You are viewing a single comment's thread from:

RE: LeoThread 2024-10-15 10:56

in LeoFinance2 months ago

Here is the daily technology #threadcast for 10/15/24. We aim to educate people about this crucial area along with providing information of what is taking place.

Drop all question, comments, and articles relating to #technology and the future. The goal is make it a technology center.

Sort:  
There are 3 pages
Pages

I certainly think there are some skills we'll lose as we hand things over to automation. I can barely remember my own phone number now, let alone the long list of numbers I used to know, and my handwriting has completely gone to pot. Hannah Fry
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

NASA has successfully launched its alien-hunting mission to Jupiter's fourth largest moon that could host the ingredients for life.

The Europa Clipper spacecraft took off from Kennedy Space Center in Florida at 12:08pm ET, tucked inside a SpaceX Falcon rocket.

The $5.2 billion probe is set to travel 1.8 billion-mile to Europa, meeting the target in April 2030, and perform more than 40 flybys to look for 'conditions suitable for life.'

DJI Air 3S review: LiDAR and improved image quality make for a nearly faultless drone

DJI just announced the dual-camera Air 3S drone and there’s some all-new cutting-edge tech hiding in the nose. A LiDAR sensor is there to provide extra crash protection at night, a time that’s often dangerous for drones.

#technology #drone

The Air 3S also has a new main camera with a larger sensor better suited for capturing video in low-light. And it now comes with the company’s ActiveTrack 360, which it first introduced in the Mini 4 Pro, allowing the device to zoom all around your subject while tracking and filming them.

There are a bunch of other little improvements, from storage to the new panoramic photo mode, all at the same $1,099 price as the Air 3 was at launch. To see if it lives up to DJI’s hype, I tested all of its new capabilities, especially the new night-flying obstacle avoidance.

Design

The Air 3S looks a lot like the Air 3 and weighs just 0.1 ounces more at 1.6 pounds. Since it’s over 250 grams, you need a permit to fly it in the US and elsewhere. Unlike the Avata 2 or Neo, the Air 3S is obviously not suited to fly around people, but DJI does offer propeller guards for a bit of extra safety.

The Air 3S’ camera module is larger as well because it holds a new 50-megapixel 1-inch sensor camera with a 24mm f/1.8 lens, up from from the previous model’s 1/1.3-inch sensor. This setup supports 4K recording up to 120 fps and 1080p at 240 fps, along with D-LogM to improve dynamic range. DJI also boosted the maximum ISO to 12,800 for improved light sensitivity. The secondary 70mm (3x) zoom camera is unchanged with a 1/1.3-inch sensor, and it supports the same video specs and color modes as the primary.

The Air 3S is DJI’s latest drone with its ActiveTrack 360 system. Once you start tracking a subject, you can change the camera angle using an onscreen control wheel. The drone will then figure out the best way to change positions while also avoiding obstacles. It does so using six vision (camera) sensors, two downward facing infrared time-of-flight sensors and the new LiDAR.

Another key change on the Air 3S is its 42GB of internal storage, up from 8GB on the Air 3 — something that’ll be handy if you forget a microSD card or fill one quickly. You can transfer files to your PC either by a Wi-Fi connection or the drone’s USB-C port. The wired method now works even if the drone is turned off.

The Air 3S can hit up to 45 mph in sport mode. That’s fast and there’s no obstacle detection with this setting, so you need to know what you’re doing. In normal mode, which does support obstacle tracking, speeds drop down to 23 mph. You get about 12.5 miles of range in the US via DJI’s latest O4 transmission system, or half that in Europe. One small change is that the Air 3S now transmits 10-bit 1080p 60p video rather than 8-bit as before, improving remote video capture quality.

Battery life on the Air 3S is excellent at 45 minutes, which equates to about 38 minutes under working conditions. With the optional Fly More kit that includes three batteries and a charging hub, pilots should be able to fly most of the day. Like the Mini 4 Pro, the charging hub supports power accumulation, so you can transfer energy from the weakest batteries to the strongest for longer flights.

Finally, the Air 3S appears to have the same propeller system as the Air 3 and emits about the same amount of noise. It’s pretty quiet for such a large drone, and once it gets up to around 150 feet, you can barely hear it.

Performance

Steve Dent for Engadget

I tested the ActiveTrack 360 function while walking, biking and driving a car to see how it worked at various speeds. When tracking my car at around 20 to 25 mph, it stayed locked on while easily avoiding trees and power lines.

I found a bike trail among some widely spaced trees and another with denser foliage. In the first section, the drone weaved smoothly around obstacles while keeping the bike in frame. After a few minutes, I was confident it wouldn’t crash and was able to focus on creating some cool shots. The drone occasionally hesitated when it encountered obstacles, which spoiled some footage. However, it never crashed or lost track of the bike.

Things weren’t quite as easy on the dense brush trail. When I changed the camera angle, the drone would occasionally pick a bad route, get hemmed in by branches and stop. A few times the propellers knicked leaves or branches, throwing the drone off keel. It never crashed or took any damage, but I replaced the propellers out of an abundance of caution.

That experience taught me that the Air 3S’s sensors can miss small obstacles, so it’s wise to stay away from those situations and use propeller guards if you can’t avoid them. You also need to make sure the drone has enough room to maneuver when using ActiveTrack 360 to change camera angles.

Steve Dent for Engadget

Later, I took the Air 3S out at night to test the LiDAR, flying it up around trees and next to buildings. Previously, this would have been very risky, but the Air 3S successfully detected and avoided all obstacles. This new low-light capability will be especially helpful when you need to bring the drone home in the dark. Keep in mind that the LiDAR only works in the forward direction, so avoid going backwards or sideways at night.

Video quality

Steve Dent for Engadget

To test the video quality, I shot aerials with the Air 3S, at dawn, in the daylight and at night. With the larger 1-inch sensor, 4K video is more detailed than ever and the camera produces vibrant, accurate colors and nice skin tones. It also does a great job in variable light conditions when using the DLogM setting. That came in handy when I was biking in the forest with a mixture of dappled sunlight and shadows. The 4K 120 fps setting is nice for creating dramatic shots with moving subjects, and there’s no discernible loss in quality compared to the 30p or 60p modes.

Elon Musk Unveils Robotaxi - "We, Robot" Breakdown

Summary below ⏬

Tesla's 2024 AI Day: Unveiling the Future of Transportation and Robotics

At Tesla's recent AI Day event, CEO Elon Musk unveiled a series of groundbreaking announcements that could reshape the future of transportation and robotics. The event, which took place on the Warner Bros. Studios lot in Hollywood, showcased Tesla's vision for autonomous vehicles and humanoid robots in a spectacle that blended cinematic flair with technological innovation.

The Robo Taxi's unveiling was accompanied by an impressive demonstration. Tesla had transformed a portion of the Warner Bros. lot into a mock city, complete with 20 Robo Taxis and additional Model 3 and Model Y vehicles, all operating autonomously. The vehicles navigated the simulated urban environment, showcasing their ability to interact seamlessly with each other and their surroundings.

One of the most intriguing aspects of the Robo Taxi is its potential to revolutionize the economics of transportation. Musk projected that the operating cost of these vehicles could be as low as 20 cents per mile, significantly undercutting traditional public transit options like buses, which he estimated at around a dollar per mile. This cost-effectiveness could make individualized mass transit a reality, potentially reshaping urban mobility.

The Roven: Scaling Up Autonomous Transport

In a surprise announcement, Musk also unveiled the "Roven," a larger autonomous vehicle capable of transporting up to 20 people or a significant amount of cargo. The Roven's design is a striking embodiment of Art Deco aesthetics, resembling futuristic trains from the 1930s. This vehicle is positioned as a solution for high-density transport needs, such as moving sports teams or large groups of people, with Musk suggesting it could potentially bring transportation costs down to as low as 5-10 cents per mile.

Optimus: The Humanoid Robot

The event wasn't limited to vehicular innovations. Musk also provided updates on Optimus, Tesla's humanoid robot project. In a dramatic demonstration of progress, Optimus robots were shown walking around and interacting with attendees at the event's afterparty, serving drinks and showcasing their capabilities. This was a far cry from the initial Optimus announcement just a few years ago, which featured a person in a robot costume.

Musk made bold predictions about the future of Optimus, suggesting that it could become "the highest selling product of all time" and "the most important product of all time." He projected that in the long term, an Optimus robot could be available for purchase at a price point of $20,000 to $30,000, making it accessible to a wide range of consumers and businesses.

The Economics of Tesla's Vision

A crucial aspect of Musk's presentation was the economic model underlying these innovations. He envisions a future where individuals can purchase a Robo Taxi for around $20,000 to $30,000 - comparable to the price of a standard new car today. However, unlike traditional vehicles that sit idle for about 90% of the time, Robo Taxis could be put to work when not in use by their owners, potentially earning money by providing rides or running errands for others.

Furthermore, Musk revealed plans to "over-spec" the Robo Taxis with computing power beyond what's necessary for autonomous driving. This excess capacity could be leveraged to offer cloud computing services, similar to Amazon's AWS, during the vehicle's downtime. This innovative approach aims to extract maximum value from each vehicle, turning them into mobile computing platforms when not in use for transportation.

Looking Ahead: Challenges and Opportunities

Despite the excitement generated by these announcements, the market response was surprisingly muted, with Tesla's stock price dropping nearly 8% the day after the event. This reaction might reflect skepticism about the ambitious timelines Musk presented or concerns about the regulatory and technical challenges that lie ahead.

Musk acknowledged his tendency to be optimistic with timelines but remained confident in Tesla's ability to deliver. He projected that unsupervised full self-driving capabilities would be available in Texas and California by next year, with Robo Taxi production beginning in 2026 or 2027.

While Tesla isn't the first to venture into autonomous taxi services - companies like Waymo have been operating in this space for years - the scale and integration of Tesla's vision set it apart. By combining autonomous vehicles, humanoid robots, and innovative economic models, Tesla is positioning itself at the forefront of a potential revolution in transportation and robotics.

As we look to the future, it's clear that Tesla's AI Day announcements represent more than just new products - they offer a glimpse into a possible future where autonomous vehicles and robots are integral parts of our daily lives. While challenges remain, the ambition and scale of Tesla's vision suggest that significant changes in transportation and robotics may be closer than we think. As always with Tesla and Elon Musk, the journey from announcement to reality will be closely watched by enthusiasts, skeptics, and industry observers alike.

JLab's $30 Go Pop ANC are its most affordable noise-canceling earbuds ever

JLab blew me away in early 2020 when the company debuted its $29 Go Air earbuds. For the first time, I realized that I had to pay closer attention to budget earbuds because some companies manage to build impressive products for less than $50. Do they offer flagship level performance and features? No. Do they cover the basics and sound good enough for casual use? Absolutely.

The company has announced several products since, including the $60 JBuds ANC (third-gen) earlier this year. It appears that JLab wasn't content with that model being its most affordable noise-canceling option. Today, the company revealed the Go Pop ANC ($30), a set of earbuds that still offer some noise-blocking abilities for half the price of the JBuds ANC. This new model is the company's most affordable noise-canceling set ever, and JLab even says it's lowest-priced set of active noise cancellation (ANC) earbuds you can buy right now at full price.

JLab says you can expect over six hours of listening time on the earbuds (ANC on), maxing out at over 24 hours when you factor in the case (ANC off). Like previous JLab models, the Go Pop ANC case has an integrated USB-C cable, so there's always have a way to plug in. Additionally, a fast charge feature gives you an hour of use in 10 minutes.

The IP55 rated earbuds are sweat and splash proof, offering more robust ingress protection than some flagship models that cost a 10 times as much. Bluetooth multipoint connectivity is here, as is the ability to customize the audio and more via the JLab app. Specifically, that piece of software will allow you to adjust the EQ and reconfigure the onboard controls, plus there's a safe hearing tool and dedicated modes for music and movies.

To accompany the Go Pop ANC's noise-cancelling functionality, there's a Be Aware transparency mode when you need it. 10mm dynamic drivers handle the sound and the two ANC MEMS microphones enable "clear" calls. The Go Pop ANC is also Google Fast Pair compatible as well, and it's available in fuchsia and teal color options in addition to the black that's pictured.

The iPad mini 7 goes big on Apple Intelligence

Meet Apple’s seventh-generation iPad mini. Arriving over three years after its predecessor, the new tablet adds a faster A17 Pro chip originally found in the iPhone 15 Pro series with support for Apple Intelligence, more entry-level storage and support for the Apple Pencil Pro.

The iPad mini 7 has the same 8.3-inch screen size as the previous model and an identical 2266 x 1488 resolution at 326 ppi. Unfortunately, the refresh rate remains a modest 60Hz; from what we can tell, the display is identical to the one in its predecessor. It’s dimensions remain unchanged as well. The front-facing camera is still in the portrait position, something that makes sense for a device of this size, and both the front- and back-facing cameras appear unchanged from the previous iPad mini.

Without any significant design upgrades, this update is more about what’s inside. The A17 Pro chip lets the new tablet use Apple Intelligence features like smarter Siri, Writing Tools, Image Playground and Genmoji. It also has more RAM and 128GB of storage in the base-level model, double that of the 2021 model. There's also a 512GB storage option for the first time.

The new iPad mini arrives on October 23 and starts at $499 for 128GB storage.

Update, October 15 2024, 9:40AM ET: This post originally said that the new iPad mini used the A18 chip; it has been updated to note that it uses the A17 Pro chip. We apologize for the error.

Scientists Use Human Stem Cells to Restore a Monkey’s Vision

Scientists successfully used human stem cells to patch a hole in a monkey's retina, restoring the primate's vision. Macular holes, a condition associated with aging, cause vision to blur and decline over time. Current fixes for the condition come at the cost of a loss of peripheral vision as they involve transplanting cells. Using stem cells to repair the problem would introduce new cells, avoiding this issue. More research is needed to determine whether the treatment is suitable for human eyes.

#technology #health #stemcells

Hey did you see the tesla robot car? And Van?

It was okay. Personally, I am not as high on the initiative as most, at least on the surface. To me, autonomous technology is going to be huge for transportation but for cargo more than people. Most miles are not fares for individuals. They are either in their car or someone they know.

Cargo, 100% of that is transported by someone other than the ultimate owner (end user).

Tesla robo-trucking is much bigger in my opinion. A new semi plant being built.

As for the robo-van, that interest me but again, configured for cargo. Think of that as a replacement for all the Amazon and grocery delivery service.

Oh I didn't even realize Tesla had semi plants already in the works. Man those truck drivers have it coming!

I hope they start saving to buy their own robo semi to rent out if that's a possibility

That shows how little the media covers Tesla. Yeah they are going to be the leading semi building in the world within a few years. The Robo aspect is a bonus.

Google Backs New Nuclear Plants to Power AI

Big Tech companies are increasingly investing into nuclear power. Google has committed to buying power generated by seven nuclear-power reactors to be built by Kairos Power. The project aims to add 500 megawatts of nuclear power starting at the end of the decade. The project site (or sites) have yet to be determined. Google will have its data centers somewhere near the reactors, but it hasn't been determined whether they will receive power directly from the plants or from the grid.

#technology #nuclearpower #energy #ai #google

It is interesting to see how the AI race is fueling a focus upon massive energy generation. This is necessary derivative for things to move forward at the pace they are.

SUper interesting. I think AI will have a ajor impact on the energy landscape going forward...

Maybe it could finally help us get to viable nuclear fusion

There is a lot of focus upon newer energy forms. Not sure fusion is ready, or will be in time. Even if solved today, takes a long time to build.

Solar-powered desalination system requires no extra batteries

MIT engineers have built a solar-powered desalination system that removes salt from water at a pace that closely follows changes in solar energy. It automatically adjusts to any sudden variation in sunlight, maximizing the utility of solar energy. The system can produce large quantities of clean water regardless of sunlight variation throughout the day without any batteries or supplemental power supply. It is geared towards desalinating brackish groundwater, a huge untapped source of potential drinking water.

#technology #solarpower #desalination #mit

OpenAI Unveils Secret Meta Prompt—And It’s Very Different From Anthropic's Approach

OpenAI has revealed the meta-prompt for its new o1 model family to help developers improve and fine-tune their products. Meta-prompts are used to instruct models to create good prompts based on user task descriptions or to improve existing prompts. This article looks at both OpenAI and Anthropic's approaches and compares them. The companies are using very different engines and their prompts reveal a lot about how each company thinks and what they expect from their products. OpenAI conceptualizes ChatGPT as a powerful computational tool while Anthropic envisions Claude as a friendly, human-like assistant.

#technology #ai #openai #anthropic #artificialintelligence

Terence Tao, mathematician: ‘It’s not good for something as important as AI to be a monopoly held by one or two companies’

This article contains an interview with Terence Tao, indisputably one of the best mathematicians in history. At 49, Tao, who was born in the Australian city of Adelaide, received his doctorate from Princeton University at 21 and was already a professor at the University of California in Los Angeles at 24. In the interview, Tao discusses the results from the recent presidential election in Venezuela, whether AI is a threat that could destroy humanity, AI systems like AlphaProof and AlphaGeometry, whether water could spontaneously explode and destroy the world, and more.

#technology #terencetao #ai #artificialintelligence #monopoly

20VC closes new $400M fund to 'make Europe great again,' says Harry Stebbings

Harry Stebbings, the U.K. podcaster who broke into the world of tech with his 20-minute interviews of venture capitalists and founders,

Harry Stebbings, the U.K. podcaster who broke into the world of tech with his 20-minute interviews of venture capitalists and founders, parlayed that fame into becoming a VC himself. Now, Stebbings has closed his third investment vehicle, and it’s his biggest yet: 20VC, the firm named after the podcast series, has closed a $400 million fund.

#newsonleo #technology

20VC's New Fund: A Focus on European Startups

20VC, a venture capital firm founded by Steve Stebbings, has raised $400 million to invest in startups across Europe. The fund will allocate $125 million to seed investments and $275 million to Series A rounds. This is a significant move, considering the current climate for founders, where funding has become increasingly challenging.

Steve Stebbings: The Face of 20VC

Steve Stebbings is the founder of 20VC, a venture capital firm that has gained significant attention for its unique approach to investing. Stebbings is not a technologist by training, having studied law at university before dropping out to start his firm. However, he has built a reputation as an online creator and content expert, leveraging his influence to open doors and secure term sheets for his fund.

A Personal Approach to Investing

Stebbings' approach to investing is highly personal, focusing on identifying great entrepreneurs and founders who have the potential to shape markets. He believes that "great founders shape markets" and that his job is to find the best founders before anyone else. This approach has helped 20VC achieve significant success, with over £10 million in revenue generated.

Operational Experience

One of the key differentiators of 20VC is its operational experience. the Firm has established "sub-funds" in categories like sales, product, and growth, which are run by teams with operator experience. This approach has helped 20VC achieve significant revenue and profitability, making it a sustainable business.

A Strong Presence in Europe

20VC has a strong presence in Europe, with a network of contacts and connections that will help identify and invest in promising startups. Stebbings credits his ability to connect with European founders, who are eager to collaborate with US investors. He has built a reputation as a trusted and influential figure in the European tech scene, with a strong track record of success.

A market Correction?

Stebbings believes that the current state of venture capital, with only 1% of companies generating 90% of the gains, is actually a market correction. He predicts that for 1% of firms, venture returns will be much larger and better than ever, as the size of the outcomes increases. This could spell more opportunity for big wins, rather than less, in the long run.

The Future of European Tech

20VC's new fund is a significant step towards addressing the investment gap between European and US technology companies. With Stebbings at the helm, the firm is poised to make a meaningful impact on the European tech scene. As Stebbings said, "We need to make Europe great again. MEGA!" With his unique approach and personal touch, 20VC is well-positioned to help European startups achieve success on the global stage.

Key Takeaways

  • 20VC has raised $400 million to invest in startups across Europe.
  • The fund will allocate $125 million to seed investments and $275 million to Series A rounds.
  • Steve Stebbings is the founder of 20VC, with a reputation as an online creator and content expert.
  • 20VC's approach to investing is highly personal, focusing on identifying great entrepreneurs and founders.
  • The firm has a strong presence in Europe, with a network of contacts and connections that will help identify and invest in promising startups.
  • 20VC believes that the current state of venture capital is a market correction, with more opportunity for big wins in the long run.

Spotify is adding music video support in 85 additional markets

Music streaming company Spotify said today that it is testing music videos for premium users in 85 additional markets.

Spotify Expands Music video Testing to 85 New Markets, Aiming to Compete with YouTube

In a significant move to enhance its music streaming experience and compete with YouTube, Spotify has announced that it is testing music videos for premium users in 85 additional markets.

#spotify #music #newsonleo

This expansion comes after the company initially experimented with music videos in limited markets in March, including the UK, Germany, Italy, the Netherlands, Poland, Sweden, Brazil, Colombia, the Philippines, Indonesia, and Kenya, with Egypt later being added to the list. However, it remains unclear if the 85 new markets include the US, a crucial market for the streaming giant.

The music video feature is currently available to paid users on both mobile and desktop devices, allowing them to easily switch between audio and video playback using a toggle. The company has also made the transition more seamless by enabling videos to start playing from where the user left off. To make music videos more discoverable, Spotify is introducing video indicators next to track names in search results, allowing users to easily identify music videos and switch to the video mode. Additionally, users can switch to full-screen video mode by turning their phone to landscape orientation.

Spotify's music video feature is designed to enhance the overall music streaming experience and encourage users to engage more with the content. According to the company, listeners who discover a song and watch the music video on the app are 34% more likely, on average, to stream the song again the following week. By offering music videos, Spotify is attempting to attract music fans who have traditionally used YouTube as their primary source for music videos. YouTube's native support for music videos has been a major draw for many users, and Spotify is hoping to capitalize on this trend by offering a similar experience.

The expansion of music videos to more markets is a significant move for Spotify, as it aims to differentiate itself from other music streaming services and attract a larger share of the market. With its vast library of songs and increasing focus on video content, Spotify is well-positioned to become a major player in the music streaming industry.

Key Takeaways:

  • Spotify is testing music videos for premium users in 85 additional markets, expanding its initial test in 11 countries.
  • The music video feature is currently limited to paid users and is available on mobile and desktop devices.
  • Users can easily switch between audio and video playback using a toggle and videos will start playing from where the user left off.
  • Video indicators will be displayed next to track names in search results to make music videos more discoverable.
  • Users can switch to full-screen video mode by turning their phone to landscape orientation.
  • Spotify's music video feature aims to enhance the overall music streaming experience and encourage users to engage more with the content.
  • The company claims that listeners who discover a song and watch the music video on the app are 34% more likely to stream the song again the following week.
  • The expansion of music videos to more markets is a significant move for Spotify, as it aims to differentiate itself from other music streaming services and attract a larger share of the market.

What SpaceX and Tesla TEACH US about CIVILIZATIONS

What SpaceX and Tesla TEACH US about CIVILIZATIONS

On October 13th, 2024, Jo Bhakdi kicked off his Civilization Series with a video discussing what we can learn from the greatest pioneers like Elon Musk to create a new civilization. The video was inspired by SpaceX's recent successful launch and attempted booster landing, which showcased incredible engineering prowess and a stark contrast between SpaceX and NASA's approaches to space exploration.

Bhakdi argues that we can learn valuable lessons from SpaceX's success in building a new civilization. First, building something from scratch, as SpaceX did, is much easier than trying to fix a broken system like NASA. Second, great things are always conceived and designed by a handful of individuals, not large organizations. Third, we need first principles to guide our efforts and create a clear vision for the future. Fourth, first principles allow us to achieve focus and concentration on a worthwhile goal. Finally, we should commit our lives to building something truly meaningful, something that will benefit future generations.

Bhakdi believes that we can use these lessons to build a new civilization that is worth living in. He outlines four key steps: defining the first principles of civilization, deriving the central pillars of the target state, creating a master plan to build those pillars, and discussing the governance, defense, and mind viruses that could threaten our civilization.

Bhakdi concludes by emphasizing the urgency of this task. He believes that we are in a critical situation and that nOW is the time to build a new civilization. He calls on viewers to join him in this effort and to help engineer a better future for humanity.

This video is a thought-provoking and inspiring look at the challenges and opportunities facing our civilization. It is a must-watch for anyone interested in the future of humanity.

Xscape is building multicolor lasers to connect chips within datacenters

Xscape, a startup founded by Columbia professors, is developing lasers to drive the fiber-optic interconnects in datacenters.

The GPUs and other chips used to train AI communicate with each other inside datacenters through “interconnects.” But those interconnects have limited bandwidth, which limits AI training performance. A 2022 survey found that AI developers typically struggle to use more than 25% of a GPU’s capacity.

#newsonleo #xscape #lasers #technology

Interconnects: The Communication Network of Datacenters

In modern datacenters, GPUs and other chips are connected through high-speed interconnects, which enable them to exchange data and instructions. These interconnects are designed to facilitate communication between the various components of the datacenter, including:

  1. GPUs: Graphics Processing Units are specialized chips designed for parallel processing of graphics and compute-intensive tasks. In AI training, GPUs are used to perform matrix multiplications, convolutions, and other compute-intensive operations.
  2. CPUs: Central Processing Units are general-purpose processors that handle tasks such as data transfer, memory management, and control flow.
  3. Memory: Memory systems, including DRAM and registers, provide a buffer for data and instructions being transferred between components.
  4. Storage: Storage systems, including hard drives and solid-state drives, provide a permanent storage for data and models.

Interconnects can be classified into several categories, including:

  1. PCIe (Peripheral Component Interconnect Express): A high-speed interconnect standard used to connect GPUs, CPUs, and other components to the motherboard.
  2. InfiniBand: A high-speed interconnect standard used to connect clusters of nodes in a datacenter.
  3. NVLink: A high-speed interconnect standard used to connect GPUs and other components in a datacenter.

Limited Bandwidth of Interconnects

The limited bandwidth of interconnects is a significant bottleneck in AI training. As AI models become increasingly complex, they require more data to be processed and transmitted between GPUs and other chips. However, the interconnects can only handle a certain amount of data per second, which limits the overall performance of the AI training process.

For example, a high-end GPU like the NVIDIA V100 can process up to 15.7 teraflops of compute power, but the interconnects can only transfer data at a rate of around 100-200 GB/s. This means that the GPU can perform many more computations than the interconnects can transfer data, resulting in a bottleneck that limits the overall performance of the AI training process.

GPU Aggregation

One potential solution to the limitations of individual GPUs is GPU aggregation. GPU aggregation involves grouping multiple GPUs together to form a single, more powerful unit. This can be achieved through various means, including:

  1. Direct Liquid Cooling (DLC): A system that uses liquid cooling to cool multiple GPUs, allowing them to operate at higher temperatures and perform more computations.
  2. Multi-GPU Systems: A system that uses multiple GPUs, each with its own memory and compute resources, to form a single, more powerful unit.
  3. GPU Clustering: A system that groups multiple GPUs together to form a single, more powerful unit, using techniques such as PCIe switching and NVLink.

GPU aggregation can help to overcome the limitations of individual GPUs by providing more compute resources and increased memory bandwidth. However, it also introduces additional complexity and cost, making it a challenging solution to implement.

Optimized Data Transfer

Another potential solution to the limitations of interconnects is optimized data transfer. Optimized data transfer involves using techniques such as:

  1. Data Compression: Compressing data to reduce the amount of data that needs to be transferred.
  2. Data Parallelism: Breaking down large datasets into smaller chunks and transferring them in parallel.
  3. Asynchronous Data Transfer: Transferring data asynchronously, allowing other components to continue processing while waiting for data to be transferred.

Optimized data transfer can help to reduce the bandwidth requirements of AI training by reducing the amount of data that needs to be transferred and allowing other components to continue processing while waiting for data to be transferred.

New Architectures

Finally, new architectures are being developed to address the limitations of current technologies. Some potential examples include:

  1. Tensor Processing Units (TPUs): Specialized chips designed for parallel processing of matrix operations, commonly used in AI training.
  2. Field-Programmable Gate Arrays (FPGAs): Chips that can be programmed to perform specific tasks, such as matrix operations, using field-programmable logic.
  3. Neuromorphic Chips: Chips that mimic the structure and function of the human brain, designed to perform tasks that are similar to those of neural networks.

New architectures can help to overcome the limitations of current technologies by providing more efficient and scalable solutions for AI training. However, they also introduce additional complexity and cost, making them a challenging solution to implement.

Conclusion

The limitations of interconnects and GPUs are significant bottlenecks in AI training, which can lead to reduced training speed, increased power consumption, and decreased model accuracy. However, researchers and developers are exploring various solutions and innovations to address these challenges, including GPU aggregation, optimized data transfer, and new architectures. By addressing these challenges, we can unlock the full potential of AI and continue to drive the development of more powerful and accurate AI models.

Amazon is using tech from a Khosla Ventures-backed startup to run robot warehouses at Whole Foods

Fulfil's technology will enable shoppers to purchase items that aren't stocked at Whole Foods, like Coca Cola soda and Kellogg's cereals.

#amazon #newsonleo #robots #wholefoods #technology

Fulfil's Technology

Fulfil's technology is based on warehouse automation, which involves the use of robots to perform tasks such as fetching and ferrying items from shelves to bags for pickup. The company's robots are designed to navigate the warehouse environment efficiently and effectively, using a combination of sensors and mapping technology to avoid collisions and optimize their routes.

In the pilot project, Fulfil's robots will be used to fetch and ferry items from shelves to bags for pickup. The robots will be equipped with grocery bags and will be able to handle a variety of items, including canned goods, coffee pods, and other staples. The system will enable Amazon to "eliminate those extra trips" shoppers take to other grocery stores, according to Tony Hoggett, who leads Amazon's worldwide grocery business.

How the System Works

The system will work as follows:

  1. Shoppers will browse the store using a smartphone app or tablet, and select the items they want to purchase online.
  2. The selected items will be added to their virtual shopping cart, which will be linked to their Amazon account.
  1. When the shopper arrives at the store, they will pick up their virtual cart and begin checking out.
  2. As they check out, the robots will fetch and ferry the items from shelves to bags for pickup.
  3. The shopper will then take their bags and leave the store, without having to wait in line or interact with cashiers.

Benefits of the System

The system will provide several benefits to shoppers and Amazon, including:

  • Convenience: Shoppers will be able to browse and purchase items online, and then pick them up in the store without having to wait in line or interact with cashiers.
  • Increased Efficiency: The system will reduce the time it takes for shoppers to checkout, making the shopping experience faster and more efficient.
  • Improved customer Experience: Shoppers will be able to get the items they need quickly and easily, without having to interact with cashiers or wait in line.
  • Increased Sales: The system will enable Amazon to offer a wider range of products to shoppers, including items that are not typically stocked by Whole foods.

Challenges and Limitations

While the system has the potential to revolutionize the way shoppers interact with Amazon and Whole Foods, there are also several challenges and limitations that need to be addressed. These include:

  • Integration with Existing Systems: The system will require integration with existing systems, including Amazon's online grocery platform and Whole Foods' store operations.
  • Logistics and Supply Chain: The system will require a robust logistics and supply chain infrastructure to ensure that items are delivered to the store in a timely and efficient manner.
  • Scalability: The system will need to be scalable to accommodate the growing demand for online grocery shopping.
  • Cost: The system will require significant investment to develop and implement, and may require changes to Amazon's business model to ensure that it is profitable.

Future Plans

Amazon and Fulfil are planning to expand the pilot project to other locations, including other Whole Foods stores and potentially other Amazon grocery stores. The company is also exploring the use of Fulfil's technology in other areas, including Amazon's warehouses and distribution centers.

Overall, the partnership between Amazon and Fulfil represents a significant step forward in the company's efforts to revolutionize the way it operates its physical stores, including Whole Foods. By integrating automated warehouse technology into its supermarkets, Amazon is well-positioned to take advantage of the growing demand for online grocery shopping and to provide customers with a more convenient and efficient shopping experience.

The first rule of any technology used in a business is that automation applied to an efficient operation will magnify the efficiency. The second is that automation applied to an inefficient operation will magnify the inefficiency.

Bill Gates

In the long term, artificial intelligence and automation are going to be taking over so much of what gives humans a feeling of purpose. Matt Bellamy
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

There's a lot of automation that can happen that isn't a replacement of humans but of mind-numbing behavior. Stewart Butterfield
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Automation is good, so long as you know exactly where to put the machine. Eliyahu Goldratt
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Automation is good, so long as you know exactly where to put the machine. Eliyahu Goldratt
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Automation is driving the decline of banal and repetitive tasks. Amber Rudd
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

The way you deal with automation is by upgrading people's skills so they can get the jobs of the future. John Delaney
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Whatever you are studying right now, if you are not getting up to speed on deep learning, neural networks, etc., you lose. We are going through the process where software will automate software, automation will automate automation. Mark Cuban
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Automation is no longer just a problem for those working in manufacturing. Physical labor was replaced by robots; mental labor is going to be replaced by AI and software. Andrew Yang
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Automation provides us with wondrous increases of production and information, but does it tell us what to do with the men the machines displace? Modern industry gives us the capacity for unparalleled wealth - but where is our capacity to make that wealth meaningful to the poor of every nation? Robert Kennedy
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Increasingly, the work we do is enabled more and more by new IT, including automation, robotics, and intelligent platforms. Pierre Nanterme
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

There is no question that automation is - and has been since the start of the Industrial Revolution - displacing workers and creating disruption within the economy and labor market. Charlie Kirk
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Rather than wringing our hands about robots taking over the world, smart organizations will embrace strategic automation use cases. Strategic decisions will be based on how the technology will free up time to do the types of tasks that humans are uniquely positioned to perform. Clara Shih
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

The structural changes of globalization and automation that has created concentrated wealth among some people who have had the right skills and the right opportunities has also created extraordinary disruption and havoc among the American middle-class. Ro Khanna
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Advances in automation, artificial intelligence and robotics, while increasing productivity, will also cause major upheavals to the workforce. John Hickenlooper
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

The impending destruction of jobs due to automation and AI technologies is definitely increasing the need for - and speed at which - we have to implement big solutions, such as a universal basic income. Andrew Yang
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

Besides black art, there is only automation and mechanization. Federico Garcia Lorca
Read more at https://www.brainyquote.com/search_results?q=automation+technology+quotes

There are 3 pages
Pages