Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

This is our Dell Technologies 7875 Workstation with dual Blackwell RTX6000 GPUs. Each one has 96GB for a total of 192GB of VRAM. It's running the training for my Robotron AI, but since that doesn't use much memory, I added vLLM instances of DeepseekR1-32B and Qwen2.5-Instruct, which serve several... show more

Dave W Plummer

103,435 subscribers

59,682 views • 3 months ago •via X (Twitter)

Education News & Politics Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

EIGHT USED RTX 3090s. 192GB OF VRAM. ONE PRIVATE BOX BUILT TO RUN 70B AI MODELS WITHOUT RENTING A DATACENTER. 00:06 the team lowers all eight GPUs into the chassis at once, turning a pile of old gaming cards into the core of a serious local AI server. each 3090 contributes 24GB of memory. combined, the rack has enough VRAM for heavyweight models, large document collections, long context and multiple jobs running side by side. these cards are five years old, but AI economics work differently from gaming. memory capacity matters more than having the newest badge, which is why used 3090s remain valuable. the server can keep one model loaded while separate GPUs handle video, images, transcription and overnight agents without sending private company data into the cloud. bookmark this build before yesterday’s gaming GPUs become tomorrow’s cheapest AI infrastructure.

EIGHT USED RTX 3090s. 192GB OF VRAM. ONE PRIVATE BOX BUILT TO RUN 70B AI MODELS WITHOUT RENTING A DATACENTER. 00:06 the team lowers all eight GPUs into the chassis at once, turning a pile of old gaming cards into the core of a serious local AI server. each 3090 contributes 24GB of memory. combined, the rack has enough VRAM for heavyweight models, large document collections, long context and multiple jobs running side by side. these cards are five years old, but AI economics work differently from gaming. memory capacity matters more than having the newest badge, which is why used 3090s remain valuable. the server can keep one model loaded while separate GPUs handle video, images, transcription and overnight agents without sending private company data into the cloud. bookmark this build before yesterday’s gaming GPUs become tomorrow’s cheapest AI infrastructure.

Gipp 🦅

118,736 views • 23 days ago

INTEL JUST SHIPPED A WORKSTATION CARD WITH TWO GPUs ON ONE PCB AND 48GB OF VRAM. FOUR CARDS GIVE A SINGLE MOTHERBOARD 192GB OF POOLED INFERENCE MEMORY FOR THE PRICE OF TWO RTX 5090s 00:14 he holds one card up to the camera, two GPU dies side by side under the cooler, each one running its own x8 PCIe lane back to the chipset the cluster sees eight discrete accelerators in software. intel's Battle Matrix stack shards a model across all eight, so a 235B parameter network loads in slices and answers requests in parallel what 192GB of VRAM unlocks: an entire 200B class model in memory without quantization. a vision agent reading 100 invoices at once. a research box that holds three frontier models loaded simultaneously, switching between them in under a second intel is the slow side of inference. nvidia is faster per token, that is the honest tradeoff. but the only other path to this much VRAM is a $40,000 nvidia rack or three networked Mac Studios four B60 cards plus the chassis lands under $5,000. power draw averages 800 watts, $55 a month in electricity. one engineer paying $400 a month for combined ChatGPT Pro and Claude Code Max pays the hardware off in less than a year

INTEL JUST SHIPPED A WORKSTATION CARD WITH TWO GPUs ON ONE PCB AND 48GB OF VRAM. FOUR CARDS GIVE A SINGLE MOTHERBOARD 192GB OF POOLED INFERENCE MEMORY FOR THE PRICE OF TWO RTX 5090s 00:14 he holds one card up to the camera, two GPU dies side by side under the cooler, each one running its own x8 PCIe lane back to the chipset the cluster sees eight discrete accelerators in software. intel's Battle Matrix stack shards a model across all eight, so a 235B parameter network loads in slices and answers requests in parallel what 192GB of VRAM unlocks: an entire 200B class model in memory without quantization. a vision agent reading 100 invoices at once. a research box that holds three frontier models loaded simultaneously, switching between them in under a second intel is the slow side of inference. nvidia is faster per token, that is the honest tradeoff. but the only other path to this much VRAM is a $40,000 nvidia rack or three networked Mac Studios four B60 cards plus the chassis lands under $5,000. power draw averages 800 watts, $55 a month in electricity. one engineer paying $400 a month for combined ChatGPT Pro and Claude Code Max pays the hardware off in less than a year

NO1ennn

48,054 views • 26 days ago

Cathie Wood on Tesla in a new interview: “We've always said Tesla is not an auto company. It is actually the convergence of three of the platforms I mentioned; Robotics, Energy Storage, and AI. Each one of those technologies has its own S-curve, now they're feeding each other, and we’re seeing that in Robotaxis. Robotaxis, we believe, will account for 90% of Tesla's valuation by the end of the decade. We're in print at $2600 (PT). That includes nothing for Optimus.”

Cathie Wood on Tesla in a new interview: “We've always said Tesla is not an auto company. It is actually the convergence of three of the platforms I mentioned; Robotics, Energy Storage, and AI. Each one of those technologies has its own S-curve, now they're feeding each other, and we’re seeing that in Robotaxis. Robotaxis, we believe, will account for 90% of Tesla's valuation by the end of the decade. We're in print at $2600 (PT). That includes nothing for Optimus.”

Nic Cruz Patane

88,440 views • 6 months ago

SITUATION EXPLAINED: Cerebras raised $5.55 billion in their IPO and closing their first day of trading valued at $66 billion, making it the biggest US tech IPO since Snowflake in 2020. Cerebras makes Wafer-Scale Engine chips built for AI inference. We asked Sarah Fong the main difference between wafer-scale chips and traditional GPUs: - GPUs are great at parallel work (graphics, training) - AI inference is sequential, AKA one token at a time This causes the "memory wall" problem: - Every GPU core needs model weights, KV cache, and activations to do its math - On a GPU, that data lives in off-chip memory (HBM) - Cores constantly load and offload from off-chip memory, which is a huge bottleneck; hardware accounts for ~70% of inference latency Cerebras' chips: -Dinner-plate sized (vs. GPUs which are palm-sized) with tens of thousands of cores -Memory sits directly on top of the cores as distributed SRAM -Weights and KV cache can be accessed at on-chip speeds in the PB/s range, compared with off-chip speeds in the TB/s range achieved by GPUs with HBM.

SITUATION EXPLAINED: Cerebras raised $5.55 billion in their IPO and closing their first day of trading valued at $66 billion, making it the biggest US tech IPO since Snowflake in 2020. Cerebras makes Wafer-Scale Engine chips built for AI inference. We asked Sarah Fong the main difference between wafer-scale chips and traditional GPUs: - GPUs are great at parallel work (graphics, training) - AI inference is sequential, AKA one token at a time This causes the "memory wall" problem: - Every GPU core needs model weights, KV cache, and activations to do its math - On a GPU, that data lives in off-chip memory (HBM) - Cores constantly load and offload from off-chip memory, which is a huge bottleneck; hardware accounts for ~70% of inference latency Cerebras' chips: -Dinner-plate sized (vs. GPUs which are palm-sized) with tens of thousands of cores -Memory sits directly on top of the cores as distributed SRAM -Weights and KV cache can be accessed at on-chip speeds in the PB/s range, compared with off-chip speeds in the TB/s range achieved by GPUs with HBM.

MTS

44,057 views • 2 months ago

THIS CHINESE BUILDER TURNED 8 RETIRED TESLA P40s INTO A 192GB PRIVATE AI SERVER THAT CAN WORK 5,760 GPU-HOURS EVERY MONTH. 00:06 he holds up one used Tesla P40 while seven more sit behind it, then installs the full stack inside a massive dual-Xeon server chassis. each card carries 24GB of VRAM, giving the system 192GB across eight GPUs for local models, video processing, transcription, private documents and AI agents. the real advantage is parallel work. some cards can serve a local model while the rest generate images, process files and run overnight automation at the same time. eight GPUs running continuously deliver 5,760 GPU-hours every 30 days. rent that workload from the cloud and the meter keeps moving for every minute the server stays online. a $412 monthly AI stack already costs $4,944 a year before serious GPU rental. he bought retired hardware once, kept the data local and built a machine that keeps working after the browser closes.

THIS CHINESE BUILDER TURNED 8 RETIRED TESLA P40s INTO A 192GB PRIVATE AI SERVER THAT CAN WORK 5,760 GPU-HOURS EVERY MONTH. 00:06 he holds up one used Tesla P40 while seven more sit behind it, then installs the full stack inside a massive dual-Xeon server chassis. each card carries 24GB of VRAM, giving the system 192GB across eight GPUs for local models, video processing, transcription, private documents and AI agents. the real advantage is parallel work. some cards can serve a local model while the rest generate images, process files and run overnight automation at the same time. eight GPUs running continuously deliver 5,760 GPU-hours every 30 days. rent that workload from the cloud and the meter keeps moving for every minute the server stays online. a $412 monthly AI stack already costs $4,944 a year before serious GPU rental. he bought retired hardware once, kept the data local and built a machine that keeps working after the browser closes.

Gipp 🦅

42,332 views • 24 days ago

TESLA ISN'T JUST AHEAD IN AI, IT'S AN ORDER OF MAGNITUDE MORE EFFICIENT THAN THE REST OF THE WORLD "I actually think Tesla is ahead of the rest of the world in intelligence density of AI by an order of magnitude or more This is going to sound like a pretty bold statement, but I know what the intelligence efficiency of the big models are like, Grok and a bunch of the other models Tesla’s AI is, in terms of its memory efficiency, I think more than an order of magnitude better. The Tesla AI is very compute-efficient and very memory-efficient One of the metrics one should consider for any given AI model is the intelligence per gigabyte. Especially when you’re constrained on RAM, having an AI that has very high intelligence density per gigabyte for a given number of gigabytes, how much functionality can you get out of it" - Elon Musk

TESLA ISN'T JUST AHEAD IN AI, IT'S AN ORDER OF MAGNITUDE MORE EFFICIENT THAN THE REST OF THE WORLD "I actually think Tesla is ahead of the rest of the world in intelligence density of AI by an order of magnitude or more This is going to sound like a pretty bold statement, but I know what the intelligence efficiency of the big models are like, Grok and a bunch of the other models Tesla’s AI is, in terms of its memory efficiency, I think more than an order of magnitude better. The Tesla AI is very compute-efficient and very memory-efficient One of the metrics one should consider for any given AI model is the intelligence per gigabyte. Especially when you’re constrained on RAM, having an AI that has very high intelligence density per gigabyte for a given number of gigabytes, how much functionality can you get out of it" - Elon Musk

X Freeze

62,736 views • 5 months ago

Deep Research is the most fun I've had with an AI tool since NotebookLM was launched! You can't imagine how incredible this agent feels for someone like me that loves doing research. I took it for a spin here and added my thoughts: I am impressed by the reasoning capabilities of the o3 model which operated under the hood. This is RL at its best once again! Synthesizing large amounts of information from different sources is no easy feat. I can tell you all about it from my experience building and scaling RAG systems. I have so many ways I plan to use this. More on this soon.

Deep Research is the most fun I've had with an AI tool since NotebookLM was launched! You can't imagine how incredible this agent feels for someone like me that loves doing research. I took it for a spin here and added my thoughts: I am impressed by the reasoning capabilities of the o3 model which operated under the hood. This is RL at its best once again! Synthesizing large amounts of information from different sources is no easy feat. I can tell you all about it from my experience building and scaling RAG systems. I have so many ways I plan to use this. More on this soon.

elvis

26,459 views • 1 year ago

I'm oddly proud and stoked about this one - live video preview w/ audio of the Robotron AI training on the web. You can watch it live here: I started simple yesterday, a canvas I updated with a snapshot of the game that I pumped over a socket from an emulator session. That evolved into a video tag with a properly encoded video stream, and today I spent a full damned day almost adding audio and keeping it in sync with the game (which runs at an uncorked and unpredictable framerate). Assuming my machine can handle the load, it's available for viewing pleasure now at the link above :-) You have to click on the preview to enable the audio, and you can turn it off/on at the bottom of the page. Thanks to dsobeski for beta-testing it!

I'm oddly proud and stoked about this one - live video preview w/ audio of the Robotron AI training on the web. You can watch it live here: I started simple yesterday, a canvas I updated with a snapshot of the game that I pumped over a socket from an emulator session. That evolved into a video tag with a properly encoded video stream, and today I spent a full damned day almost adding audio and keeping it in sync with the game (which runs at an uncorked and unpredictable framerate). Assuming my machine can handle the load, it's available for viewing pleasure now at the link above :-) You have to click on the preview to enable the audio, and you can turn it off/on at the bottom of the page. Thanks to dsobeski for beta-testing it!

Dave W Plummer

15,271 views • 4 months ago

I'm going to be controversial here. I would love to pay to have my entire series animated. I would hope one day my books get picked up and this becomes a reality. But being realistic it's not going to happen. I will draw, write my own works and commission artists where i can. But seeing this, is a dream come true for me, even if its Ai. I'm glad that the original picture is my own hand. Hate all you want. It's okay to like things and I like seeing this in a way that would take a life time snd a shit load of money to do. How much I want to see this for real one day.

I'm going to be controversial here. I would love to pay to have my entire series animated. I would hope one day my books get picked up and this becomes a reality. But being realistic it's not going to happen. I will draw, write my own works and commission artists where i can. But seeing this, is a dream come true for me, even if its Ai. I'm glad that the original picture is my own hand. Hate all you want. It's okay to like things and I like seeing this in a way that would take a life time snd a shit load of money to do. How much I want to see this for real one day.

Zack Riley 🇦🇺

22,791 views • 2 months ago

INTEL JUST QUIETLY BUILT THE CHEAPEST PATH TO 128GB OF LOCAL AI VRAM, AND IT BEATS NVIDIA'S DGX SPARK BY 2X FOR THE SAME $4,000 00:11 four Arc Pro B70s racked in one chassis. 32GB of GDDR6 each, $949 MSRP, roughly $1,000 street per card stack four and you get 128GB of VRAM for about $4,000 total. a lab benchmarked the quad against a DGX Spark and a single RTX Pro 6000. the four B70s doubled the Spark or better they trail the Pro 6000, but that card is $10,000. so the real choice is now $4K for 128GB from Intel or $10K for 96GB from NVIDIA the article covers the DGX Spark at $3,999 as the single-desk local AI ceiling. this Intel quad matches the price, beats the memory throughput, and nobody is talking about it yet save this before Intel's VRAM-per-dollar advantage becomes the default homelab build ↓

INTEL JUST QUIETLY BUILT THE CHEAPEST PATH TO 128GB OF LOCAL AI VRAM, AND IT BEATS NVIDIA'S DGX SPARK BY 2X FOR THE SAME $4,000 00:11 four Arc Pro B70s racked in one chassis. 32GB of GDDR6 each, $949 MSRP, roughly $1,000 street per card stack four and you get 128GB of VRAM for about $4,000 total. a lab benchmarked the quad against a DGX Spark and a single RTX Pro 6000. the four B70s doubled the Spark or better they trail the Pro 6000, but that card is $10,000. so the real choice is now $4K for 128GB from Intel or $10K for 96GB from NVIDIA the article covers the DGX Spark at $3,999 as the single-desk local AI ceiling. this Intel quad matches the price, beats the memory throughput, and nobody is talking about it yet save this before Intel's VRAM-per-dollar advantage becomes the default homelab build ↓

NO1ennn

136,365 views • 6 days ago

CNN’s Kasie Hunt: “The Heritage Foundation has documented a total of 25 instances of prosecution for voter fraud where citizenship was an issue. Just 25. I mean has this ever actually affected an American election?” DHS Secretary Markwayne Mullin: “Well, 25 is 25 too many. It's kinda like one individual that dies from the hands of an illegal is one too many. It’s all preventable. One person voting illegally is one too many.” CNN’s Kasie Hunt: “Do you agree it doesn’t happen very much?” DHS Secretary Markwayne Mullin: “No, I’m not saying it doesn’t happen very much or not. If you start looking and scrubbing some of these [voter] rolls, we see thousands of individuals that are registered to vote that have had mail-in ballots [sent] to them that are passed away.” “We see instances were 20 to 30 ballots being mailed to the same address.”

CNN’s Kasie Hunt: “The Heritage Foundation has documented a total of 25 instances of prosecution for voter fraud where citizenship was an issue. Just 25. I mean has this ever actually affected an American election?” DHS Secretary Markwayne Mullin: “Well, 25 is 25 too many. It's kinda like one individual that dies from the hands of an illegal is one too many. It’s all preventable. One person voting illegally is one too many.” CNN’s Kasie Hunt: “Do you agree it doesn’t happen very much?” DHS Secretary Markwayne Mullin: “No, I’m not saying it doesn’t happen very much or not. If you start looking and scrubbing some of these [voter] rolls, we see thousands of individuals that are registered to vote that have had mail-in ballots [sent] to them that are passed away.” “We see instances were 20 to 30 ballots being mailed to the same address.”

RedWave Press

165,322 views • 1 month ago

Tom Holland reveals he had a long, intense talk with Robert Downey Jr. before stepping onto the set of The Odyssey. “I spoke with Robert at length before I started the movie and i was lucky that I had Anne, we had some of our first days together. I was lucky that I had Matt and Chris is an artist director, he's so helpful, he's so kind. One of the things that I love, and Matt has said this while we've been on tour that every movie he has made has led to this moment, and the same for Chris. So for me as a young actor to get to see these titans of our business use every tool that they had in the workshop to build this movie is just a treasure trove of lessons. And I really feel I left this experience a better actor”

Tom Holland reveals he had a long, intense talk with Robert Downey Jr. before stepping onto the set of The Odyssey. “I spoke with Robert at length before I started the movie and i was lucky that I had Anne, we had some of our first days together. I was lucky that I had Matt and Chris is an artist director, he's so helpful, he's so kind. One of the things that I love, and Matt has said this while we've been on tour that every movie he has made has led to this moment, and the same for Chris. So for me as a young actor to get to see these titans of our business use every tool that they had in the workshop to build this movie is just a treasure trove of lessons. And I really feel I left this experience a better actor”

👽

3,876,775 views • 8 days ago

A data center was just built where this woman lives in New Jersey, she just got her first power bill “I just got my first power bill under the new rules in which I pay for AI to plug into our power grid — Is my bill more than double the highest I've ever paid for any of the hottest years on record? Yeah I used to pay something like $220, $235. It's $600 dollars. Northern New Jersey standard usage. I mean, I use lights. I have a fan going. If you look at the usage levels on my bill, it's on par with what I did last year and the year before. It's nothing greater or less than, but it went from, like, $235 to $666. $666.39.”

A data center was just built where this woman lives in New Jersey, she just got her first power bill “I just got my first power bill under the new rules in which I pay for AI to plug into our power grid — Is my bill more than double the highest I've ever paid for any of the hottest years on record? Yeah I used to pay something like $220, $235. It's $600 dollars. Northern New Jersey standard usage. I mean, I use lights. I have a fan going. If you look at the usage levels on my bill, it's on par with what I did last year and the year before. It's nothing greater or less than, but it went from, like, $235 to $666. $666.39.”

Wall Street Apes

1,199,973 views • 11 months ago

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 views • 8 months ago

Good morning from Washington. Time for a quick tour. You know what you won’t find in my office? A desk. Or a chief of staff. We have an open office with all day interaction and collaboration between my employees. As for the chief of staff? I tried having a single point of failure before. Didn’t love it. Now our four directors report directly to me. Our office works faster, more efficient and better than it ever has under this set up. Everyone in the office has a voice and they are encouraged to use it. Our entire office is on the same page with the way we serve our constituents. And we love it. Now

Good morning from Washington. Time for a quick tour. You know what you won’t find in my office? A desk. Or a chief of staff. We have an open office with all day interaction and collaboration between my employees. As for the chief of staff? I tried having a single point of failure before. Didn’t love it. Now our four directors report directly to me. Our office works faster, more efficient and better than it ever has under this set up. Everyone in the office has a voice and they are encouraged to use it. Our entire office is on the same page with the way we serve our constituents. And we love it. Now

Rep. Nancy Mace

544,697 views • 5 months ago

DeepSeek R1 671B is ridiculously good, but running at 158 tokens per second is something out of this world. A model *this good* running *this fast* is 3 years ahead of this time. The secret: DeepSeek in this short video isn't running on GPUs. It's running on custom-made chips! This is running on SambaNova and their SN40L chips, which are designed and optimized for running AI workflows. SambaNova is sponsoring my work. We designed GPUs to run games and later realized we could use them for AI. This is very different from creating a chip specifically for AI from the ground up. Just think about this for a second: One single SNL40 chip can simultaneously hold 100+ models (trillions of parameters) in memory! An agentic workflow that uses multiple models simultaneously won't need to swap models from memory even once! By the way, the Llama family of models *screams* on the SambaNova Cloud. Here are some numbers from my latest tests: • Llama 3.3 70B: 367.32 t/s • Llama 3.2 1B: 2381.82 t/s • Llama 3.2 3B: 1335.71 t/s • Llama 3.1 8B: 929.58 t/s • Llama 3.1 70B: 369.16 t/s • Llama 3.1 405B: 103.37 t/s You can access all of these models for free!

DeepSeek R1 671B is ridiculously good, but running at 158 tokens per second is something out of this world. A model this good running this fast is 3 years ahead of this time. The secret: DeepSeek in this short video isn't running on GPUs. It's running on custom-made chips! This is running on SambaNova and their SN40L chips, which are designed and optimized for running AI workflows. SambaNova is sponsoring my work. We designed GPUs to run games and later realized we could use them for AI. This is very different from creating a chip specifically for AI from the ground up. Just think about this for a second: One single SNL40 chip can simultaneously hold 100+ models (trillions of parameters) in memory! An agentic workflow that uses multiple models simultaneously won't need to swap models from memory even once! By the way, the Llama family of models screams on the SambaNova Cloud. Here are some numbers from my latest tests: • Llama 3.3 70B: 367.32 t/s • Llama 3.2 1B: 2381.82 t/s • Llama 3.2 3B: 1335.71 t/s • Llama 3.1 8B: 929.58 t/s • Llama 3.1 70B: 369.16 t/s • Llama 3.1 405B: 103.37 t/s You can access all of these models for free!

Santiago

131,019 views • 1 year ago

Not sure if I'm the only one but I think WHOOP isn't very good tracking lifting weights It generally doesn't auto detect weight lifting (instead showing it as ACTIVITY) and then rates it as low strain compared to cardio, probably because it's lower BPM But that doesn't really make sense, I can do an extremely heavy lifting workout that hurts for days after but WHOOP will show a strain of 6 Then I go for a light bicycle ride and it's a strain of 12, simply because it gets my heart BPM up This problem is confirmed to me by people at WHOOP but since it's so hard to measure the strain of lifting with their sensors they prefer to not estimate lifting too heavy, instead they want people to write down their specific lifting workouts in the WHOOP app everytime But like nobody has time for that Also since it never auto detects weight lifting (eventhough it's the only workout type I ever set, so it should just default), I have to go back to every single workout in the app of the last couple of days to manually change it to Weight Lifting which is tedious (especially with that link workout popup everytime + the calendar resetting to today straight after) I think WHOOP is made for runners but they forgot people also need to lift weights and it has a bias to cardio This actually has been influencing a lot of my friends to run more which is good but I still think it's a massive bias as strength training is super important for longevity too Will Ahmed Viviano

Not sure if I'm the only one but I think WHOOP isn't very good tracking lifting weights It generally doesn't auto detect weight lifting (instead showing it as ACTIVITY) and then rates it as low strain compared to cardio, probably because it's lower BPM But that doesn't really make sense, I can do an extremely heavy lifting workout that hurts for days after but WHOOP will show a strain of 6 Then I go for a light bicycle ride and it's a strain of 12, simply because it gets my heart BPM up This problem is confirmed to me by people at WHOOP but since it's so hard to measure the strain of lifting with their sensors they prefer to not estimate lifting too heavy, instead they want people to write down their specific lifting workouts in the WHOOP app everytime But like nobody has time for that Also since it never auto detects weight lifting (eventhough it's the only workout type I ever set, so it should just default), I have to go back to every single workout in the app of the last couple of days to manually change it to Weight Lifting which is tedious (especially with that link workout popup everytime + the calendar resetting to today straight after) I think WHOOP is made for runners but they forgot people also need to lift weights and it has a bias to cardio This actually has been influencing a lot of my friends to run more which is good but I still think it's a massive bias as strength training is super important for longevity too Will Ahmed Viviano

@levelsio

148,931 views • 9 months ago

For more than a year, I've poured my heart into creating my Alice in Wonderland Themed Airbnb. It's taken countless hours, plenty of setbacks, and more hard work than I ever imagined. There are still days when it feels like the finish line is a long way off, but when I look back at where it started, I can't help but feel proud. This place has become so much more than a project, it's a dream I've held onto and slowly brought to life, one piece at a time. Every detail reflects a little bit of my imagination and a lot of my heart. I know it's not finished yet, but seeing it come together reminds me that the best things are worth the time, patience, and love they require. I can't wait for the day I finally open the door and share this little Wonderland with the world. ❤️

For more than a year, I've poured my heart into creating my Alice in Wonderland Themed Airbnb. It's taken countless hours, plenty of setbacks, and more hard work than I ever imagined. There are still days when it feels like the finish line is a long way off, but when I look back at where it started, I can't help but feel proud. This place has become so much more than a project, it's a dream I've held onto and slowly brought to life, one piece at a time. Every detail reflects a little bit of my imagination and a lot of my heart. I know it's not finished yet, but seeing it come together reminds me that the best things are worth the time, patience, and love they require. I can't wait for the day I finally open the door and share this little Wonderland with the world. ❤️

Kat Stryker

18,023 views • 11 days ago