Specifically for real-time, low latency applications. Things like autonomous driving, local weather prediction and the like. For the gamers, offloading graphics processing and simply rendering the output on the screen at home.
Miscellaneous News
- Thread starter bd popeye
- Start date
Massively smart play for US to try and assassinate Liang, because just imagine who China kills in return? Sam Altman? Zuckerberg? The bitcoin guy? Vivek (lmao)?
Losing those would boost rather than damage US tech industry.
Imagine an army led by generals that are so bad that fragging them actually improves their combat capability: this army doesn't need to fear decapitation strikes, ever.
plawolf
Lieutenant General
It would be negligent for Liang and all his team to not be protected by the MSS, and the MSS doesn’t tolerate negligence.
Trying to kill him will be no easy feat. Trying to kill him and get away with it will be far harder.
Can the CIA kill him? Probably if they tried really really hard. But are the prepared to lose potentially multiple operatives and risk starting a war to do it?
Trying to kill him will be no easy feat. Trying to kill him and get away with it will be far harder.
Can the CIA kill him? Probably if they tried really really hard. But are the prepared to lose potentially multiple operatives and risk starting a war to do it?
I also saw pics of their office building. Can't find the link right now but it's literally just an office. And people are trying to visit.
I hope they up the security. Not even a joke.
I hope they up the security. Not even a joke.
He used to tell the story that his parents were persecuted by the Castro regime to get sympathy votes from Cuban community. I guess that lying truly is a key skill to be Secretary of State. (Pempeo: what did i say?)
That is a tricky one. So let me break it down first.
Latency and throughput are two "evils" in software engineering that depends on compute, network and storage.
Real-time is widely being abused nowadays. There is a so-called real-time distribution of Linux, which is a joke to me, because real-time and Linux kernel are oxymoron. Real-time requirements are often divided into hard real-time and near real-time. The defining characteristic of real-time software is deterministic requirement between BSP and kernel.
So to answer your question, which implies some computer-as-a-service for real-time low-latency applications, the short answer is not now. For example, AWS offers Local Zone for low-latency use cases, but not even close to be feasible to real-time low-latency applications that I did in the past.
That comes down to applications such as autonomous driving, in the context of real-time low-latency, DeepSeek, software architecture. IIRC, Huawei has deployed a multi-tier architecture for its ADS services. Huawei knows carrier-grade systems, solutions and software better than any other NEV vendor in China. And in the world of software engineering, carrier-grade system architecture is inherent hard real-time. So naturally, I guess, Huawei knows the technical barriers in implementing true L3/L4 autonomous driving solutions, which requires some form of backend in the cloud, public or private. I think ADS has already implemented its "AI agent" in each HIMA cars that are connected to its backend running its Pan Gu (盘古) LLM. It is this technological integration that I said in the past that what Huawei is trying to implement for autonomous driving is extremely difficult to accomplish.
I am interesting in wait and see what kind of real-time low-latency public services that can successfully integrate with 5G and AI. I guess ADS might likely be the first of its kind.
I saw pics of the smallish office building and now as their fame has spread-Chinese people taking selfies outside building etc(harmless of itself)-but since Liang has met and talked withVP Li Qiang himself he would be noticed and assigned MSS security as would his team.The reason I am so worried is that the wonderful J36/J50 076 etc are in the realm of the hypothetical war scenario-BUT this DEEPSEEK hit was real ,now and palpable-a gut punch in real-time right where it hurts to the USA/westoids.IIRC in the late 1950's -the Americans tried to sabotage/assassinate scientists/engineers in Lanzhou to stop Chinese nuclear development-this is just as big-on guard
From my understanding, AGI will have a much greater impact than basically any other technological advance. And this time, it's actually something that cannot be lost. Even a 2 years gap in reaching AGI is devastation. And this is clearly not the case for other things such as 6th gen.
In before US bans the laws of physics from operating in China. Add breathing air too.
We have already reached this point.
GALLUP: American public's
This explains the exodus to XHS very well, despite the rampant propaganda.
Hear me out, because I also had the same concern, why give it up with no return on investment and relinquish a tech advantage?
Losing Money (actually, they made a lot!)
DeepSeek is a hedge fund that actually made more money shorting Nvidia's $600 billion drop (biggest one-day drop in US history) than the entire profit of ClosedAI of one single human lifetime (i.e., ClosedAI has never reported profit, and likely never will after now open sourcing models) DeepSeek is primarily a hedge fund (quant trading company) that made more money from shorting Nvidia than it could possibly fathom from DeepSeek as a standalone closed private product.
Losing Tech Advantage (actually, it's no "secret sauce")
With no disrespect to DeepSeek's innovation, but the optimization technique they shared was not surpsing and well-known in published research few years before. For example, compression of key indices to reduce RAM, 8 bit floating point instead of 32 bit floating point to reduce RAM, multi-token prediction instead of single-token prediction to reduce RAM, mixture-of-experts to reduce compute requirements, all of these are known techniques, they allegedly applied it on top of distillation of o1/4o's output. In no way is this "moat" enduring in perpetuity, and likely will be independently discovered by others in due time. So they made it widely known, to take advantage of shorting Nvidia.
Did I mention that it completely undermines ClosedAI's business model and the whole technology sanction regime? They shorted the entire AI industry and make BANK. That's the point.
It seems that that FT girl has not been able to send her DS intelligence to MI6. Someone should inform the MPs that Hangzhou is neither cold nor abundant in energy. Scotland should stick to what it is good at, making Scotches.