正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

$INFQ GPU + CPU + QPU Frontier models are hitting saturation on tokens (data volume) and context windows (memory). $INFQ CTO walks through how they and $NVDA are working together to add a Quantum Processing Unit to the chain as the next inflection point.

YeahDave

32,111 subscribers

57,181 次观看 • 3 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

If you invest in quantum, you invest in the US national security. $INFQ CEO Matt Kinsella says the GPS will be the first thing to go in case of a full scale conflict. The company’s sensors can replace GPS with precise timing. $INFQ sells to NASA and the US Department of War. It looks like $INFQ could be selected for quantum sensors as explained in the Executive Order 14413 on Quantum Innovation last week.

If you invest in quantum, you invest in the US national security. $INFQ CEO Matt Kinsella says the GPS will be the first thing to go in case of a full scale conflict. The company’s sensors can replace GPS with precise timing. $INFQ sells to NASA and the US Department of War. It looks like $INFQ could be selected for quantum sensors as explained in the Executive Order 14413 on Quantum Innovation last week.

Investor Denis

21,943 次观看 • 1 个月前

"Quantum is here, and it's time to make sure we are capitalized to address the opportunity that's ahead of us," Infleqtion CEO Matt Kinsella says on going public on the New York Stock Exchange under the ticker $INFQ.

"Quantum is here, and it's time to make sure we are capitalized to address the opportunity that's ahead of us," Infleqtion CEO Matt Kinsella says on going public on the New York Stock Exchange under the ticker $INFQ.

Yahoo Finance

25,920 次观看 • 5 个月前

In 2025, I told you to buy quantum and we made millions together. In 2026, I told you to buy quantum AGAIN. This time, you'll retire within the next 4 years. Here are the BEST quantum stocks to buy right now: 1. IonQ $IONQ 2. International Business Machines $IBM 3. Infleqtion $INFQ 4. Rigetti Computing $RGTI 5. D-Wave Quantum $QBTS 6. Xanadu $XNDU 7. Quantum Computing $QUBT 8. Arqit Quantum $ARQQ 9. Horizon Quantum $HQ All my buy and sell signals in Discord @

In 2025, I told you to buy quantum and we made millions together. In 2026, I told you to buy quantum AGAIN. This time, you'll retire within the next 4 years. Here are the BEST quantum stocks to buy right now: 1. IonQ $IONQ 2. International Business Machines $IBM 3. Infleqtion $INFQ 4. Rigetti Computing $RGTI 5. D-Wave Quantum $QBTS 6. Xanadu $XNDU 7. Quantum Computing $QUBT 8. Arqit Quantum $ARQQ 9. Horizon Quantum $HQ All my buy and sell signals in Discord @

StockWhale

152,401 次观看 • 2 个月前

🎞️ Curious to learn more about Match Chain? Take a look at this helpful explainer that walks us through how Match ID and Match Hub are working together to power the internet’s evolution from web2 to web3. Read more 👉

🎞️ Curious to learn more about Match Chain? Take a look at this helpful explainer that walks us through how Match ID and Match Hub are working together to power the internet’s evolution from web2 to web3. Read more 👉

Matchain

254,695 次观看 • 2 年前

$NVDA is entering the consumer PC chip market with RTX Spark, an $ARM based CPU/GPU/AI chip for Windows laptops and mini-PCs. The flagship version includes 20 CPU cores, 6,144 GPU cores, and up to 128GB of unified LPDDR5X memory, aimed at local AI agents, creators, developers, and gaming. Nvidia says 30+ laptops and 10+ desktops are already in development with partners including Microsoft, Dell, HP, Asus, Lenovo, MSI, Acer, and Gigabyte.

$NVDA is entering the consumer PC chip market with RTX Spark, an $ARM based CPU/GPU/AI chip for Windows laptops and mini-PCs. The flagship version includes 20 CPU cores, 6,144 GPU cores, and up to 128GB of unified LPDDR5X memory, aimed at local AI agents, creators, developers, and gaming. Nvidia says 30+ laptops and 10+ desktops are already in development with partners including Microsoft, Dell, HP, Asus, Lenovo, MSI, Acer, and Gigabyte.

Wall St Engine

109,479 次观看 • 2 个月前

The models we use every day are brilliant strangers. They forget your organization the moment a chat ends, then relearn it on the next query. Engram fixes that. It learns your world once and reuses that memory, matching frontier systems on 1-10% of the tokens. Microsoft, Notion, and Harvey are already testing it within their organizations. Congratulations to the team, and hear directly from Dan Biderman (CEO and co-founder) and Sabri Eyuboglu (CTO and co-founder) with "Leigh Marie" Braswell ⬇️

The models we use every day are brilliant strangers. They forget your organization the moment a chat ends, then relearn it on the next query. Engram fixes that. It learns your world once and reuses that memory, matching frontier systems on 1-10% of the tokens. Microsoft, Notion, and Harvey are already testing it within their organizations. Congratulations to the team, and hear directly from Dan Biderman (CEO and co-founder) and Sabri Eyuboglu (CTO and co-founder) with "Leigh Marie" Braswell ⬇️

Kleiner Perkins

37,765 次观看 • 1 个月前

I had to test it myself to believe this unreal inference speed. 3,000 tokens/s for 1 user on standard datacenter GPUs. They leveraged a hidden efficiency gap in how GPUs generate tokens. Kog just achieved 3,000 tokens/s on 8× AMD MI300X GPUs and 2,100 on 8× NVIDIA H200 (FP16, no speculative decoding). Their tech preview is on a 2B model, and they show how their techniques will scale to large frontier MoE models at similar speeds. That's a huge number because normal low-batch GPU decoding for 2B to 8B models is usually closer to 100 to 300 tokens/s per request, so Kog is claiming something like a 10X to 30X jump in the speed one user actually feels. Their trick: they are getting the speed by treating LLM decoding as a memory streaming problem, not mainly a math problem. For 1 user at batch size 1, the GPU is not doing big, efficient matrix-matrix work like in training or large-batch serving; it is repeatedly pulling the model’s active weights from high-bandwidth memory for each new token, so speed depends on how smoothly those weights keep flowing. Normal inference stacks keep breaking that flow. They run many separate GPU programs for different parts of the model, move intermediate results through memory, wait at synchronization points, talk back to the CPU for scheduling or sampling, and then repeat this token after token. Kog’s answer is to co-design 3 things that are usually tuned separately: the runtime, the low-level GPU code, and the model architecture. The biggest engineering move is the monokernel, where the whole decode pass runs as 1 persistent GPU-resident program, including sampling, so the system does not keep stopping for kernel launches, CPU scheduling, and intermediate memory round trips. They also rebuilt synchronization, because their own measurements say grid sync was eating around 35% of token-generation time; instead of making every compute unit wait at a broad barrier, each unit waits only for the exact data it needs. On AMD MI300X, they also map memory access around the chiplet layout, because memory latency changes depending on which die makes the request. Then their Laneformer model uses Delayed Tensor Parallelism, which lets cross-GPU communication happen in the background instead of blocking every layer.

I had to test it myself to believe this unreal inference speed. 3,000 tokens/s for 1 user on standard datacenter GPUs. They leveraged a hidden efficiency gap in how GPUs generate tokens. Kog just achieved 3,000 tokens/s on 8× AMD MI300X GPUs and 2,100 on 8× NVIDIA H200 (FP16, no speculative decoding). Their tech preview is on a 2B model, and they show how their techniques will scale to large frontier MoE models at similar speeds. That's a huge number because normal low-batch GPU decoding for 2B to 8B models is usually closer to 100 to 300 tokens/s per request, so Kog is claiming something like a 10X to 30X jump in the speed one user actually feels. Their trick: they are getting the speed by treating LLM decoding as a memory streaming problem, not mainly a math problem. For 1 user at batch size 1, the GPU is not doing big, efficient matrix-matrix work like in training or large-batch serving; it is repeatedly pulling the model’s active weights from high-bandwidth memory for each new token, so speed depends on how smoothly those weights keep flowing. Normal inference stacks keep breaking that flow. They run many separate GPU programs for different parts of the model, move intermediate results through memory, wait at synchronization points, talk back to the CPU for scheduling or sampling, and then repeat this token after token. Kog’s answer is to co-design 3 things that are usually tuned separately: the runtime, the low-level GPU code, and the model architecture. The biggest engineering move is the monokernel, where the whole decode pass runs as 1 persistent GPU-resident program, including sampling, so the system does not keep stopping for kernel launches, CPU scheduling, and intermediate memory round trips. They also rebuilt synchronization, because their own measurements say grid sync was eating around 35% of token-generation time; instead of making every compute unit wait at a broad barrier, each unit waits only for the exact data it needs. On AMD MI300X, they also map memory access around the chiplet layout, because memory latency changes depending on which die makes the request. Then their Laneformer model uses Delayed Tensor Parallelism, which lets cross-GPU communication happen in the background instead of blocking every layer.

Rohan Paul

13,244 次观看 • 2 个月前

Physicists are still working on how to make general relativity and quantum mechanics consistent. But there's one problem where they are confident they've made the two theories work together: how many bits of information can you store in a black hole?

Physicists are still working on how to make general relativity and quantum mechanics consistent. But there's one problem where they are confident they've made the two theories work together: how many bits of information can you store in a black hole?

Dwarkesh Patel

29,959 次观看 • 2 个月前

Devs are using Memory Bank as a project architecture and planning tool before they write any code. "I prefer starting in Cline, using Memory Bank to build out the context files as a roadmap, and letting Cline take the wheel." How to use Memory Bank for project planning 🧵

Devs are using Memory Bank as a project architecture and planning tool before they write any code. "I prefer starting in Cline, using Memory Bank to build out the context files as a roadmap, and letting Cline take the wheel." How to use Memory Bank for project planning 🧵

Cline

96,628 次观看 • 1 年前

Palantir CTO Shyam Sankar says Chinese open source models pose a "threat to American Prosperity": "These Chinese open source models are really the result of distillation attacks." "Most of this is just stolen American IP from frontier labs, and I think the frontier labs should do more to protect that IP, and that's in their own economic interest." "The gravest threat to us is not homicide, it's suicide." "It's an inability to turn these tokens into economic value and generate American prosperity for it." "You see that with the proposed moratorium in New York on data centers. Turning our back on AI would be as consequential a mistake as turning our back on the atom in the '70s." "It's just the beginning, but we're starting to see a burgeoning US open model ecosystem forming." "NVIDIA's Nemotron models are very good." "I think you're going to see in the next month a slate of announcements of other American companies, non-frontier labs, neo labs, who are putting out open weight models that companies are going to be able to cheaply fine-tune and capture their own alpha in weights they control." Via Bloomberg

Palantir CTO Shyam Sankar says Chinese open source models pose a "threat to American Prosperity": "These Chinese open source models are really the result of distillation attacks." "Most of this is just stolen American IP from frontier labs, and I think the frontier labs should do more to protect that IP, and that's in their own economic interest." "The gravest threat to us is not homicide, it's suicide." "It's an inability to turn these tokens into economic value and generate American prosperity for it." "You see that with the proposed moratorium in New York on data centers. Turning our back on AI would be as consequential a mistake as turning our back on the atom in the '70s." "It's just the beginning, but we're starting to see a burgeoning US open model ecosystem forming." "NVIDIA's Nemotron models are very good." "I think you're going to see in the next month a slate of announcements of other American companies, non-frontier labs, neo labs, who are putting out open weight models that companies are going to be able to cheaply fine-tune and capture their own alpha in weights they control." Via Bloomberg

Jawwwn

180,421 次观看 • 25 天前

Most C Programmers don't have a clue as to how the CPU works. Here is a lecture from my course on the C Language that introduces how to visualise the CPU as a C (or general software) Programmer.... More Details are at: Being able to model the behaviour of CPU provides a solid base to reason about your code. The summary is - - The CPU is a state machine which maintains it's states in registers. - It has busses to memory for requesting data/instruction read/writes. - Interrupts to attend to events. I discuss this and more in my course on the C Language.

Most C Programmers don't have a clue as to how the CPU works. Here is a lecture from my course on the C Language that introduces how to visualise the CPU as a C (or general software) Programmer.... More Details are at: Being able to model the behaviour of CPU provides a solid base to reason about your code. The summary is - - The CPU is a state machine which maintains it's states in registers. - It has busses to memory for requesting data/instruction read/writes. - Interrupts to attend to events. I discuss this and more in my course on the C Language.

Piyush Itankar

28,019 次观看 • 1 年前

Interesting that in three of my recent conversations with Krishna at Anthropic, Dylan Patel, and Gavin Baker, each said that frontier tokens are capturing the majority of the economic value. Gavin: "An overwhelming amount of the economic returns to AI at the model layer have been at the frontier. That's surprising to me, and I think it's been surprising to a lot of people. This is one of the most important questions to be answered, and you need to have a hypothesis on it as an investor." Krishna: "We think the returns to frontier intelligence are extremely high. Customers invest really heavily in more tokens with the newer models. The ones at the frontier clearly are capturing this economic value, driving meaningful ROI for customers. The returns to frontier intelligence are not slowing down." Dylan: "No one gives a crap about GPT-4 class models. They want the frontier because the frontier lets them create the economically valuable things."

Interesting that in three of my recent conversations with Krishna at Anthropic, Dylan Patel, and Gavin Baker, each said that frontier tokens are capturing the majority of the economic value. Gavin: "An overwhelming amount of the economic returns to AI at the model layer have been at the frontier. That's surprising to me, and I think it's been surprising to a lot of people. This is one of the most important questions to be answered, and you need to have a hypothesis on it as an investor." Krishna: "We think the returns to frontier intelligence are extremely high. Customers invest really heavily in more tokens with the newer models. The ones at the frontier clearly are capturing this economic value, driving meaningful ROI for customers. The returns to frontier intelligence are not slowing down." Dylan: "No one gives a crap about GPT-4 class models. They want the frontier because the frontier lets them create the economically valuable things."

Patrick OShaughnessy

121,110 次观看 • 2 个月前

PowerInfer - a high-speed inference engine for deploying LLMs locally. Just came across this super interesting project on speeding up inference. It's not MoE but it's a simple approach that exploits the high locality in LLM inference to design a GPU-CPU hybrid inference engine. Hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons (the majority) are computed on the CPU. This approach significantly reduces GPU memory demands and CPU-GPU data transfer. It achieves an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs on a single NVIDIA RTX 4090 GPU. It's on only 18% lower than that achieved by a top-tier server-grade A100 GPU. It also significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy. There is a lot more innovation around inference that's coming fast. Really encouraged by the study on sparse computation to enhance the computational efficiency of LLMs. It's now possible to use PowerInfer with Llama 2 and Faclon 40B. Mistral-7B support is coming soon!

PowerInfer - a high-speed inference engine for deploying LLMs locally. Just came across this super interesting project on speeding up inference. It's not MoE but it's a simple approach that exploits the high locality in LLM inference to design a GPU-CPU hybrid inference engine. Hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons (the majority) are computed on the CPU. This approach significantly reduces GPU memory demands and CPU-GPU data transfer. It achieves an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs on a single NVIDIA RTX 4090 GPU. It's on only 18% lower than that achieved by a top-tier server-grade A100 GPU. It also significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy. There is a lot more innovation around inference that's coming fast. Really encouraged by the study on sparse computation to enhance the computational efficiency of LLMs. It's now possible to use PowerInfer with Llama 2 and Faclon 40B. Mistral-7B support is coming soon!

elvis

261,622 次观看 • 2 年前

OpenAI's Chief Futurist Joshua Achiam breaks with the consensus on takeoff, arguing intelligence hits a physical ceiling and then raw compute decides who wins: "People expect that there's no ceiling for the amount of intelligence that you could have in a model, and they think of recursive self-improvement as this loop that's gonna happen at some point or another. RSI starts happening and model intelligence takes off and it goes to the moon, and they don't see a ceiling." "I think on physical grounds, there's gotta be a maximum amount of computation that you can have per unit volume and energy in the physical universe. And that sort of implies there's a maximum amount of intelligence per unit volume and unit of energy." "If that's the case, eventually, seeing how fast AI model capabilities are increasing right now, eventually everyone hits that saturation point and everyone's got roughly equivalently capable models from a raw intelligence perspective." "The open source frontier lags the closed source frontier by some number of months, but the fact that it's months is crazy. So eventually everyone is probably working with equally maximally capable models, and then it's amount of compute that you're able to throw at a problem that determines who wins." Joshua Achiam

OpenAI's Chief Futurist Joshua Achiam breaks with the consensus on takeoff, arguing intelligence hits a physical ceiling and then raw compute decides who wins: "People expect that there's no ceiling for the amount of intelligence that you could have in a model, and they think of recursive self-improvement as this loop that's gonna happen at some point or another. RSI starts happening and model intelligence takes off and it goes to the moon, and they don't see a ceiling." "I think on physical grounds, there's gotta be a maximum amount of computation that you can have per unit volume and energy in the physical universe. And that sort of implies there's a maximum amount of intelligence per unit volume and unit of energy." "If that's the case, eventually, seeing how fast AI model capabilities are increasing right now, eventually everyone hits that saturation point and everyone's got roughly equivalently capable models from a raw intelligence perspective." "The open source frontier lags the closed source frontier by some number of months, but the fact that it's months is crazy. So eventually everyone is probably working with equally maximally capable models, and then it's amount of compute that you're able to throw at a problem that determines who wins." Joshua Achiam

MTS

60,213 次观看 • 20 天前

J.P. Morgan JUST said that quantum computing is, "a core focus" for them and they are, "investing BILLIONS." If you want to become a multi-billionaire this year, you need to go all-in on quantum right now - there's no waiting. Here's the best quantum stocks to buy: 1. IonQ $IONQ 2. D-Wave Quantum $QBTS 3. Rigetti Computing $RGTI 4. Quantum Computing $QUBT 5. Infleqtion $INFQ 6. Xanadu $XNDU 7. Arqit Quantum $ARQQ 8. Horizon Quantum $HQ 9. X-Energy (honorable mention) $XE 10. ETF's include: $QTUM, $QNTM, and $CHPX All my buy and sell signals in Discord @

J.P. Morgan JUST said that quantum computing is, "a core focus" for them and they are, "investing BILLIONS." If you want to become a multi-billionaire this year, you need to go all-in on quantum right now - there's no waiting. Here's the best quantum stocks to buy: 1. IonQ $IONQ 2. D-Wave Quantum $QBTS 3. Rigetti Computing $RGTI 4. Quantum Computing $QUBT 5. Infleqtion $INFQ 6. Xanadu $XNDU 7. Arqit Quantum $ARQQ 8. Horizon Quantum $HQ 9. X-Energy (honorable mention) $XE 10. ETF's include: $QTUM, $QNTM, and $CHPX All my buy and sell signals in Discord @

StockWhale

148,518 次观看 • 3 个月前

Microsoft and Atom Computing are coming together to ultimately build the world’s most powerful quantum machine. Learn more on how we are advancing the future of quantum computing: #AzureQuantum #MSIgnite

Microsoft and Atom Computing are coming together to ultimately build the world’s most powerful quantum machine. Learn more on how we are advancing the future of quantum computing: #AzureQuantum #MSIgnite

Microsoft

47,163 次观看 • 1 年前

Floki Trading Bot is currently hosting a massive RICE AI ( ◉ - ◉ ) trading competition on BNB Chain, with 2M $RICE tokens in rewards. This video walks you through how the competition works, how to earn points, and how to climb the leaderboard! 👉

Floki Trading Bot is currently hosting a massive RICE AI ( ◉ - ◉ ) trading competition on BNB Chain, with 2M $RICE tokens in rewards. This video walks you through how the competition works, how to earn points, and how to climb the leaderboard! 👉

Floki Trading Bot

16,433 次观看 • 10 个月前

Bank of America just said that, “quantum could be the biggest revolution for humanity since discovering fire.” If you want to become a billionaire, go all-in on quantum stocks. These are the best ones: 1. IonQ $IONQ 2. Rigetti Computing $RGTI 3. D-Wave Quantum $QBTS 4. Quantum Comptuing $QUBT 5. Xanadu Quantum $XNDU 6. Infleqtion $INFQ 7. Arqit Quantum $ARQQ 8. Horizon Quantum $HQ 9. SEALSQ $LAES All my buy and sell signals in Discord @

Bank of America just said that, “quantum could be the biggest revolution for humanity since discovering fire.” If you want to become a billionaire, go all-in on quantum stocks. These are the best ones: 1. IonQ $IONQ 2. Rigetti Computing $RGTI 3. D-Wave Quantum $QBTS 4. Quantum Comptuing $QUBT 5. Xanadu Quantum $XNDU 6. Infleqtion $INFQ 7. Arqit Quantum $ARQQ 8. Horizon Quantum $HQ 9. SEALSQ $LAES All my buy and sell signals in Discord @

StockWhale

256,539 次观看 • 3 个月前

At Microsoft, we are focused on empowering Frontier Transformation through Copilot, Microsoft IQ, and Agent 365. Businesses are not static, and neither are the AI systems that support them. We are helping organizations build, observe, and tune agentic workflows against the outcomes they seek and the ROI they demand. Our open, model-diverse, and heterogenous platform powers that improvement loop. Read my latest earnings blog to see how customers are bringing Frontier Transformation to life, demonstrating how intelligence, trust, and human ambition come together across industries.

At Microsoft, we are focused on empowering Frontier Transformation through Copilot, Microsoft IQ, and Agent 365. Businesses are not static, and neither are the AI systems that support them. We are helping organizations build, observe, and tune agentic workflows against the outcomes they seek and the ROI they demand. Our open, model-diverse, and heterogenous platform powers that improvement loop. Read my latest earnings blog to see how customers are bringing Frontier Transformation to life, demonstrating how intelligence, trust, and human ambition come together across industries.

Judson Althoff

24,468 次观看 • 16 天前