. [1] github.com/huamidev/GestaltX…
5
3
25
6,327
Something interesting to share: I can query the EID and IMEI from the China Mobile Carrier App. This iPhone has two IMEI, I can use either one to get this result. But at the China Mobile store, I’m sure they only scan IMEI1, and they told me scanning IMEI1 is enough. When switching between eSIMs, that China Mobile eSIM will always use IMEI2, just as if something is already bound to one IMEI.
3
610
It’s crazy one CCX spans 32c!
At this time, AMD has not provided any information regarding a reduction in the chip size of the Zen 6c core compared to the Zen 6 core. However, in a conversation with Mike Clark, he told me that the focus was less on chip area and more on performance.
1
370
Sometimes I really want to ask some PhDs from various fields: do you find that current AI knowledge is actually helpful for conducting rigorous research? In my experience with the CPU microarchitecture field, the knowledge of Opus 5.5 is completely wrong most of the time without context and a web search. I nearly lost my mind with frustration when I tried to have it help me improve my research paper.
2
4
497
I have tried Gemini 4 Argon, and the text improvement task in my research paper is MUCH MUCH better than Opus 5.5 and GPT-6 Astra!
1
77
I can’t imagine how much "slop" we’ll see this year at conferences like ISCA in papers relying solely on simulators. Because the quality of a simulator implementation really comes down to the developer’s personal integrity and their awareness of the design space limitations.
1
250
所以我现在审稿更看重是不是一个新的需求 + 扎实的分析,最后(大概率)由 AI 进行设计空间探索带来的创新,已有需求的排列组合能力已经是 AI 完全能够替代的了。
深有同感。读博这几年,我熬夜看过的几百篇顶会,所谓的“全新架构”,十有八九就是把 Transformer 的某个模块替换一下,套在一个新数据集上。这就是极其标准的“学术线性插值”。 以前大家还能靠这种“微创新+高质量排版”混毕业、拿教职。但现在,你把这几篇基操论文丢给 Claude,让它去找结合点、设计消融实验,它一秒钟能给你生成十个比你还严谨的 idea。 在拼“缝合能力”这件事上,碳基生物已经彻底干不过硅基了。 以后学术界真得洗牌了。如果博士的能力还停留在做“资料检索+排列组合”,那毕业即失业。 真正能活下来的,是那些能提出真问题、能搭建底层新范式,或者能去一线搞到真实脏数据的人。别再把生命浪费在写那些大模型一秒钟就能生成的八股文上了
3
409
I genuinely hope my AI subscription allows unused tokens from last week to carry forward, much like certain telecom providers shift unused cellular data into the next month.
1
228
Shortly before the ISCA submission deadline, I decided to add some FPGA-based experiments. I revisited a project I had shelved a few months back: qemu-system-fpga, which connects the I/O of an FPGA-based soft-core directly to QEMU. I also implemented DMA support for virtio, making the process of booting the FPGA soft-core as straightforward as launching a QEMU instance. Performance testing shows impressive I/O efficiency: running a 160MHz Rocket-Chip on a Xilinx VU37P, I managed to hit ~100 Mbps with iperf3. In comparison, 9pfs is a bit slower (Maybe NFS should be the right choice). Incredibly, Opus 5.5 did this in a day! github.com/cyyself/qemu-syst…
6
351
Now I think LLM can be compilers for everything, even their weights.
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
1
2
222
但我觉得存在一个现实是,很多很多论文的作者(哪怕是在每个领域最顶级的几个会议上)发自内心都不觉得自己的研究有现实意义,但是为了毕业拿学位却不得不这么做,这种事情也加剧了上述情况的出现。少数真正有意义的研究通常大家还是十分 respect 的。
科研圈最可怕的异化,可能不是博士生失去了科研能力,而是逐渐失去了从科学发现本身获得快乐的能力。 一些博士生已经被学术评价体系训练出了一套非常扭曲的奖励机制。 看到一篇有意思的论文,第一反应不是好奇它发现了什么,而是这篇工作凭什么能中顶会。 看到同门发了 Paper,第一反应不是祝贺,而是焦虑自己为什么还没有。 看到竞争对手被 Reject,甚至会产生一种莫名的优越感。 科研本来应该是一场不断扩大人类知识边界的正和游戏,却逐渐被一些人活成了比较排名、争夺 Credit 和学术地位的零和博弈。 这里有一个更深刻的问题: 当一个人长期无法从研究本身获得正反馈,他就可能开始从别人的失败中寻找相对优势。 因为科学发现太慢,论文发表太不确定,而比较、嘲讽和否定别人,却能提供即时的心理奖励。 这不是科学家的天性,而可能是长期处于高度竞争、资源稀缺和单一评价体系下形成的适应性行为。 更讽刺的是: 一个以培养独立思考和探索未知为目标的制度,可能最终培养出了一批最害怕失败、最依赖外部评价的人。 当然,严厉的学术批评是科学进步的必要条件。但批评研究中的错误,与通过贬低研究者获得优越感,完全是两回事。 我越来越觉得,衡量一个 PhD Program 的质量,不应该只看它一年发表了多少顶会。 还应该看,几年博士训练之后,学生是否依然拥有好奇心、创造欲,以及对科学本身的热爱。 科研最大的失败,不是几年没有发出顶会,而是几年之后,你已经不知道除了发表顶会,还有什么事情值得让自己兴奋。 我们原本是因为想理解世界才走进科研。 不要最后只学会了如何在科研圈里赢过别人。
1
7
1,170