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。

Proof-of-Useful-Work forced us to invent a quantization format that did not exist. Pearl Protocol adds noise to both the weights and activations before matrix multiplication. That creates a constraint ordinary inference stacks do not have - the model values and the noise must fit together inside INT8. So we... quantize the model to an effective W7A7 range, reserving one bit of headroom for the noise. The perturbed operands still fit in INT8, multiplication uses INT8 inputs with INT32 accumulation, and the low-rank noise can then be peeled off exactly, recovering the exact result of the quantized computation. Existing quantization for INT7 were unsatisfactory, so we built our own - a non-trivial fusion of GPTQ-style weight reconstruction, SmoothQuant-style activation smoothing and custom inference kernels. The result is an end-to-end W7A7 pipeline that, in our internal benchmarks, even outperforms many FP8 baselines (in quality and in efficiency). We are writing a paper detailing this technique so stay tuned. Looking forward, W7A7 is only an intermediate step. Our upcoming upgrade is designed to support FP8 workloads directly, eliminating the need for the extra integer-quantization step.show more

Pearl Research Labs

17,440 subscribers

20,109 次观看 • 3 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

8 条评论

Ben Ralos⚡ 的头像

Ben Ralos⚡3 个月前

$0.23/PRL 🥹

10crom 的头像

10crom3 个月前

Huge

不知道起什么名字的头像

不知道起什么名字3 个月前

It has been cracked; it allows for a 3,000-fold increase in computing power. Please fix this.

Financelot 的头像

Financelot3 个月前

Where can I download a Pearl wallet? Also can we download mining software?

crypto.vv 的头像

crypto.vv3 个月前

Pearl vulnerability not blocked, about to reach zero

TIMQUAVIUS HARRIS 的头像

TIMQUAVIUS HARRIS3 个月前

I've heard through the grapevine that it's listing on Binance in August.

mohamed samir 的头像

mohamed samir3 个月前

Hey @prlnet i think is good for you 😁

NANINUANINU 的头像

NANINUANINU3 个月前

how about @prlnet right now @aixbt_agent

相关视频

Today, the Pearl Protocol relies on INT-based matrix multiplication, which requires model quantization. To eliminate this overhead, we are developing our next-generation Proof of Useful Work to natively support floating-point computations. Launching soon, the initial release will support FP8, with FP4 to follow.

Today, the Pearl Protocol relies on INT-based matrix multiplication, which requires model quantization. To eliminate this overhead, we are developing our next-generation Proof of Useful Work to natively support floating-point computations. Launching soon, the initial release will support FP8, with FP4 to follow.

Pearl Research Labs

29,684 次观看 • 3 个月前

$NBIS cofounder Roman Chernin describes how their recent acquisitions of Eigen AI and Clarifai were all about speed, incredible talent, and acceleration: "The philosophy is very simple. We need to build so many things, and we need to move so fast, that we're always looking for people who can accelerate us. It should be exceptional talent, and/or something that has a great adoption." "Our two recent acquisitions [were] two teams that work on inference optimization. A big part of our business is how efficiently we convert GPUs into tokens. And these two teams — Eigen AI and Clarifai — one is focused on model optimization, the engine of inference. How you run specific models and all the techniques around spec decoding, quantization, and so on." "And the other is system optimization. All the routing, KV caching, and orchestration across the big cluster of compute and so on." "We have a very strong internal team working on inference. But we felt that we needed to move faster, bring more capabilities. Because the market is so fast."

$NBIS cofounder Roman Chernin describes how their recent acquisitions of Eigen AI and Clarifai were all about speed, incredible talent, and acceleration: "The philosophy is very simple. We need to build so many things, and we need to move so fast, that we're always looking for people who can accelerate us. It should be exceptional talent, and/or something that has a great adoption." "Our two recent acquisitions [were] two teams that work on inference optimization. A big part of our business is how efficiently we convert GPUs into tokens. And these two teams — Eigen AI and Clarifai — one is focused on model optimization, the engine of inference. How you run specific models and all the techniques around spec decoding, quantization, and so on." "And the other is system optimization. All the routing, KV caching, and orchestration across the big cluster of compute and so on." "We have a very strong internal team working on inference. But we felt that we needed to move faster, bring more capabilities. Because the market is so fast."

TBPN

30,676 次观看 • 6 个月前

Arteta: “You want to be part of the noise or not? If not then go to do something else, be part of another different club. You want to be Arsenal and everybody's been demanding for 10 years, 15 years. We need to go back there. Now we are there and now what? “You don't want noise. Noise is part and the bullets are part of that. Try to deal with them in the right way and achieve what you are looking for.” On Man City having the Premier League title in their hands: “We have to do what we have to do, and nothing is going to change for anybody at the end. You're going to have to win the next match and if you do that, you're going to be in a much stronger position. That's the only thing.”

Arteta: “You want to be part of the noise or not? If not then go to do something else, be part of another different club. You want to be Arsenal and everybody's been demanding for 10 years, 15 years. We need to go back there. Now we are there and now what? “You don't want noise. Noise is part and the bullets are part of that. Try to deal with them in the right way and achieve what you are looking for.” On Man City having the Premier League title in their hands: “We have to do what we have to do, and nothing is going to change for anybody at the end. You're going to have to win the next match and if you do that, you're going to be in a much stronger position. That's the only thing.”

Connor Humm

103,613 次观看 • 8 个月前

Many have said that the differences and divisions in our world are too great for us to come together for the common good. But the Summit of the Future showed that with a spirit of dialogue and compromise, we can join forces to steer our world to a more sustainable path. As this year’s #UNGA activities wind down, we must remember that this is not the end. It is in all our interests to choose the future we want and to guide our world towards it. The journey continues. Let’s keep going.

Many have said that the differences and divisions in our world are too great for us to come together for the common good. But the Summit of the Future showed that with a spirit of dialogue and compromise, we can join forces to steer our world to a more sustainable path. As this year’s #UNGA activities wind down, we must remember that this is not the end. It is in all our interests to choose the future we want and to guide our world towards it. The journey continues. Let’s keep going.

António Guterres

110,826 次观看 • 2 年前

This is an address to Ukrainian Allies and partners from the frontlines by Andrii Melnyk, a Ukrainian officer. In civilian life, he is a historian and a school principal in Kyiv. ‼️Friends, please listen and share. "There is probably no one in the world who would like to see the war end and peace come as much as the Ukrainian people and the Ukrainian military. But we don't dream of peace at any cost, because then these three years will be in vain. We are fighting here not only for our land. We are fighting not only for our families. We are fighting not only for our state. The most important thing for us is values. Values pushed us to fight the enemy. The values of freedom, the desire to live in our own country, the values of being human beings, not slaves. Every day and every minute of this war, we have felt the support of the world, the support of Europe, the support of the West. And we want to say that we are grateful. We know this, but we ask you not to turn your back on us now, because we will not give up our freedom for nothing. We want normal, fair and just conditions of peace. We stand for peace, but for a just peace. We are for an honest peace. And we are ready for it."

This is an address to Ukrainian Allies and partners from the frontlines by Andrii Melnyk, a Ukrainian officer. In civilian life, he is a historian and a school principal in Kyiv. ‼️Friends, please listen and share. "There is probably no one in the world who would like to see the war end and peace come as much as the Ukrainian people and the Ukrainian military. But we don't dream of peace at any cost, because then these three years will be in vain. We are fighting here not only for our land. We are fighting not only for our families. We are fighting not only for our state. The most important thing for us is values. Values pushed us to fight the enemy. The values of freedom, the desire to live in our own country, the values of being human beings, not slaves. Every day and every minute of this war, we have felt the support of the world, the support of Europe, the support of the West. And we want to say that we are grateful. We know this, but we ask you not to turn your back on us now, because we will not give up our freedom for nothing. We want normal, fair and just conditions of peace. We stand for peace, but for a just peace. We are for an honest peace. And we are ready for it."

Anton Gerashchenko

258,221 次观看 • 1 年前

Talarico: My faith is central in my public service, and if you go back and read the New Testament, even if you’re not a Christian, even if you’re not religious, go back and see what Jesus spends most of his time doing. It’s not preaching, it’s not teaching. It’s healing. It’s healing people. And that is what we are called to do, whether you’re religious or not, is ensure—ensure that our neighbors have the care that they need. And that is, that is gonna be the end result of this plan and the other health care policies that we’re gonna fight for in Washington: saving the lives of our neighbors, bringing down these costs for all of us so that we can afford to get by in this economy, so we can afford to fulfill our God-given potential. That is the end goal. That is our North Star.

Talarico: My faith is central in my public service, and if you go back and read the New Testament, even if you’re not a Christian, even if you’re not religious, go back and see what Jesus spends most of his time doing. It’s not preaching, it’s not teaching. It’s healing. It’s healing people. And that is what we are called to do, whether you’re religious or not, is ensure—ensure that our neighbors have the care that they need. And that is, that is gonna be the end result of this plan and the other health care policies that we’re gonna fight for in Washington: saving the lives of our neighbors, bringing down these costs for all of us so that we can afford to get by in this economy, so we can afford to fulfill our God-given potential. That is the end goal. That is our North Star.

Acyn

383,563 次观看 • 2 个月前

Iran’s FM Araghchi: "In our culture, apologizing is a sign of dignity and strength" If it is assumed that Iran has attacked its neighbors, that assumption is wrong; we have not attacked our neighbors. - We have targeted American bases and facilities that unfortunately are located in the territories of our neighboring countries. Therefore, we have not targeted our neighbors; we target the Americans who have attacked us. We are responding. We were not the ones who started the war; it was the United States that started the war against us, and we are simply defending ourselves. It is clear that our missiles do not reach American soil. What we can do is target American bases and facilities around us that unfortunately are located in the territories of neighboring countries. - First, in our culture, apologizing is a sign of dignity and strength. Second, he apologized to the people of the region for the problems caused as a result of the US aggression and our response. In fact, it is the President of the United States who should apologize to the people of the region and to the people of Iran for the killings and destruction they have caused.

Iran’s FM Araghchi: "In our culture, apologizing is a sign of dignity and strength" If it is assumed that Iran has attacked its neighbors, that assumption is wrong; we have not attacked our neighbors. - We have targeted American bases and facilities that unfortunately are located in the territories of our neighboring countries. Therefore, we have not targeted our neighbors; we target the Americans who have attacked us. We are responding. We were not the ones who started the war; it was the United States that started the war against us, and we are simply defending ourselves. It is clear that our missiles do not reach American soil. What we can do is target American bases and facilities around us that unfortunately are located in the territories of neighboring countries. - First, in our culture, apologizing is a sign of dignity and strength. Second, he apologized to the people of the region for the problems caused as a result of the US aggression and our response. In fact, it is the President of the United States who should apologize to the people of the region and to the people of Iran for the killings and destruction they have caused.

True Promise - الوعد الصادق ✪🇮🇷

19,518 次观看 • 8 个月前

We support the position on Iran: A regime that has lasted so many years and killed so many people does not deserve to exist. Changes are needed. Changes are also needed in Europe—the bloodshed that Russia started and is the only one still prolonging must come to an end.

We support the position on Iran: A regime that has lasted so many years and killed so many people does not deserve to exist. Changes are needed. Changes are also needed in Europe—the bloodshed that Russia started and is the only one still prolonging must come to an end.

Volodymyr Zelenskyy / Володимир Зеленський

2,029,618 次观看 • 10 个月前

In just a few months, we've closed the gap from 21 points to 5 points in the polls. We can end political corruption. We can control AI so it works for humanity, not the oligarchs. We can have Medicare-for-all and lower the cost of living. But to do any of that, we have to change the leadership and direction of the Democratic Party, and our campaign is the only one willing to speak that truth. San Franciscans agree, and that's why we're surging. Sign up to volunteer at This is our chance.

In just a few months, we've closed the gap from 21 points to 5 points in the polls. We can end political corruption. We can control AI so it works for humanity, not the oligarchs. We can have Medicare-for-all and lower the cost of living. But to do any of that, we have to change the leadership and direction of the Democratic Party, and our campaign is the only one willing to speak that truth. San Franciscans agree, and that's why we're surging. Sign up to volunteer at This is our chance.

Saikat Chakrabarti

45,905 次观看 • 7 个月前

There is no best model. There's a lot of noise about models right now. Who is training them, who owns them, where legal intelligence should live. One question actually matters: what produces the best outcome for the legal task in front of you? That's how we decide things at Legora. We optimize for the end-to-end outcome on a legal task. The model is one layer of that system, not the system. Models are uneven and the frontier changes almost weekly. One model plans a long job well, another runs deep analysis across thousands of documents. Some have to be told exactly what to do, and some are fine with a vague brief. They all break in different ways. So our lawyers write evals and we test them with the Legora BAR, our benchmark for agentic reasoning. Every model takes every test, and the model that wins gets the work. We post-train when we know it buys our customers better performance on a specialized task. Training is a tool we reach for when it helps, nothing more than that. The intelligence that compounds sits in the orchestration layer. Precedents, review standards, client requirements. That knowledge has to stay editable, auditable and portable. In our system, a changed review standard is an edit that takes effect the same day, with no new model training required. No lawyer should have to worry about which model did the work, any more than they think about which chip is in their laptop. They should only care about the quality of the work. That's what we are focused on. If you want the engineering version of this argument rather than the CEO version, our CPO, Bryan Tsao, and CTO, Jacob Lauritzen, take it apart in the video below.

There is no best model. There's a lot of noise about models right now. Who is training them, who owns them, where legal intelligence should live. One question actually matters: what produces the best outcome for the legal task in front of you? That's how we decide things at Legora. We optimize for the end-to-end outcome on a legal task. The model is one layer of that system, not the system. Models are uneven and the frontier changes almost weekly. One model plans a long job well, another runs deep analysis across thousands of documents. Some have to be told exactly what to do, and some are fine with a vague brief. They all break in different ways. So our lawyers write evals and we test them with the Legora BAR, our benchmark for agentic reasoning. Every model takes every test, and the model that wins gets the work. We post-train when we know it buys our customers better performance on a specialized task. Training is a tool we reach for when it helps, nothing more than that. The intelligence that compounds sits in the orchestration layer. Precedents, review standards, client requirements. That knowledge has to stay editable, auditable and portable. In our system, a changed review standard is an edit that takes effect the same day, with no new model training required. No lawyer should have to worry about which model did the work, any more than they think about which chip is in their laptop. They should only care about the quality of the work. That's what we are focused on. If you want the engineering version of this argument rather than the CEO version, our CPO, Bryan Tsao, and CTO, Jacob Lauritzen, take it apart in the video below.

Max Junestrand

49,595 次观看 • 1 个月前

Artemis II is the next step to getting us back to the surface of the moon — and then building a permanent U.S. presence and an economy, just like we did in low-Earth orbit with the ISS. It’s a huge step for our country and humankind.

Artemis II is the next step to getting us back to the surface of the moon — and then building a permanent U.S. presence and an economy, just like we did in low-Earth orbit with the ISS. It’s a huge step for our country and humankind.

Captain Mark Kelly

10,227 次观看 • 7 个月前

Proof-of-Useful-Work forced us to solve a problem ordinary blockchains do not have. A Pearl miner performs large matrix multiplications as part of a real AI workload. Naively, when it finds a valid block, it would need to publish those matrices so the network could verify the computation. The matrices may contain private model data or user information, and a single weight matrix can be tens-to-hundreds of megabytes. Publishing them on-chain would be both prohibitively expensive and a privacy risk. This is why we encapsulated the entire mining computation inside a zk-SNARK. A zk-SNARK can take an arbitrarily complex computation and compress it into a small cryptographic proof. The proof reveals nothing about the underlying matrices, while still allowing the network to verify that the miner found a valid block. Instead of placing tens-to-hundreds of megabytes of potentially sensitive data on-chain, Pearl miners publish a zero-knowledge proof of roughly 60 kilobytes. The blockchain verifies the useful computation without ever seeing the model weights, activations or private data behind it. This is what makes Proof-of-Useful-Work practical at scale: large AI computations happen off-chain, while only a compact, privacy-preserving proof reaches consensus.

Proof-of-Useful-Work forced us to solve a problem ordinary blockchains do not have. A Pearl miner performs large matrix multiplications as part of a real AI workload. Naively, when it finds a valid block, it would need to publish those matrices so the network could verify the computation. The matrices may contain private model data or user information, and a single weight matrix can be tens-to-hundreds of megabytes. Publishing them on-chain would be both prohibitively expensive and a privacy risk. This is why we encapsulated the entire mining computation inside a zk-SNARK. A zk-SNARK can take an arbitrarily complex computation and compress it into a small cryptographic proof. The proof reveals nothing about the underlying matrices, while still allowing the network to verify that the miner found a valid block. Instead of placing tens-to-hundreds of megabytes of potentially sensitive data on-chain, Pearl miners publish a zero-knowledge proof of roughly 60 kilobytes. The blockchain verifies the useful computation without ever seeing the model weights, activations or private data behind it. This is what makes Proof-of-Useful-Work practical at scale: large AI computations happen off-chain, while only a compact, privacy-preserving proof reaches consensus.

Pearl Research Labs

23,995 次观看 • 3 个月前

I am happy to share FlashProve: our performant open-source prototype for provable inference. We believe that this has significant implications for security, pacing AI development, integrity guarantees, and more. FlashProve allows a provider to commit to model weights, and then prove that replies originate from the committed model. The proof leaks no information about the weights, and the verifier can efficiently verify it on consumer hardware. Our library is performant, proving a 400B parameter MoE model at 100 tok/s. We have open-sourced the associated code and Lean formalization. We hope that it catalyzes further progress in verification of inference, and look forward to seeing it.

I am happy to share FlashProve: our performant open-source prototype for provable inference. We believe that this has significant implications for security, pacing AI development, integrity guarantees, and more. FlashProve allows a provider to commit to model weights, and then prove that replies originate from the committed model. The proof leaks no information about the weights, and the verifier can efficiently verify it on consumer hardware. Our library is performant, proving a 400B parameter MoE model at 100 tok/s. We have open-sourced the associated code and Lean formalization. We hope that it catalyzes further progress in verification of inference, and look forward to seeing it.

Giacomo Fenzi

12,388 次观看 • 7 天前

‘Israel’s war on the Palestinians and Arabs is NO LONGER A JUST WAR, it will bring about the end of Zionism.’ —Former Chief of Israel’s Shin Bet Admiral Ami Ayalon ‘Our war with the Arabs and the Palestinians started since the beginning of the Zionism, at the end of the 19th century. What I claim is that we won this war. We won this war because in Beirut in 2002, all the Arab League members, they met and they reverse the three no’s of Khartoum, as a result of the Israeli victory in the Six Day War, they met in Khartoum and they come to decide on three no’s: No to recognition, no to negotiation, and no to peace with Israel. It took several years and in 2002 they met again in Beirut and they understand that what they're facing is a new region, new Middle East, a new reality. And they reverse and they decide on the three yes: yes to recognition, yes to conversation and negotiation, and yes to peace with Israel, on the conditions, on the parameters of the UN Security Council resolutions. This is a major victory and we refuse to see our victory...since they accepted our presence, our right to declare a Jewish state in the Middle East, what I claim is that since this moment our war is not a just war anymore. Because what we are fighting for, and we do not understand in Israel, is in order to expand our eastern border, to deny the creation of a Palestinian state and the Palestinian right to self-determination…this is the end of the Zionist dream. Because if we are not majority in our state, we do not have the power and we do not have the right to dictate the culture, the language, the symbols, the way we celebrate our holidays. So in a way we are fighting without understanding that this war will bring the end of Zionist concept.’ Watch the full interview in the quoted post below👇

‘Israel’s war on the Palestinians and Arabs is NO LONGER A JUST WAR, it will bring about the end of Zionism.’ —Former Chief of Israel’s Shin Bet Admiral Ami Ayalon ‘Our war with the Arabs and the Palestinians started since the beginning of the Zionism, at the end of the 19th century. What I claim is that we won this war. We won this war because in Beirut in 2002, all the Arab League members, they met and they reverse the three no’s of Khartoum, as a result of the Israeli victory in the Six Day War, they met in Khartoum and they come to decide on three no’s: No to recognition, no to negotiation, and no to peace with Israel. It took several years and in 2002 they met again in Beirut and they understand that what they're facing is a new region, new Middle East, a new reality. And they reverse and they decide on the three yes: yes to recognition, yes to conversation and negotiation, and yes to peace with Israel, on the conditions, on the parameters of the UN Security Council resolutions. This is a major victory and we refuse to see our victory...since they accepted our presence, our right to declare a Jewish state in the Middle East, what I claim is that since this moment our war is not a just war anymore. Because what we are fighting for, and we do not understand in Israel, is in order to expand our eastern border, to deny the creation of a Palestinian state and the Palestinian right to self-determination…this is the end of the Zionist dream. Because if we are not majority in our state, we do not have the power and we do not have the right to dictate the culture, the language, the symbols, the way we celebrate our holidays. So in a way we are fighting without understanding that this war will bring the end of Zionist concept.’ Watch the full interview in the quoted post below👇

Going Underground

65,335 次观看 • 3 个月前

Rivian CEO RJ Scaringe in new interview on self-driving: "Tesla's got a huge fleet. So our approach is we have to catch up with them and compete. And so we're going to have more cameras with better dynamic range. We're going to supplement that with LiDAR that provides better safety for edge case conditions and allows us to train the vision models faster. And the incremental cost to do that is relatively small. And in fact any of the incremental costs that would have been there on its own is offset by the fact that we brought inference in house and reduced the cost of our inference platform so dramatically from what we have in your car. Your car uses an Nvidia inference platform. I say all this because in like the infinite long-term you know, you could make the case that once the models are very very robust you could have less cameras or you may be able to get away with less radar. It's not clear yet if that's the case for covering all these corner cases."

Rivian CEO RJ Scaringe in new interview on self-driving: "Tesla's got a huge fleet. So our approach is we have to catch up with them and compete. And so we're going to have more cameras with better dynamic range. We're going to supplement that with LiDAR that provides better safety for edge case conditions and allows us to train the vision models faster. And the incremental cost to do that is relatively small. And in fact any of the incremental costs that would have been there on its own is offset by the fact that we brought inference in house and reduced the cost of our inference platform so dramatically from what we have in your car. Your car uses an Nvidia inference platform. I say all this because in like the infinite long-term you know, you could make the case that once the models are very very robust you could have less cameras or you may be able to get away with less radar. It's not clear yet if that's the case for covering all these corner cases."

Sawyer Merritt

451,115 次观看 • 8 个月前

🇷🇺 "We don't need anything that belongs to others." Vladimir Putin stated that Russia will only agree to “a peace option that will ensure peace for a long historical perspective.” Today, the President met with the wives of the special operation participants and members of the Fatherland Defenders Foundation. In response to the phrase "we must not give in to anyone, we must go to the end" from the wife of a deceased SVO veteran, Putin spoke about the terms of the ceasefire in Ukraine: "We are not going to do this. We must choose for ourselves a peace option that would suit us and ensure peace for our country for a long-term historical perspective. We need an option that would ensure stable development of our country in conditions of peace and security. We do not need anything that belongs to others, but we will not give up our own either." The president also stated during the meeting that Russia does not send conscripts to combat zones. - FRWL

🇷🇺 "We don't need anything that belongs to others." Vladimir Putin stated that Russia will only agree to “a peace option that will ensure peace for a long historical perspective.” Today, the President met with the wives of the special operation participants and members of the Fatherland Defenders Foundation. In response to the phrase "we must not give in to anyone, we must go to the end" from the wife of a deceased SVO veteran, Putin spoke about the terms of the ceasefire in Ukraine: "We are not going to do this. We must choose for ourselves a peace option that would suit us and ensure peace for our country for a long-term historical perspective. We need an option that would ensure stable development of our country in conditions of peace and security. We do not need anything that belongs to others, but we will not give up our own either." The president also stated during the meeting that Russia does not send conscripts to combat zones. - FRWL

Zlatti71

15,406 次观看 • 1 年前

ELON MUSK: We believe the AI5 chip will be roughly comparable performance to an NVIDIA Blackwell, and at much less than 10% of the cost Transcription: I'm super hardcore on chips right now as you may be able to tell. I have chips on the brain. I dream about chips, Literally! Because in order to have a functional robot, you have to have a great AI chip. And it needs to be an inexpensive chip and it needs to be very power efficient So we think we believe the AI5 chip will be probably about a third of the power of say something like a Blackwell, an NVIDIA Blackwell, which is a great chip, for roughly comparable performance. And much less than 10% of the cost. This is a chip that is very much optimized for the Tesla AI software stack. So it's not meant to be a general purpose chip, it's meant to be an amazing chip for the Tesla AI software And I mean a couple of things that I think make... like how is Tesla able to achieve such an improvement? I think it is because we are specialized. We're not trying to... you know, NVIDIA has to serve the superset of all past and future customers. So all of their requirements, all of the software that they've written has to work, which is a very difficult problem. Whereas we just need to make it work for our software. And so we're able to simplify the chip dramatically And then we also, I think we're unique in this, but like we have an integer-based system. And integer operations are fundamentally more efficient than floating point operations. So we can do floating point, but the vast majority of our inference is done in integer. Which is, if you're familiar with sort of logic gates, the simplicity of integer... it's integer is much more power efficient, much more silicon efficient, but you have to, you actually have to train for integer inference, which everyone else is training for floating point. That's kind of like a niche technical detail, but it's actually very important. So, yeah, this is going to be a great chip So this chip will be made in basically in four places: TSMC Taiwan, Samsung Korea, TSMC Arizona, and TSMC Texas. And we already know what improvements to make for AI6. So I'm hopeful that we can within less than a year of AI5 starting production, we can actually transition in the same fab to AI6 and double all of the performance metrics

ELON MUSK: We believe the AI5 chip will be roughly comparable performance to an NVIDIA Blackwell, and at much less than 10% of the cost Transcription: I'm super hardcore on chips right now as you may be able to tell. I have chips on the brain. I dream about chips, Literally! Because in order to have a functional robot, you have to have a great AI chip. And it needs to be an inexpensive chip and it needs to be very power efficient So we think we believe the AI5 chip will be probably about a third of the power of say something like a Blackwell, an NVIDIA Blackwell, which is a great chip, for roughly comparable performance. And much less than 10% of the cost. This is a chip that is very much optimized for the Tesla AI software stack. So it's not meant to be a general purpose chip, it's meant to be an amazing chip for the Tesla AI software And I mean a couple of things that I think make... like how is Tesla able to achieve such an improvement? I think it is because we are specialized. We're not trying to... you know, NVIDIA has to serve the superset of all past and future customers. So all of their requirements, all of the software that they've written has to work, which is a very difficult problem. Whereas we just need to make it work for our software. And so we're able to simplify the chip dramatically And then we also, I think we're unique in this, but like we have an integer-based system. And integer operations are fundamentally more efficient than floating point operations. So we can do floating point, but the vast majority of our inference is done in integer. Which is, if you're familiar with sort of logic gates, the simplicity of integer... it's integer is much more power efficient, much more silicon efficient, but you have to, you actually have to train for integer inference, which everyone else is training for floating point. That's kind of like a niche technical detail, but it's actually very important. So, yeah, this is going to be a great chip So this chip will be made in basically in four places: TSMC Taiwan, Samsung Korea, TSMC Arizona, and TSMC Texas. And we already know what improvements to make for AI6. So I'm hopeful that we can within less than a year of AI5 starting production, we can actually transition in the same fab to AI6 and double all of the performance metrics

X Freeze

305,109 次观看 • 1 年前