ByteDance’s Pico reveals its latest VR headset as it aims to compete with Meta Quest 2

subsidiary Pico has . The Pico 4 will initially be available in Japan, South Korea, the UK, France, Germany, Spain, Italy and eight other European countries. Pico hasn’t revealed US release plans as yet, but it aims to bring the device to Singapore and Malaysia by the end of the year, and China at a later date.

The headset — which has a Qualcomm XR2 processor, an Adreno 650 GPU and 8GB of RAM — can be used as a standalone device. Pico claims the battery, which is in the rear strap to help keep things balanced, offers around three hours of use on a single charge, as notes. The device weighs 295 grams without the strap and 586 grams when it’s attached.

You can also connect Pico 4 to a gaming PC for higher-end VR experiences. That might be necessary to make full use of the dual displays, which offer higher than 4K resolution at 4,320 x 2,160 resolution for each eye. The displays have a 90Hz refresh rate and 105-degree field of view, according to   

The Pico 4 uses inside-out tracking with no need for external beacons. It comes with Pico 4 motion controllers (which have vibration features) and there are four external cameras, as points out. According to the Pico website, the device will offer full-color passthrough — something Meta is working on for its Project Cambria headset. 

A person wearing the Pico 4 virtual reality headset.


Given that ByteDance also owns , it shouldn’t be surprising that there’s a way to view videos from that app. You’ll be able to share VR experiences to TikTok as well. There will be hundreds more things to watch in VR and 360 formats. Pico is working to bring live sports and “avatar-based concerts” to the platform as well. 

As for games, there are 165 of them in the Pico store and more being added each week. The headset will support the likes of Peaky Blinders: The King’s Ransom, Demeo, , All-in-One Summer Sports VR and Just Dance VR (which will arrive in 2023 as a Pico exclusive). There will also be SteamVR support.

Meanwhile, there are plans to launch a metaverse-style experience called Pico Worlds next year. Unlike in Meta’s , Pico’s avatars appear to have legs.

Considering the price and specs, it’s Pico is trying to compete with Meta Quest 2 (Meta recently of that product). Whether the brand can hang with Meta on the content front remains to be seen. Users are unlikely to be able to play Beat Saber, for instance. Pico also revealed its latest device just a few weeks before Meta will show off at least one VR headset, likely to be the higher-end Project Cambria model, .

A Pico 4 with 128GB of storage costs €429 (around $422). A model with double the storage capacity will run you €499 (approximately $491). Preorders open next month and the headset will ship on October 18th. Pico also to release some accessories next year. A more accurate fitness tracker, a wireless dongle for PC connectivity and a carrying case will each cost €50 (or around $49).

All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission. All prices are correct at the time of publishing.

Repost: Original Source and Author Link


8BitDo reveals wireless versions of its Xbox-style Ultimate Controller

Well-renowned peripheral maker 8BitDo has revealed three new versions of its , including two wireless options. There are Bluetooth, 2.4GHz and wired variants, all of which are available to pre-order now. The controllers will ship on October 28th.

Both the Bluetooth and 2.4GHz models have rechargeable lithium-ion batteries and come with a charging dock. All three have a switch that allows you to swap between three profiles on the fly. You can remap the controller and adjust other settings using 8BitDo’s Ultimate Software on Windows, Android and iOS (but only on PC for the wired controller). The Xbox-style controller also has two back paddle buttons.

8BitDo Ultimate Controller


The Bluetooth version works with Nintendo Switch and Windows, though you may need to connect the included 2.4GHz adapter (which nestles inside the charging dock) to your PC. and it’s available in black and white. The 2.4GHz model is compatible with Windows, Android and Raspberry Pi, 8BitDo says. That one comes in white, black and pastel pink. It’s all yours .

As for the wired USB variant, that’s compatible with Switch, Windows, Android and Raspberry Pi. Other features include trigger vibration and enhanced grip. You can choose between black and white colorways for the wired version, which will .

8BitDo Ultimate Controller


All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.

Repost: Original Source and Author Link


Black Hat 2022 reveals enterprise security trends

Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.

The blast radius of cyberattacks on an enterprise is projected to keep growing, extending several layers deep into software supply chains, devops and tech stacks. Black Hat 2022’s presentations and announcements for enterprise security provide a sobering look at how enterprises’ tech stacks are at risk of more complex, devastating cyberattacks. Held last week in Las Vegas and in its 25th consecutive year, Black Hat‘s reputation for investigative analysis and reporting large-scale security flaws, gaps and breaches are unparalleled in cybersecurity.

The more complex the tech stack and reliant on implicit trust, the more likely it is to get hacked. That’s one of several messages Chris Krebs, the former and founding director of the U.S. Cybersecurity and Infrastructure Security Agency (CISA), delivered in a keynote to the audience at the Black Hat 2022 conference last week. Krebs mentioned that weaknesses often start from building overly complex tech stacks that create more attack surfaces for cybercriminals to then attempt to exploit.

Krebs also emphasized how critical software supply chain security is, explaining that enterprises and global governments aren’t doing enough to stop another attack at the scale of SolarWinds.

“Companies that are shipping software products are shipping targets,” he told the keynote audience.


MetaBeat 2022

MetaBeat will bring together thought leaders to give guidance on how metaverse technology will transform the way all industries communicate and do business on October 4 in San Francisco, CA.

Register Here

Cybercriminals “understand the dependencies and the trust connections we have on our software services and technology providers, and they’re working up the ladder through the supply chain,” Krebs added.

Additionally, eliminating implicit trust is table stakes for reducing supply chain attacks, a point Krebs alluded to throughout his keynote. 

Enterprise security: Reducing the growing blast radius 

Infrastructure, devops, and enterprise software vulnerabilities discovered by researchers made the enterprise-specific sessions worth attending. In addition, improving identity access management (IAM) and privileged access management (PAM), stopping ransomware attacks, reducing Azure Active Directory (AD) and SAP HTTP server attacks, and making software supply chains more secure dominated the enterprise sessions. 

Continuous integration and continuous delivery (CI/CD) pipelines are software supply chains’ most dangerous attack surfaces. Despite many organizations’ best efforts to integrate cybersecurity as a core part of their devops processes, CI/CD software pipelines are still hackable.

Several presentations at the conference explored how cybercriminals can hack into software supply chains using remote code execution (RCE) and infected code repositories. One session in particular focused on how advanced hackers could use code-signing to be indistinguishable from a devops team member. 

Another illustrated how hackers quickly use source code management (SCM) systems to achieve lateral movement and privilege escalation across an enterprise, infecting repositories and gaining access to software supply chains at scale.

Tech stacks are also becoming a more accessible target as cybercriminals’ skills increase. One presentation on how Azure AD user accounts can be backdoored and hijacked by exploiting external identity links to bypass multifactor authentication (MFA) and conditional access policies showed just how an enterprise can lose control of a core part of their tech stack in only minutes. 

A separate session on SAP’s proprietary HTTP server explained how cybercriminals could leverage two memory corruption vulnerabilities found in SAP’s HTTP server using high-level protocol exploitation techniques. CVE-2022-22536 and CVE-2022-22532 are remotely exploitable and could be used by unauthenticated attackers to compromise any SAP installation globally.

Malware attacks continue to escalate across enterprises, capable of bypassing tech stacks that rely on implicit trust and disabling infrastructure and networks. Using machine learning (ML) to identify potential malware attacks and thwart them before they happen using advanced classification techniques is a fascinating area of research. Malware Classification with Machine Learning Enhanced by Windows Kernel Emulation presented by Dmitrijs Trizna, security software engineer at Microsoft, provided a hybrid ML architecture that simultaneously utilizes static and dynamic malware analysis methodologies. 

During an interview prior to his session, Trizna explained that  “AI [artificial intelligence] is not magic, it’s not the silver bullet that will solve all your (malware) problems or replace you. It’s a tool that you need to understand how it works and the power underneath. So don’t discard it completely; see it as a tool.”

Trizna makes ML code for the models he’s working on available on GitHub.  

Cybersecurity vendors double down on AI, API and supply chain security 

Over 300 cybersecurity vendors exhibited at Black Hat 2022, with most new product announcements concentrating on API security and how to secure software supply chains. In addition, CrowdStrike’s announcement of the first-ever AI-based indicators of attack (IOA) reflects how fast cybersecurity providers are maturing their platform strategies based on AI and ML advances. 

CrowdStrike’s announcement of AI-powered IOAs is an industry first

Their AI-based IOAs announced at Black Hat combine cloud-native ML and human expertise, a process invented by CrowdStrike more than a decade ago. As a result, IOAs have proven effective in identifying and stopping breaches based on actual adversary behavior, irrespective of the malware or exploit used in an attack.

AI-powered IOAs rely on cloud-native ML models trained using telemetry data from CrowdStrike Security Cloud, as well as expertise from the company’s threat-hunting teams. IOAs are analyzed at machine speed using AI and ML, providing the accuracy, speed and scale enterprises need to thwart breaches. 

“CrowdStrike leads the way in stopping the most sophisticated attacks with our industry-leading indicators of attack capability, which revolutionized how security teams prevent threats based on adversary behavior, not easily changed indicators,” said Amol Kulkarni, chief product and engineering officer at CrowdStrike. “Now, we are changing the game again with the addition of AI-powered indicators of attack, which enable organizations to harness the power of the CrowdStrike Security Cloud to examine adversary behavior at machine speed and scale to stop breaches in the most effective way possible.” 

AI-powered IOAs have identified over 20 never-before-seen adversary patterns, which experts have validated and enforced on the Falcon platform for automated detection and prevention. 

“Using CrowdStrike sets Cundall apart as one of the more advanced organizations in an industry that typically lags behind other sectors in I.T. and cybersecurity adoption,” said Lou Lwin, CIO at Cundall, a leading engineering consultancy. “Today, attacks are becoming more sophisticated, and if they are machine-based attacks, there is no way an operator can keep up. The threat landscape is ever-changing. So, you need machine-based defenses and a partner that understands security is not ‘one and done.’ It is evolving all the time.” 

CrowdStrike demonstrated AI-powered IOA use cases, including post-exploitation payload detections and PowerShell IOAs using AI to identify malicious behaviors and code.  

AI-generated IOA fortifies existing defenses using cloud-based ML and real-time threat intelligence to analyze events at runtime and dynamically issue IOAs to the sensor. The sensor then correlates the AI-generated IOAs (behavioral event data) with local events and file data to assess maliciousness. CrowdStrike says AI-powered IOAs operate asynchronously alongside existing layers of sensor defense, including sensor-based ML and IOAs. Image credit: CrowdStrike.

For many enterprises, API security is a strategic weakness 

Cybersecurity vendors see the opportunity to help enterprises solve this challenge, and several announced new solutions at Black Hat. Vendors introducing new API security solutions include Canonic Security, Checkmarx, Contrast Security, Cybersixgill, Traceable, and Veracode. Noteworthy among these new product announcements is Checkmarx’s API Security, which is a component of their well-known Checkmarx One platform. Checkmarx is known for its expertise in securing CI/CD process workflows

 API Security can identify zombie and unknown APIs, perform automatic API discovery and inventory and perform API-centric remediation. In addition, Traceable AI announced several improvements to their platform, including identifying and stopping malicious API bots, identifying and tracking API abuse, fraud and misuse, and anticipating potential API attacks throughout software supply chains.

Stopping supply chain attacks before they get started 

Of the more than 300 vendors at Black Hat, the majority with CI/CD, devops, or zero-trust solutions promoted potential solutions for stopping supply chain attacks. It was the most hyped vendor theme at Black Hat. Software supply chain risks have become so severe that the National Institute of Standards and Technology (NIST) is updating its standards, including NIST SP 1800-34, concentrating on systems and components integral to supply chain security. 

Cycode, a supply-chain security specialist, announced it has added application security testing (SAST) and container-scanning capabilities to its platform, as well as introducing software composition analysis (SCA). 

Veracode, known for its expertise in security testing solutions, introduced new enhancements to its Continuous Software Security Platform, including software bill of materials (SBOM) API, support for software composition analysis (SCA), and support for new frameworks including PHP Symfony, Rails 7.0, and Ruby 3.x. 

The Open Cybersecurity Schema Framework (OCSF) meets an enterprise security need  

CISOs’ most common complaint regarding endpoint detection and response (EDR), endpoint management, and security monitoring platforms is that there is no common standard for enabling alerts across platforms. Eighteen leading security vendors have collaborated to take on the challenge, creating the Open Cybersecurity Schema Framework (OCSF) project. The project includes an open specification that enables the normalization of security telemetry across a wide range of security products and services. Open-source tools are also available to support and accelerate OCSF schema adoption.

Leading security vendors AWS and Splunk are cofounders of the OCSF project, with support from CrowdStrike, Palo Alto Networks, IBM Security and others. The goal is to continually create new products and services that support the OCSF specifications, enabling standardization of alerts from cyber monitoring tools, network loggers, and other software, to simplify and speed up the interpretation of that data. 

“At CrowdStrike, our mission is to stop breaches and power productivity for organizations,” said Michael Sentonas, chief technology officer, CrowdStrike. “We believe strongly in the concept of a shared data schema, which enables organizations to understand and digest all data, streamline their security operations, and lower risk. As a member of the OCSF, CrowdStrike is committed to doing the hard work to deliver solutions that organizations need to stay ahead of adversaries.”

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.

Repost: Original Source and Author Link


AMD accidentally reveals its first Ryzen 7000 desktop processors

You don’t have to wait until the fall to have an idea of what AMD’s Ryzen 7000 desktop CPUs will be like — the company has unintentionally provided clues of its own. Videocardz and Gizmodo note AMD’s resource page contained a since-removed list of early Ryzen 7000 processor models. The focus is primarily on higher-end chips, including two Ryzen 9 variants (the 7900X and 7950X), one Ryzen 7 (the 7700X) and a Ryzen 5 model (the 7600X). There’s no Ryzen 3 chips, although that last part isn’t shocking when AMD has historically focused on enthusiast parts in the early stages of CPU rollouts.

The list didn’t include technical details. In its Computex demo, however, AMD showed a 16-core CPU that reached a 5.5GHz clock speed. That might represent the Ryzen 9 7950X. All of the 7000 series will be based on a new Zen 4 architecture that delivers twice the Level 2 cache per core, maximum boost speeds above 5GHz, AI acceleration and support for technologies like DDR5 memory and PCIe 5.0. You’ll need an AM5-compatible motherboard to make the leap, but AMD is promising a 15 percent or higher increase in single-threaded performance.

It’s still unclear when the Ryzen 7000 desktop line will ship, or how much it will cost. You’ll also have to wait longer if you’re hoping for high-end laptop CPUs, as AMD won’t deliver Dragon Range until 2023. Even so, the teaser gives you an idea of what to expect when Zen 4 finally reaches stores. AMD isn’t reinventing its product strategy, so you can plan your PC upgrade accordingly.

All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.

Repost: Original Source and Author Link


Homeland Security bug bounty reveals huge number of flaws

The outcome of a bug bounty program for the Department of Homeland Security (DHS) has been revealed, and it’s not particularly encouraging news for a government agency synonymous with cyber security.

Participants of DHS’ first-ever bug bounty program, named “Hack DHS,” confirmed that they found a worrying number of security bugs.

Stock Depot/Getty Images

They discovered a total of 122 security vulnerabilities in external DHS systems, according to The Register and Bleeping Computer. Twenty-seven bugs were recognized as “critical severity” flaws.

The Hack DHS initiative saw more than 450 security researchers participate in the program. For their efforts, the government agency paid out a total reward of $125,600 that was distributed amongst the ethical hackers.

As aptly highlighted by The Register, the aforementioned payout figure pales in comparison to what other organizations pay to bug bounty hunters.

For example, Intel has previously offered up to $100,000 for successfully uncovering specific vulnerabilities.

Other technology giants like Microsoft offer 10s of thousands of dollars for finding flaws, while Apple paid a single individual nearly the entirety of the Hack DHS bounty by giving him $100,000 for hacking a Mac.

Google, meanwhile, has awarded nearly $30 million to individuals enrolled in its own bug bounty programs. In one particular case, the company gave a self-taught teenage hacker $36,000 for reporting a certain bug.

Considering the fact that one of the Department of Homeland Security’s key responsibilities involves cyber security, many may understandably be concerned that such a high amount of security bugs were found in the first place. Moreover, the somewhat lackluster payment tiers associated with Hack DHS could be a potential deterrent to future interested parties.

All things considered, it seems the DHS is not as secure as many Americans would have hoped it would be.

A physical lock placed on a keyboard to represent a locked keyboard.
piranka/Getty Images

Homeland Security’s quest to become more secure

Hack DHS was originally introduced in December 2021. Any hacker who joined the program would have to provide a comprehensive breakdown of any vulnerability they find. They also have to detail how that flaw can be targeted and exploited by potential threat actors, as well as explain how it can be specifically utilized to access and extract data from DHS systems.

Once these security defects are put through a verification process by “DHS security experts,” which takes 48 hours to analyze after a bug is detected and submitted, they are generally patched within 15 days or so. In some cases, it takes the government agency longer than half a month to fix the more intricate flaws.

The government agency’s bug bounty program will be conducted via a tiered rollout consisting of three stages. The first phase, payouts, has been completed, while the upcoming second stage will see security researchers hand-picked by the DHS taking part in a live hacking event.

As for the final phase, The Register reports that DHS will share information that it hopes will influence additional bug bounty programs.

The popularity of bug bounty programs is increasingly becoming more prominent in an era where cybercriminals have been intensifying their attempts to infiltrate major companies, especially in the technology space.

For example, Intel unveiled Project Circuit Breaker, an expansion to its bug bounty program that was introduced to recruit “elite hackers.” Google also updated its Vulnerability Reward Program last year by launching a new bug platform.

Elsewhere, Google recently confirmed that a record number of dangerous zero-day exploits were identified in 2021, while cybercrimes are more widespread than ever before.

Editors’ Choice

Repost: Original Source and Author Link


Microsoft reveals new secret weapon against cybercrime

Microsoft announced a new cybersecurity-based initiative that will allow small businesses and huge enterprises alike to tap into the tech giant’s in-house security services and personnel.

Named Microsoft Security Experts, the program will offer security services in the form of three distinct platforms.


As reported by PCMag, the firm asserted in its announcement post that the current “security landscape has become increasingly challenging and complex for our customers.”

With the pandemic accelerating the growth of individuals’ and businesses’ digital presence more than ever before, it revealed that cybercrime threats have increased “at an alarming rate over the last year”.

In fact, Microsoft provided some statistics to show how dire the situation has become when it comes to digital security. In 2011, the costs associated with cybercrime totaled $3 trillion, while that number increased two-fold to $6 trillion in 2021. By 2025, that figure is forecasted to soar to $10.5 trillion on an annual basis, according to Cybersecurity Ventures.

In 2021 alone, the company said its Microsoft Security service blocked over 9.6 billion malware threats, in addition to detecting and blocking upward of 35.7 billion phishing and other malicious emails.

As for the attacks that evaded security and mitigation systems, the consequences proved to be devastating. In 2021, nearly $7 billion was stolen from individuals due to various cybercrimes — malware and phishing scams as well as others.

Furthermore, more than 35 ransomware strains are actively being monitored by Microsoft Security. The platform is also keeping tabs on 250 unique threat actors “across observed nation-state, ransomware, and criminal activities.”

Its technology allows Microsoft to block more than 900 brute force password theft attempts, which are attempted every single second, so don’t forget to reinforce your passwords.

The importance of adequate security measures pertaining to passwords shouldn’t be understated. We’ve recently seen some of the world’s largest corporations being infiltrated solely due to weak passwords.

Microsoft stressed, however, that technology in and of itself can simply not handle the threat of cybercrime without additional assistance.

“Technology is critical, but it’s the combination of leading technologies, comprehensive threat intelligence, and highly skilled people that makes for a truly effective security posture. The challenge is that in this critical moment when cybersecurity has reached an inflection point, our nation is facing a cybersecurity talent shortage…”

A digital depiction of a laptop being hacked by a hacker.
Digital Trends Graphic

Talent is not enough

An interesting statistic was mentioned in the blog post: Nearly one in three security jobs are currently vacant within the U.S. alone. As a result of this state of affairs and the lack of talent needed to fill said positions, the time of detection for a breach has reached an “alarming” 287 days.

Microsoft acknowledged that although talent may be readily available, general access to “highly skilled expertise remains a challenge.”

As such, the technology company is responding to the situation through an expansion of its existing service capabilities via Microsoft Security Experts.

“It’s getting harder every day for organizations to build and maintain a full security team, let alone one with the ever-expanding skillset required to meet the range of today’s security demands.

Security Experts combines expert-trained technology with human-led services to help organizations achieve more secure, compliant, and productive outcomes.

Our vision is to deliver this new category of services across security, compliance, identity, management, and privacy. The first step on that journey is offering new and expanded services for security.

Microsoft is uniquely positioned to help our customers and their partners meet today’s security challenges. We secure devices, identities, apps, and clouds — the fundamental fabric of our customers’ lives — with the full scale of our comprehensive multicloud, multiplatform solutions. Plus, we understand today’s security challenges because we live this fight ourselves every single day.

Now, our world-class security expertise is your security expertise.”

As for the services housed under the newly formed service category, these include three “managed services that can help you scale your team of experts to fit your needs — without the challenges of hiring and training them.”

Three new security services businesses can use

Microsoft Defender Experts for Hunting is for businesses or customers that already have an advanced security system in place, but require further help to actively monitor and combat threats that materialize within Microsoft Defender data (including endpoints), Office 365, cloud applications, and identity.

Microsoft Defender Experts for XDR, meanwhile, has been designed to supply customers with a service that can strengthen and enlarge the capacity of their security operations center.

Finally, ​​Microsoft Security Services for Enterprise has been reserved for the larger businesses that are “looking for more comprehensive, high-touch managed services from Microsoft experts.”

“This comprehensive, expert-led service combines proactive threat hunting and managed XDR, leveraging Microsoft’s complete security information and event management (SIEM) and XDR stack to protect all cloud environments and all platforms.

Dedicated Microsoft security experts manage onboarding, daily interactions, practice modernization, and incident response for you. Microsoft Security Services for Enterprise is sold through a custom statement of work and is available today. Interested enterprise customers should contact their Account Executive to learn more.”

Microsoft has previously brought attention to the lack of interest in the cybersecurity field. It highlighted how there will be 3.5 million vacant cybersecurity jobs by 2025.

As part of its efforts, it hopes to “help close the gap in the profession” by expanding its cybersecurity skills campaign to 23 more countries.

Hacking as a whole is reaching new heights as of late.

For example, cybercriminals have even found ways to target a PC’s power supply and manipulate it to cause a fire inside one’s own home.

Putting that dangerous scenario aside, a study published by Google confirmed that a record number of zero-day exploits, which have been given the distinction of being “one of the most advanced attack methods,” were attempted against leading technology firms, including Apple, Microsoft, and Google itself last year.

Editors’ Choice

Repost: Original Source and Author Link


M2 MacBook Air teardown reveals bad news for performance

The newly released MacBook Air featuring Apple’s latest proprietary M2 chip has the gotten teardown treatment from the Max Tech YouTube channel, revealing a basic look at the internals of the notebook and some bad news about the system’s performance.

The featured model included 256GB SSD in a single NAND storage chip. MacRumors noted this specs setup has the potential to result in a notebook that performs 30% to 50% slower in SSD benchmark tests than MacBook Air models with higher specs or older MacBook Air models with the same specs. This was also a problem with the recently launched M2 MacBook Pro.

MacBook Air motherboard, as shown by Max Tech teardown.

The video also discussed some interesting details during the teardown, such as the more elongated design, which allows for a larger and more powerful battery cell in the M2 MacBook Air. This model features a 52.6-watt hour battery, a slight upgrade from the 49.9 watt-hour battery in the M1 MacBook Air. While subtle, Eremenko noted that the new notebook has more internal volume, which allows it to hold a larger battery, which is expected to power the device for up to 18 hours at a time.

The notebook features speakers located at the front of the paneling. They are also front-firing, which is intended to make for better audio.

Finally, the teardown mentions that the M2 MacBook Air includes an ultra-wideband chip within its motherboard. The component appears to currently not be functional; however, it has the potential to be unlocked in the future for uses, including Air Tags and lossless wireless audio.

M2 MacBook Air Teardown: Apple’s SECRET Revealed (& SSD)

The M2 MacBook Air has been available for pre-order since July 8 and is now on sale in stores. Prior reports detailed that the notebook was quickly back -ordered and experienced shipment delays as late as mid-August, as soon as pre-orders opened.

Many Apple products have experienced shipping delays throughout the year; however, it is not clear whether the M2 MacBook Air shipping delays have been caused by consumer demand or by supply chain issues.

The M2 MacBook Air model, which is featured in the video, starts at $1,199.

Editors’ Choice

Repost: Original Source and Author Link


Nvidia reveals H100 GPU for AI and teases ‘world’s fastest AI supercomputer’

Nvidia has announced a slew of AI-focused enterprise products at its annual GTC conference. They include details of its new silicon architecture, Hopper; the first datacenter GPU built using that architecture, the H100; a new Grace CPU “superchip”; and vague plans to build what the company claims will be the world’s fastest AI supercomputer, named Eos.

Nvidia has benefited hugely from the AI boom of the last decade, with its GPUs proving a perfect match for popular, data-intensive deep learning methods. As the AI sector’s demand for data compute grows, says Nvidia, it wants to provide more firepower.

In particular, the company stressed the popularity of a type of machine learning system known as a Transformer. This method has been incredibly fruitful, powering everything from language models like OpenAI’s GPT-3 to medical systems like DeepMind’s AlphaFold. Such models have increased exponentially in size over the space of a few years. When OpenAI launched GPT-2 in 2019, for example, it contained 1.5 billion parameters (or connections). When Google trained a similar model just two years later, it used 1.6 trillion parameters.

“Training these giant models still takes months,” said Nvidia senior director of product management Paresh Kharya in a press briefing. “So you fire a job and wait for one and half months to see what happens. A key challenge to reducing this time to train is that performance gains start to decline as you increase the number of GPUs in a data center.”

Nvidia says its new Hopper architecture will help ameliorate these difficulties. Named after pioneering computer scientist and US Navy Rear Admiral Grace Hopper, the architecture is specialized to accelerate the training of Transformer models on H100 GPUs by six times compared to previous-generation chips, while the new fourth-generation Nivida NVlink can connect up to 256 H100 GPUs at nine times higher bandwidth than the previous generation.

The H100 GPU itself contains 80 billion transistors and is the first GPU to support PCle Gen5 and utilize HBM3, enabling memory bandwidth of 3TB/s. Nvidia says an H100 GPU is three times faster than its previous-generation A100 at FP16, FP32, and FP64 compute, and six times faster at 8-bit floating point math.

“For the training of giant Transformer models, H100 will offer up to nine times higher performance, training in days what used to take weeks,” said Kharya.

The company also announced a new data center CPU, the Grace CPU Superchip, which consists of two CPUs connected directly via a new low-latency NVLink-C2C. The chip is designed to “serve giant-scale HPC and AI applications” alongside the new Hopper-based GPUs, and can be used for CPU-only systems or GPU-accelerated servers. It has 144 Arm cores and 1TB/s of memory bandwidth.

The new Grace CPU “superchip” consists of two CPUs connected together.
Image: Nvidia

In addition to hardware and infrastructure news, Nvidia also announced updates to its various enterprise AI software services, including Maxine (an SDK to deliver audio and video enhancements, intended to power things like virtual avatars) and Riva (an SDK used for both speech recognition and text-to-speech).

The company also teased that it was building a new AI supercomputer, which it claims will be the world’s fastest when deployed. The supercomputer, named Eos, will be built using the Hopper architecture and contain some 4,600 H100 GPUs to offer 18.4 exaflops of “AI performance.” The system will be used for Nvidia’s internal research only, and the company said it would be online in a few months’ time.

Over the past few years, a number of companies with strong interest in AI have built or announced their own in-house “AI supercomputers” for internal research, including Microsoft, Tesla, and Meta. These systems are not directly comparable with regular supercomputers as they run at a lower level of accuracy, which has allowed a number of firms to quickly leapfrog one another by announcing the world’s fastest.

However, during his keynote address, Nvidia CEO Jensen Huang did say that Eos, when running traditional supercomputer tasks, would rack 275 petaFLOPS of compute — 1.4 times faster than “the fastest science computer in the US” (the Summit). “We expect Eos to be the fastest AI computer in the world,” said Huang. “Eos will be the blueprint for the most advanced AI infrastructure for our OEMs and cloud partners.”

Repost: Original Source and Author Link


Naughty Dog reveals more details about ‘The Last of Us’ remake for PS5 and PC

As if official confirmation was actually needed after the leak earlier today, a remake of 2013’s The Last Of Us is coming to PlayStation 5 on September 2nd. Sony put together a trailer showing some of the gorgeous visuals of The Last of Us Part I and noted that the remake is coming to PC as well. The bundle also includes the excellent Left Behind expansion.

The developers used original performances from Ashley Johnson, Troy Baker, and the rest of the cast, but utilized a new AI and refreshed the combat. The effects and exploration have been enhanced as well. The team is harnessing 3D audio and the DualSense controller’s adaptive triggers and haptic feedback. There will be more accessibility features than the original game had too, building on the extensive options in The Last of Us Part II. More details will be revealed in the coming months.

The remake is a full-price game at $70. Opt for the $80 Digital Deluxe edition and you’ll be able to unlock some items and features early, including a speedrun mode, explosive arrows(!), weapon skins and modifiers for faster crafting and healing. The $100 Firefly edition includes all of those bonuses as well as a steelbook cover for the physical version and four issues of the The Last of Us: American Dreams comic. There are pre-order perks for every version too.

That’s not the only news about the franchise that was revealed today. It’s long been acknowledged that a standalone multiplayer games set in the world of The Last of Us was in the works, and now Naughty Dog has revealed more details about what’s in store. Studio co-president Neil Druckmann, the director of the first two games in the series, showed the first concept art at Summer Game Fest.

TLOU multiplayer game

Naughty Dog

Druckmann said it will be “as big of as any of our single-player games that we’ve done, and in some ways bigger.” It has its own story and a new cast of characters, along with a fresh setting. It’s led by a team of Naughty Dog veterans who have worked on the series as well as the Uncharted games. More details will be announced next year. 

Following that announcement, Druckmann touched on the HBO’s upcoming The Last of Us series. While lots of set photos have popped up over the last year, he showed the second official still from the show. It shows Joel (Pedro Pascal) and (Bella Ramsay) taking cover. Bump up the brightness on the image, though, and you’ll see something lurking in the background.

The Last of Us show

PlayStation Productions/HBO

It also emerged that Johnson and Baker, who played Ellie and Joel in the games, will appear in the show, but their roles haven’t been revealed as yet. Filming will wrap on the first season tomorrow. While the series is expected to arrive next year, Druckmann said that fans will hear more about it “very soon.”

All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.

Repost: Original Source and Author Link


The first ‘Final Fantasy XVI’ gameplay trailer reveals a summer 2023 release window

Final Fantasy XVI is due to hit PlayStation 5 in the summer of 2023 — and Square Enix has a hot new trailer to prove it. The title’s first gameplay trailer showcases the massive monsters called Eikons and the Dominant humans that inhabit them, and presents chaotic, cinematic battle scenes in a medieval-inspired world.

FFXVI was officially revealed in September 2020 and developers promised to drop more information about it by the end of the following year. But in December 2021, the development team announced the game was nearly six months behind schedule due to the COVID-19 pandemic. Now that we’ve waited an extra half a year, the gameplay trailer is finally here.

“In our latest trailer, we’ve introduced several new Eikons, as well as provided a more detailed peek at our action-packed battle system and the freedom it gives players,” FFXVI director Hiroshi Takai said on the PlayStation blog. “As for development progress, I’m happy to announce that the game is fully playable from start to finish; though, from optimization to brush-up, there is still a mountain of challenges to tackle as we head into our final push.”

It seems like Square Enix is back on track with its FFXVI development timeline, aiming to release the game on PS5 next summer. It’s set to be exclusive to PS5 for a limited time.

All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.

Repost: Original Source and Author Link