TheThe problem with going off-grid is the lack of connection.

The urge to get away from it all without losing access to Slack and Instagram was a #vanlife trend long before COVID-19. The pandemic only accelerated it, fueled by social distancing rules, office closures, and flexible remote work policies that enable more people to set up shop from any location they choose. Few do so, however, because change is hard, and going truly remote often means being out of range of cell towers — but not satellites. That’s where Starlink RV comes in.

SpaceX launched its internet from space service in public beta in October 2020. The service has steadily improved ever since we tested it in May 2021, when we found it to be “unreliable, inconsistent, and foiled by even the merest suggestion of trees.” The latest advancements include the release of a smaller rectangular dish and sanctioned support for portability, most expressly with the launch of the Starlink RV service. Starlink RV allows owners to take their $599 / €639 Dishy McFlatface anywhere (on the same continent) there’s coverage, which now means large swaths of North America and nearly all of Western Europe. You can even pause and unpause the $135 / €124 per month service so that you’re only paying for the months you need it.

Importantly, Starlink RV targets people on the go, be it weekend campers, overlanders and vanlifers who live and work in their rigs year-round, or retirees with an RV or vacation home where they reside for months at a time. As such, Starlink RV is competing against pricey unlimited mobile data plans and signal boosters that attempt to fill holes in coverage, not home internet services. Unlike the fixed Starlink Residential service, which requires a perfect line of sight to the sky to be useful, subscribers to Starlink RV can move their Dishy at will, and likely be much more forgiving when the choice is between degraded service and no service at all.

I’ve been testing Starlink RV for the last two weeks in a variety of locations: from atop a Ventje T5 camper van parked in a German forest where tall trees block satellite visibility; at a crowded festival in Bavaria with overloaded cell towers; at a Dutch beach where the prospects of mobile data are bleak; and in a severe thunderstorm at my home in central Amsterdam. For me, Starlink RV’s performance was an epiphany and cause to get serious about leaving the shackles of my urban existence behind.

SpaceXSpaceX currently offers two flavors of Starlink services: Residential Starlink and Starlink RV. There’s also the Residential Starlink plus Portability option, which is a jumbled hybrid of the two. Each service starts with the same dish and Wi-Fi router kit ($599 / €639) but differs in terms of expected speeds, monthly fees, the ability to use the service when traveling, and the option to easily pause the service.

Starlink RV is SpaceX’s most flexible offering. The Starlink RV monthly subscription costs $135 / €124 compared to $110 / €99 per month for the Residential service (adding Portability costs an additional $25 / €25 each month). Each service comes with heavily caveated performance “goals” of 50–250Mbps downloads and 10–20Mbps uploads with 20–40ms of expected latency and “no data caps,” although it does warn against misuse and abuse.

One big advantage of Starlink RV is the ability to pause the service (and fees!) at any time and then resume it at a later date when you need it again. That can save subscribers a bundle of money if they’re only traveling a few months each year. The big disadvantage is that Starlink RV users are “always de-prioritized” compared to Residential subscribers. According to SpaceX, that could result in speeds closer to 5–100Mbps down and 1–10Mbps up when used in congested areas or during times of high usage.

Starlink RV found internet where mobile data was slow or didn’t exist at all.

If my service was indeed being de-prioritized, it was done with a relatively light touch at my testing locations and times — though I did experience a wide range of performance over the two week period. Using the iOS Speedtest app on my phone positioned less than 10 feet from the Starlink Wi-Fi router, I measured download speeds as slow as 44.2Mbps and as fast as 235Mbps, with uploads ranging from 8.9Mbps to 29.6Mbps. Speeds sometimes varied from one minute to the next, but mostly I was seeing download speeds on the average of 150–200Mbps and uploads between 10–15Mbps.

That’s not particularly fast by home internet standards in Europe. But, in the four remote locations where I set up Starlink RV in Germany and the Netherlands, Starlink was faster (sometimes 100 times faster) than the mobile data networks from T-Mobile and Vodafone, which sometimes failed to connect at all.

Starlink RV worked even when cell towers were overloaded at a weekend festival.

Hank sitting next to the Wi-Fi router as the Starlink RV dish connects through the trees.

For example, when parked alongside hundreds of other campers at a 4x4 show in the hills of Bavaria, Starlink’s satellites gave me 189 / 1 1.1Mbps (up / down) versus Vodafone’s 11.1 / 1.9Mbps due to the surge of people suddenly choking the nearest tower. On a relatively remote Dutch beach, I clocked 217 / 8.9Mbps while T-Mobile limped along at 0.7 / 0.16Mbps and Vodafone at 2.37 / 0.09Mbps. At one particular forest outside Dortmund, Germany, where we parked our camper to make lunch and walk the dog, Starlink was hitting 49 / 12Mbps through trees (more on that later), while Vodafone and T-Mobile failed to connect at all.

RV-grade signal boosters from companies like Weboost start at around $450 and would have likely improved cellular data speeds in some of my testing. You can sometimes double or even quadruple speeds with a booster, but even then they’d pale in comparison to the speeds achieved with the Starlink RV. And you can’t boost a signal that doesn’t exist.

To put these speeds into perspective, Zoom video conferencing requires up to 6Mbps down and 2Mbps up, while livestreaming needs at least 6Mbps up. But speed isn’t everything. Frequent dropouts can make video conferencing and streaming impossible, and latency above 60ms can lose an online campaign for gamers.

Regarding latency, the Starlink app was usually reporting lag around 50ms, with a range of 28–88ms. The more detailed Speedtest app usually reported idle latency (no other traffic) at around 50ms as well, with a range of 32–69ms. It also reported download latency (ping lag measured during downloads) between 161–293ms and upload latency (lag measured during uploads) between 71–169ms. YouTuber TTTHEFINEPRINTTT has some early positive impressions and live footage of gameplay over Starlink RV if that’s your thing. Regardless, Starlink still has some work to do to meet its latency goal of 20–40ms.

Knowing that I was traveling with Starlink RV meant that I was always looking for campsites with excellent visibility of the stars to avoid the well-documented dropouts and slowdowns caused by obstructions. My rectangular dish is 12 inches (50cm) wide and 19 inches (30cm) long. At 9.2 pounds (4.2kg), it’s nearly half the weight of the original 16-pound (7.2kg) dish. I stuffed my dish, router, and cables into a plush laundry bag. I kept the stand separate to avoid scratches, putting both the laundry bag and stand inside an old kitesurfing backpack to maximize portability.

Setting up the Starlink RV hardware takes only a few minutes after you’re parked. I usually placed the dish on the ground or on top of my camper and then ran its 75-foot (23m) cable back to the Wi-Fi router, where it attaches via a fiddly Micro USB connector that rarely lines up properly. Then you plug the router into your camper’s AC port, shore power, or big portable battery with a built-in inverter and watch Dishy rotate to life before turning to lock onto an overhead satellite. It would often take as long as 15 minutes after rolling into a new location before I had functioning Starlink internet — enough time to appreciate the space-age tech as I struggled with slow to nonexistent mobile networks.

When operating in an open field or at the beach, for example, Starlink’s perfect line-of-sight connectivity worked for hours on end without any network drops at all, allowing me to attend video conferences, make calls over Wi-Fi, watch TikTok videos, and stream Netflix and YouTube videos without issue.

I did experience some network dropouts and interruptions when using Starlink near obvious obstructions like tall buildings or trees. But I was still able to do things like message my kids, check that my home automations were running, work in Slack, research my next destination in Google Maps and Safari, check email, and myriad other things as the urge struck. About the only thing I tried and failed at was a FaceTime call that dropped mid-session. Fortunately, the whole kit is easy to pack up and move to another location with fewer obstructions if fast, reliable internet is more important than the view.

I also tested Starlink RV in the rain, including a severe thunderstorm in Amsterdam with plenty of lightning, wind, and a heavy downpour that lasted for about 20 minutes before exhausting itself into a steady rain. My service was interrupted at the onset, resulting in several “no signal received” messages logged in the Starlink app that lasted between 15–90 seconds over the first six minutes of the storm. Service didn’t stop completely; it just felt like really flaky Wi-Fi and certainly would have wreaked havoc had I been on a Zoom call for work. Things soon returned to normal with just a couple of 4-second “no signal received” messages over the next few hours of drizzle. In general, light rain didn’t seem to create any issues.

Other worthwhile mentions:

• SpaceX says the rectangular dish consumes an average of 48–74W — or 20W when idle. I think it might be better than that, having observed it jumping between 33–62W with an average draw closer to 42W when plugged into a Jackery portable battery with a built-in power meter. Power comes at a premium for RV owners, so it’s good to see SpaceX making regular improvements here.
• The Wi-Fi router supports 802.11a/b/g/n/ac on dual-band 2.4GHz and 5GHz. The IP54 rating makes it resistant to splashing water and rain, but it’s intended for indoor use only. It can easily cover a campsite for at least a hundred feet in each direction if you’d like to share your service with others (or charge for it). There’s no ethernet port, though, so you’ll need to buy an adapter from SpaceX for that.
• I often forgot to hit the “Stow” button in the Starlink app (under settings) before unplugging the power from the router in my rush to get on the road again. Fortunately, you can still put the dish into the stowed position by removing the stand, setting Mr. McFlatface’s face onto a flat service, and plugging the power back in. The dish folds down after a few seconds, which is much better than waiting several minutes for the Stow button to reappear again in the app after the satellite service reconnects. It doesn’t go completely flat in the stowed position, but it’s flat enough to make it easier to stow away in the recesses of a vehicle.
• Although Starlink RV’s upload speeds can be slow compared to fixed internet or strong mobile data connections, creators who need to upload large videos from the road can start the process at night and be done by the time they wake up in the morning. That’s way more convenient than having to search for a cafe that doesn’t mind you leeching their slow public Wi-Fi all day. Just be careful not to drain all your vehicle’s power if you’re running Starlink RV all night long.
• Using Starlink RV in a moving vehicle will void the warranty, according to SpaceX. “While our teams are actively working to make it possible to use Starlink on moving vehicles, Starlink is not yet configured to be safely used in this way,” reads a SpaceX support page. That’s why you won’t (yet) find an RV mount in the Starlink shop alongside all the other accessories.
• Starlink RV can only be used on the same continent as your registered shipping address — something you should note if you’re planning to outfit your Unimog overlander with Starlink RV for your trip to Morocco or Iceland.
• The no wait list feature of Starlink RV is legit. I got mine in less than a week. Some Residential folks have been waiting months for their Starlink kits to arrive.
• There’s a software setting in the Starlink app that sends extra power to the dish to melt snow. Cool.

The Starlink RV Wi-Fi router and dish require about 42W of power on average when active.

TheThe audiences for Starlink RV and Starlink Residential differ in two fundamental ways. First, Starlink RV users are the types who are surprised and grateful to have any connectivity at all, while Residential users expect rock-solid connectivity at all times. Second, Starlink RV users can more easily move their rolling home to avoid obstructions, which is something Starlink Residential users can’t do.

To be frank, I’m kind of blown away by the transformative experience of using Starlink RV over the last few weeks. I’ve been a budding vanlifer for years, scouring Instagram on the reg for Sprinter 4x4 porn. Sitting in a remote field and watching Dishy lock onto one of the thousands of SpaceX satellites orbiting overhead reminded me of the first time I used GPS to magically navigate myself home. With Starlink RV, the magic is realizing that I can now take my home on the road and navigate the next chapter of my career. Hey Nilay, let’s talk.

Photography by Thomas Ricker / The Verge

Starlink RV review: the dawn of space internet to go

We’re gonna need a bigger desk. Image: Samsung

Samsung is getting ready to release its massive 55-inch curved Odyssey Ark monitor in August, according to a report from Korean outlet ETNews (via SamMobile). The monitor, which was announced in January at CES, has reportedly gone through a few certification programs that have to be done before it can go on sale.

Details about this monitor are still scant. The company has said that it’ll have a 16:9 4K panel and that the stand will support pivot, tilt, and rotation. It also announced that it would come out in the second half of 2022 (which the reported August window falls squarely in). But let’s be honest, when you show up with a monitor that promises to physically tower over you while you’re using it, people will pay attention even if you don’t announce the price or refresh rate.

Samsung says the dial is for managing “lighting and the interface.” Because buttons on the back would be very hard to reach. Image: Samsung

Many of us here at The Verge are very excited for this monitor — when it came time to hand out awards for CES 2022, we gave it Best in Show. But as with many CES announcements, as the months wear on, it can be hard to remember anything from the flood of gadgets and tech; did Samsung really announce an extremely curved, extremely large monitor that can stretch over and around you, or was that just a fever dream? Rumors that it could actually be coming out relatively soon prove that it was, in fact, actually announced and reminds us that we are looking forward to it.

One thing that adds a bit of credence to the report is that Samsung has successfully released other monitors that it announced around CES this year. The decidedly less ambitious (but very cute) M8 monitor has already started hitting store shelves, as has a smaller (read: reasonably sized) curved gaming monitor, the Odyssey Neo G8. While that’s obviously not proof that the Ark is up next, it is good to see that Samsung has a track record of shipping its CES monitors this year.

Samsung’s monstrous 55-inch Odyssey Ark monitor could go on sale in August

Now, Microsoft has announced that it is open-sourcing its Grounded Open Dialogue Language Model (GODEL) for use by a wider audience.

GODEL combines two important functionalities in a single language model. The first is being task-oriented and the second is to still make the conversation a realistic and social one. Typically, chatbots try to either be goal-driven - like support assistants - or they just attempt to be social without a purpose.

GODEL is a class of pre-trained models that basically allow a dialog agent such as a chatbot to provide answers based on the current conversation while also generating external and related information to make your interaction more realistic and grounded. An example provided by Microsoft is a conversation about restaurant recommendations branching out to discuss recipes, cooking techniques, and more.

Microsoft wants GODEL to be unrestricted in terms of the queries it can answer and expand upon, while still making the overall conversation useful.

GODEL is actually an evolution of DialoGPT, which Microsoft Research released in 2019. Microsoft's automated- and human-based testing has shown its latest conversational AI to be far superior to DialoGPT, which is why the firm believes that it is ready for broader use in diverse environments. Below, you can see an example of a conversation with GODEL in which it responds to a question about a current event that was not a part of its training set:

GODEL is now available as an open-source repository on GitHub here and you can also check out the associated research paper here. It is offered in three variants along with the code you would need to re-train GODEL on a specific set of information.

Microsoft opens-sources GODEL language model for realistic conversations

And with the launch of the Arc A380, the company also published a system requirement guide since the Arc GPUs are new products that system builders aren't much familiar with it. In its guidelines, Intel clarified that PCIe Resiable BAR (ReBAR) feature must be enabled for "optimal" performance with Arc. Alongside that, support for PCIe ReBAR on Intel platforms was also confirmed. You can view the supported hardware configurations in this article.

In its guide, the company hinted at supporting AMD's implementation of ReBAR, also called Smart Access Memory (SAM). It stated:

Additional platforms/motherboards with Resizable BAR / Smart Access Memory enabled may also support Intel® Arc™ A-Series graphics.

Today, in a statement issued to Neowin, Intel has confirmed that Smart Access Memory (SAM) support is indeed coming to Arc which means AMD Ryzen platforms will be on an equal playing field with their Intel counterparts.

Here is Intel's official statement:

As we described in a blog post last month, the Arc graphics product rollout involves a staggered introduction on targeted platforms to most effectively serve our customer base. We are supporting Intel platforms with resizable BAR and will add support for AMD platforms with Smart Access Memory as Intel Arc graphics cards become available for sale as components.

Of course, just like on Intel, not every AMD platform will support SAM. Generally, you need a Ryzen 3000 series processor (except the 3000 G-series APUs) and pair up with a 500 series chipset AM4 board. Upcoming AM5 motherboards will also very likely support SAM.

Recently, in a comparison using the Arc A380 on an AMD platform, it was found that Arc suffered severe performance impact on the AMD system compared to the Intel machine. This is likely due to the lack of Smart Access Memory support alongside lack of some other optimizations. Though we must remember that Arc is very new to the market and that in the future, such teething issues are expected to be ironed out.

Intel officially confirms AMD Smart Access Memory (SAM) support is coming to Arc

According to the first independent review, although the new Arc A380 was decent, it did not manage to beat either AMD or Nvidia's offerings in actual games, even though it shined in synthetic benchmarks. The testing was done on an Intel i5-12400 CPU paired with a PCIe 4.0-capable Asus B660 board.

However, a new review on an AMD platform is adding more to the story of Arc. The test results suggest that Intel's Arc graphics cards may not yet be optimized for AMD platforms just yet.

The Gunnir Arc A380 Photon was coupled with a Ryzen 5 5600 and a PCIe 4.0-ready Asus TUF Gaming B550M-PLUS WiFi II motherboard. The gaming numbers show performance regression for Arc in almost all of the game titles and sometimes the margins are massive. For example, in Forza Horizon 5, the Arc A380 is 17% slower on the AMD platform (73%) compared when it's run on Intel (88%).

Intel has already confirmed that Arc requires PCIe Resizable BAR (ReBAR) for optimal performance and so the test was conducted with the feature enabled on both platforms. However, it's possible that Arc is un-optimized still for AMD's ReBAR implementation, also called Smart Access Memory (SAM).

The good news is that Intel has hinted that it will optimize Arc for AMD platforms later in its official announcement. This is the polar opposite of what the company did with its previous Iris Xe desktop card launch when it outright declared no plans for supporting AMD or older Intel platforms.

Source: 远古时代装机猿 (Bilibili) via Greymon55 (Twitter)

Intel's new Arc A380 desktop GPU hates being coupled with an AMD Ryzen CPU, for now at least

AR, meet ML: IKEA app lets you erase and replace your furniture

Broadcom takeover of VMware could be derailed by EU antitrust probe

The on-stage demo showcased a use-case where Alexa read a bedtime story to a child in the voice of his dead grandmother. Prasad stated that:

This required inventions where we had to learn to produce a high-quality voice with less than a minute of recording versus hours of recording in the studio. The way we made it happen is by framing the problem as a voice conversion task and not a speech generation path. We are unquestionably living in the golden era of AI, where our dreams and science fictions are becoming a reality.

Amazon claims that voice can be synthesized using just one minute of audio as an input.

Although Amazon didn't describe other use-cases, the one that it demonstrated does have fairly odd vibes to it. Maybe Amazon meant it to be wholesome moment but hearing the synthesized audio of a dearly departed relative through a smart speaker just isn't my cup of tea.

Details about availability and other potential use-cases are scarce at the moment. It remains to be seen how Amazon will market this "feature" to a wider audience. Who knows, maybe Amazon will market it as the digital Ouija board for this generation.

Source: TechCrunch

Amazon wants Alexa to speak to you in the voice of your deceased relative

The Samsung ISOCELL HP3, a 1/1.4" format sensor, packs 200 million pixels (200MP), which are 12% smaller than the predecessor.

Samsung says the new model offers a more accurate and quicker auto-focusing thanks to the Super QPD technology. It uses a single lens over four-adjacent pixels for phase detection in horizontal and vertical directions. The ISOCELL HP3 also supports 8K at 30 FPS and 4K at 120 FPS video recording.

To provide a better low-light experience, the ISOCELL HP3 uses the Tetra2 technology for combining four or 16 pixels into one larger pixel. This approach allows the sensor to gather more light while taking 50MP or 12.5MP images.

Another noteworthy change in Samsung's latest creation is an upgraded Smart-ISO Pro feature that merges image information from two images taken with high and low ISO values to create one picture with a high dynamic range. The new version now uses triple ISO mode to widen the sensor's dynamic range and produce 14-bit images instead of the predecessor's 12-bit color depth.

Samsung plans to start producing the ISOCELL HP3 sensor later this year, so you can expect phones with improved 200MP cameras to hit the market somewhere in 2023.

Source

Google recently entered into commercial agreements with several news outlets in Europe. The company has reportedly signed a similar agreement with the Wikimedia Foundation. Google has agreed to pay Wikipedia for content displayed by its search engine that is sourced from the online encyclopedia.

Additionally, Internet Archive, a non-profit that runs a site called the Wayback Machine, will also receive commercial services, presumably from Google, for free. The Wayback Machine has been quite critical for Wikipedia as it routinely saves snapshots of the website. These have often been used to fix Wikipedia links and may have also helped to reverse the damage inflicted by internet trolls.

Google was the first paying customer for Wikimedia Enterprise, the commercial venture of the Wikimedia Foundation which was launched last year. "We're thrilled to be working with them both as our longtime partners," said Wikimedia's Lane Becker about the new development.

Wikipedia, one of the Internet’s largest and most visited websites, relies heavily on volunteers for content. The website also seeks donations to maintain its operations. This arrangement will not change. Simply put, the website will continue to remain free for viewers and editors. The foundation confirmed that it won’t be changing the arrangement it has for individual users.

Google, on the other hand, is one of Wikipedia’s biggest clients. Google pulls a massive amount of content from Wikipedia for its "Knowledge Panel", which is essentially a sidebar that accompanies the main search results. The Knowledge Panel offers short but important information that correlates with the search query. Google also cites Wikipedia in the information panels it adds to some YouTube videos.

Google has previously given money to Wikipedia, but it has been in the form of donations and grants, confirmed Google's Tim Palmer:

We have long supported the Wikimedia Foundation in pursuit of our shared goals of expanding knowledge and information access for people everywhere.

Needless to mention, the recently signed contract goes far beyond donations or charity. The foundation did not reveal the actual terms of the contract. However, it seems Google may have to pay Wikimedia every time the search results have information sourced from Wikipedia.

Source: Yahoo Finance

Google to pay for Wikipedia content displayed in 'Knowledge Panel' and search results

Mark Zuckerberg announced on Wednesday that Facebook Pay has officially become Meta Pay, but his post is mainly focused on what the company hopes to build in the future — “a wallet for the metaverse.” While Meta Pay is essentially the same tool as it was before, just with Meta branding, Zuckerberg says the company is working on something that will let users manage their identities, items, and payment methods while making their way through the digital world that Meta bets will be the future.

The company has hinted at these plans before — its May announcement of the Meta Pay rebranded also mentioned that it was “scoping out what a single wallet experience might look like” with regards to making payments and storing your identity and digital items. Zuckerberg’s post, though, focuses mostly on how the company’s metaverse wallet will enable proof of digital ownership; he mentions how it could be used to create and buy “digital clothing, art, videos, music, experiences, virtual events, and more.”

Zuckerberg admits the idea is in its early days.

The concept of a wallet that proves you own digital items sounds a lot like crypto wallets that hold NFTs. And while Zuckerberg’s post doesn’t mention the blockchain at all, the company has recently been working on integrating NFTs into Instagram and Facebook. Still, the company hasn’t really said whether it plans on actually building its metaverse using crypto or whether it’ll just take inspiration from it. (Given Meta’s history, the latter may be the best option; its attempt at making an actual cryptocurrency didn’t end well, and it’s reportedly looking into a digital currency that’s not based on the blockchain.)

Zuckerberg also says the plan is for Meta’s wallet to be interoperable, letting you bring your goods to “any metaverse experience.” While Zuckerberg has admitted that he’s not sure what interoperability will look like between companies, Meta recently helped form a standards group with the goal of getting everyone on the same page when it comes to language and tech related to the metaverse.

The wallet that Zuckerberg details is likely a ways off — especially if it’s meant to be interoperable based on standards that don’t exist yet. However, that hasn’t stopped the company from selling digital items in the meantime; it already has a store that sells clothes for your digital avatars.

Zuckerberg details his plan to move your digital items across the metaverse

Essentially, it is to make the job of gamers and shoppers easier. Currently, only comparison against Nvidia GeForce cards is present but it is possible that in the future, Intel Arc GPUs will also be provided. Also, right now, the tool only has data for the present generation of cards, ie, Radeon RX 6000 and Nvidia GeForce RTX 3000.

Here's how AMD describes its new GPU Comparion Tool:

At AMD, we enable consumers to choose the best GPU for their individual needs. See the performance of AMD Radeon™ 6000 Series GPUs in the latest games and settings using the comparison tool below. For full transparency, a competitive comparison with the Nvidia RTX 3000 Series graphics cards is also provided. All performance numbers have been generated and verified by testing in AMD internal labs.

Although first party benchmarks should always be taken with at least a pinch of salt, at first glance, the new GPU comparison tool does seem to be providing an unbiased comparison. For example, in Ubisoft's Assassins' Creed Valhalla, where AMD typically does exceptionally well, the red bars (representing Radeon GPUs) are longer. And the gap generally grows in favor of AMD as we lower the resolution:

Meanwhile, in Deathloop at 4K, the GeForce RTX cards completely dominate, which is indeed the expected performance behavior:

The tool also has an FPS/$ metric to determine the better value option. AMD says it used Newegg's prices to make this comparison.

In the footnotes of the page, AMD has provided more information on the test bench.

GPU Comparison Tool Configurations

• AMD Ryzen 9 5900X for AMD Radeon RX 6700 and higher configuration. AMD Ryzen 5 5600X for other configurations.

• AMD Ryzen 9 5900X for Geforce 3060Ti and higher configurations. AMD Ryzen 5 5600X for other configurations.

You can find the GPU comparison tool on AMD's official website at this link.

AMD releases GPU Comparison Tool so you need not look up third-party benchmarks

60+ fps gaming

The Arc desktop cards are currently available only in China, and as such, a Chinese reviewer took the card for a test run in order to see how well it measures up against AMD's RX 6400 and Nvidia GTX 1650. The Arc A380 Photon variant from Intel's AIB partner Gunnir was tested.

The card does excellent in 3DMark Time Spy, which is a synthetic DirectX 12 benchmark. It also beats the Navi 24 RX 6400 and 6500 XT in the ray-tracing Port Royal benchmark, though this was to be expected as the Navi 24 GPU has a really low amount of Infinity Cache with just 16MB. Here, the RTX 3050 blows everything away but it is also in a completely different price bracket compared to the Arc A380 and the RX 6400/6500 XT.

However, synthetic benchmarks is where the fun ends for the new Arc A380. In actual game tests, it looks like the A380 is beaten by both the RX 6400 as well as the Nvidia GTX 1650. The tested games are League of Legends, PUBG, GTA V, Tomb Raider, Forza Horizon 5 and Red Dead Redemption 2, three titles from DX11, two for DX12 and one for Vulkan.

Bear in mind though that the test platform here is a PCIe 4.0-capable TUF GAMING B660M-PLUS D4 motherboard. This means on a different system which is PCIe 3.0 only, the 6400 probably won't enjoy the performance lead that it does here as it comes with only four lanes of PCIe (x4).

Source and images: Shenmedounengce (Bilibili) via Olrak (Twitter), Wild_C (Twitter)

Review shows Intel Arc A380 is beaten by AMD RX 6400 and Nvidia GTX 1650

Although the term "metaverse" has failed to gain widespread public traction in the past few months, it seems that big tech isn't giving up just yet. Now, multiple organizations have joined hands to form a consortium called Metaverse Standards Forum. The list of founding partners can be seen in the graphic above, and you'll notice several heavy-hitters including Microsoft, Sony, Meta, Qualcomm, Nvidia, Epic Games, Huawei, Adobe, and more.

The idea behind the forum is to promote the development of open and interoperable standards necessary to build the metaverse. But of course, the press release doesn't define what tangible shape this metaverse would take.

Membership is open to any organization willing to take part in this effort, and it's free-of-cost. The consortium has mentioned the need to build spatial technologies such as interactive 3D graphics, VR, digital twins, content collaboration, and more to reach "new levels of scale and immersiveness". But again, the press release doesn't tangibly define what it means by that phrase.

Vice president of Metaverse at Meta Vishal Shah had the following to say about the collaboration:

Building a metaverse for everyone will require an industry-wide focus on common standards. The Metaverse Standards Forum can drive the collaboration that’s needed to make this possible, and Meta is committed to this work. Creators, developers and companies will all benefit from the technologies and experiences that will be made possible by common protocols.

Of course, an endeavor of this scale means that either metaverse will become mainstream or that it will fail spectacularly. We will find out more once forum meetings begin next month.

You might also notice that Apple is suspiciously absent from the list of founding partners for the Metaverse Standards Forum. Whether that is due to Apple's lack of interest in this area or the firm's nature of gravitating more towards proprietary solutions remains to be seen.

Meta, Microsoft, Sony form Metaverse Standards Forum, Apple nowhere to be seen

You can view the images below to see the old and new UI of Google News:

Old

New

There's also a new fact-checking section that shows the original claim next to a headline. Following the headline, it shows fact-checks from independent sources that tell whether the claim is false or not.

The "Your Topics" section is now customizable as well. Simply click on the Customize button on the top right and you can follow the topics you really care about, while removing the rest.

With the redesign, Google News is officially returning to Spain as well. In December 2014, Google dropped the local news offering in Spain as a law would have required paying publishers to use their content, including headlines.

However, a newspaper coalition pushed Google to restore news and Google announced plans for a revival last November. According to the new deal, Google could negotiate licenses with individual publishers instead of having to pay all of them.

The updated website is also supporting the Google News Initiative that is a campaign to support newsrooms for a long-lasting, diverse, and innovative news industry. Google has also started taking applications for the Global News Equity Fund that is meant for news for minorities and underrepresented groups.

Google is also funding $1 million for the Data-Driven Reporting Project that assists communities that have data-heavy investigations.

How Google News on desktop looked in 2002 versus how it looks now in 2022. Source - Google

This year, Google News marks its 20th anniversary. The redesign now gives local news more priority, while also making sure what people read is relevant to them.

Google News marks its 20th anniversary with updated UI, more focus on local news

Leak of next-gen Intel NUC combines a 12th-gen CPU with Intel’s discrete Arc GPU

The new device, as reported by the South China Morning Post, would benefit Chinese internet censors by detecting the brainwaves triggered by lewd content. Pornography is illegal in China.

The prototype device was made “for bad information detection”, said Xu Jianjun, director of the electrical engineering experiment centre at Beijing Jiaotong University, who published the work in Journal of Electronic Measurement and Instrumentation.

It was tested on 15 male university students who would wear the device while sat in front of a computer screen – with an alarm going off when explicit imagery was detected by the brain.

The developers say the new device can automatically adapt to the brainwaves of a human censor, able to filter out other signals that would come from emotions or weariness.

Chinese authorities employ professional censors called jian huang shi - porn appraisers – to check social media for what they deem inappropriate content.

Usually artificial intelligence can be used to detect explicit pictures and videos, but sometimes the algorithm makes errors. Human brains can still detect pornographic content more quickly and with greater accuracy.

Many of these appraisers are women, but since the test was only ran on men researchers believe there could be differences between genders. The accuracy of the machine is reportedly 80 per cent.

However, one unnamed researcher said that there could be ethical issues. “There is no law to regulate the use of such devices or protect the data they collected,” an individual from the University of Science and Technology of China in Hefei reportedly said.

Source

The landmark discovery, published in Nature today, was nine years in the making.

"This is the most exciting discovery of my career," senior author and quantum physicist Michelle Simmons, founder of Silicon Quantum Computing and director of the Center of Excellence for Quantum Computation and Communication Technology at UNSW told ScienceAlert.

Not only did Simmons and her team create what's essentially a functional quantum processor, they also successfully tested it by modeling a small molecule in which each atom has multiple quantum states – something a traditional computer would struggle to achieve.

This suggests we're now a step closer to finally using quantum processing power to understand more about the world around us, even at the tiniest scale.

"In the 1950s, Richard Feynman said we're never going to understand how the world works – how nature works – unless we can actually start to make it at the same scale," Simmons told ScienceAlert.

"If we can start to understand materials at that level, we can design things that have never been made before.

"The question is: how do you actually control nature at that level?"

The latest invention follows the team's creation of the first ever quantum transistor in 2012.

(A transistor is a small device that controls electronic signals and forms just one part of a computer circuit. An integrated circuit is more complex as it puts lots of transistors together.)

To make this leap in quantum computing, the researchers used a scanning tunneling microscope in an ultra-high vacuum to place quantum dots with sub-nanometer precision.

The placement of each quantum dot needed to be just right so the circuit could mimic how electrons hop along a string of single- and double-bonded carbons in a polyacetylene molecule.

The trickiest parts were figuring out: exactly how many atoms of phosphorus should be in each quantum dot; exactly how far apart each dot should be; and then engineering a machine that could place the tiny dots in exactly the right arrangement inside the silicon chip.

If the quantum dots are too big, the interaction between two dots becomes "too large to independently control them", the researchers say.

If the dots are too small, then it introduces randomness because each extra phosphorus atom can substantially change the amount of energy it takes to add another electron to the dot.

The final quantum chip contained 10 quantum dots, each made up of a small number of phosphorus atoms.

Double carbon bonds were simulated by putting less distance between the quantum dots than single carbon bonds.

Polyacetylene was chosen because it's a well-known model and could therefore be used to prove that the computer was correctly simulating the movement of electrons through the molecule.

Quantum computers are needed because classical computers cannot model large molecules; they are just too complex.

For example, to create a simulation of the penicillin molecule with 41 atoms, a classical computer would need 10^86 (1 followed by 86 zeros) transistors, which is "more transistors than there are atoms in the observable universe".

For a quantum computer, it would only require a processor with 286 qubits (quantum bits).

Because scientists currently have limited visibility as to how molecules function at the atomic scale, there's a lot of guess work in the creation of new materials.

"One of the holy grails has always been making a high temperature superconductor," says Simmons. "People just don't know the mechanism for how it works."

Another potential application for quantum computing is the study of artificial photosynthesis, and how light is converted to chemical energy through an organic chain of reactions.

Another big problem quantum computers could help solve is the creation of fertilizers. Triple nitrogen bonds are currently broken under high temperature and pressure conditions in the presence of an iron catalyst to create fixed nitrogen for fertilizer.

Finding a different catalyst that can make fertilizer more effectively could save a lot of money and energy.

Simmons says the achievement of moving from quantum transistor to circuit in just nine years is mimicking the roadmap set by the inventors of classical computers.

The first classical computer transistor was created in 1947. The first integrated circuit was built in 1958. Those two inventions were 11 years apart; Simmons' team made that leap two years ahead of schedule.

This article was published in Nature.

Source

It’s called Proteus Image: Amazon

Amazon has announced its “first fully autonomous mobile robot,” meant to move large carts throughout its warehouses. The robot is called Proteus, and Amazon says it can safely navigate around human employees, unlike some of its past robots that it kept separated in a caged area.

Amazon says Proteus robots have “advanced safety, perception, and navigation technology,” and a (strangely silent) video shows the robots shining a green light in front of themselves as they move around. When a human steps into the beam, the robot stops moving, then resumes after the person moves away.

The company has also announced several other robotic systems. One, called Cardinal, is a robotic arm that can lift and move packages weighing up to 50 pounds, which Amazon hopes to deploy in warehouses next year. The company says that its computer vision systems let it pick out and lift individual packages, even if they’re in a pile.

Amazon’s post also shows off tech that could let employees ditch the hand scanners they use to log barcodes. Instead, workers stand in front of a camera system that recognizes the packages without pausing to scan the label. There isn’t a lot of detail on how it works other than some combination of machine learning and a 120 fps camera system, but the effect is similar to what we’ve seen from the company’s Just Walk Out tech that lets it build cashierless stores. We’ve reached out to Amazon to ask exactly what the system is looking at and will let you know if we hear back.

As is often the case with new robot technology, there are potential labor concerns. Despite recent reports that Amazon could soon struggle to find workers, the company says it’s not looking to build robots instead of hiring people. A lead at Amazon’s robotics division explicitly told Forbes that “replacing people with machines is just a fallacy” that could lead to a company going out of business. However, robots could play a role in setting a pace of work that humans struggle to safely keep up with, something we’ve already seen happen at the company with automated management systems. The new scanning system especially seems like it could create unrealistic expectations about how fast workers should be moving.

For its part, Amazon claims all its new robots could actually help improve safety. Cardinal operates in places where workers would otherwise lift and twist heavy packages, a movement that can lead to injuries, and Proteus could “reduce the need for people to manually move heavy objects.” The company’s also working on a robot that would deliver containers to workers rather than having them bend or climb to reach items.

Amazon’s CEO recently promised to systemically address the injury rates at the company’s warehouses, though he has minimized reports that its workers are hurt at double the industry rate by calling the company’s rates “misunderstood.”

Amazon announces its first fully autonomous mobile warehouse robot

As has been the tradition, with the latest generation of PCIe too, the bandwidth of has doubled netting a total throughput of 128GT/s or 128Gbps uni-directionally on one lane (x1). Summing up, on PCIe x16 slot, like those for discrete graphics cards, the total theoretical throughput is 512GB/s bi-directionally. Meanwhile, an NVMe SSD, which typically pair with x4 PCIe slots, can provide up to 64GB/s uni-directionally. The final specifications will be released in 2025.

Here are the highlights of the new PCIe 7.0 standard:

• Delivering 128 GT/s raw bit rate and up to 512 GB/s bi-directionally via x16 configuration
• Utilizing PAM4 (Pulse Amplitude Modulation with 4 levels) signaling
• Focusing on the channel parameters and reach
• Continuing to deliver the low-latency and high-reliability targets
• Improving power efficiency
• Maintaining backwards compatibility with all previous generations of PCIe technology

With the new PCIe Gen7, PCI-SIG says that the 800 Gig Ethernet (800 GbE), and other similar data intensive markets can be addressed far more effectively. You can find more details on the official press release here.

PCIe 7.0 to bring insane speeds of 128GT/s for even faster SSDs, 800 Gig ethernet, and more

Miners, e-cafes, and scalpers are now trying to sell their hardware stock on streams and auctions. As a result, users can snag a second-hand GPU, such as the RTX 3060 Ti, for $350 or even less. Many popular graphics cards going for MSRP or even less is quite a sight to behold after astronomically high prices and scarce availability during the last two years.

As tempting as it might be to snag a powerful Nvidia or AMD GPU for a price lower than its MSRP, it is not the best idea to go after a graphics card that went through seven rings of mining hell. Potential buyers should be aware that the mining GPUs are often not in their best conditions after spending months in always-on, always-100% mode.

With manufacturers increasing their supply and prices going down like never before, you may better spend a little more and get a new graphics card with a warranty and peace of mind. As a bonus, you can enjoy the view of scalpers getting desperate to get at least some money from their stock.

Miners flood market with GPUs they no longer need as cryptocurrencies crash

Lenovo ThinkStation P360 Ultra crams 16 CPU cores and a GPU into a tiny desktop

In the not-so-distant past, the computing, mobile, and especially the automobile industry faced a dearth in the availability of semiconductors for their vehicles due to a fall in industrial outputs stemming from COVID-19 lockdowns. Even though their production schedules took a hit, Samsung and TSMC, the leading manufacturers, continued to develop faster chipsets for electronic devices in an ongoing bid to retain old customers as well as to attract new ones.

Having over 90% of advanced semiconductor device capacity, Taiwan remains the top chip producer. The rest is covered by South Korea as suggested by data from Semiconductor Industry Association.

Currently, both TSMC and Samsung Foundry are trying to attract industry leaders Apple, Nvidia, Qualcomm, AMD, MediaTek, etc. with their advanced nodes, upping the ante by employing Gate All Around (GAA) technology to offer 3nm chips, a notch faster than their already available 4nm chips.

Looking ahead, plans are underway by both competing firms to mass-produce 2nm chips intended to hit the market by 2025.

Source: Nikkei Asia via SamMobile

TSMC is building four new 3nm semiconductor chip plants

If you are even remotely interested in computer hardware and PCs in general, you have likely heard about AMD's Ryzen processors. That's because after being nearly non-existent in the CPU market for half a decade in the early 2010s, AMD came back swinging with its powerful yet affordable Ryzen CPUs. And at the heart of these Ryzen chips, powering them, is the Zen micro-architecture.

While Michael T Clark, a Corporate Fellow Design Engineer at AMD was the chief architect of Zen, overseeing the entire project was none other than legendary chip guru Jim Keller, who is hailed as genius by the majority of the tech community.

Earlier this year, when sharing a bit about his time at AMD during a conference called "Future of Compute" held by the Indian Institute of Science (IISc), Keller remarked that AMD "stupidly cancelled" the K12 ARM project.

If you can't recall what K12 was, basically it was sister project alongside Zen that Keller was tasked to run. And while Zen was based on x86, K12 was being built using the Arm architecture. At the time, AMD and Keller were working on a really interesting project dubbed "Skybridge" which married Arm and x86 on the same socket as they were being designed with pin compatibility.

And even though AMD did launch an Arm server chip in the form of the Opteron A1100 SoC, its custom K12 core never saw the light of the day.

At the conference, Keller also revealed that while at AMD, he worked on Zen 2 and Zen 3 as well. This means that the upcoming Zen 4 architecture is going to be the first AMD design which was not influenced by the veteran chip designer. After working at AMD, Keller jumped ship to rivals Intel where he allegedly worked on a Zen 5 killer project purportedly called "Royal Core".

Source: CSAChannel IISC (YouTube) via Wccftech

Legendary chip guru Jim Keller who helped design Zen calls AMD stupid for axing Arm K12

While that is indeed a good omen for those looking forward to M2-based products, one of the disappointing aspects of the M2 is the lack of AV1 codec support. The new chip supports up to 8K H.264 (AVC) and H.265 (HEVC), as well as its in-house ProRes video format for up to 8K playback.

The official press release from Apple says:

• The media engine includes a higher-bandwidth video decoder, supporting 8K H.264 and HEVC video.
• Apple’s powerful ProRes video engine enables playback of multiple streams of both 4K and 8K video.

This is despite Apple being a part of the Alliance for Open Media (AOMedia) consortium, which has been behind the AV1 codec since 2018.

However, all this could be set to change as Apple is seemingly working on AV1. This is suggested by changes spotted inside its AVFoundation's Core Media framework libraries with a new global variable listed as "kCMVideoCodecType_AV1".

You can find this in Apple's developer documentation here. For those wondering what AVFoundation is, it is basically Apple's framework for working with time-based audiovisual (AV) media (moving media) on the iOS, macOS, watchOS and tvOS. Hence, there may be broad support for AV1 finally coming to multiple Apple products throughout its ecosystem. If things materialize, this could be a major addition to Apple's arsenal as most of its competitors already support the codec. For example, Intel supports not just AV1 decoding but encoding too.

It will be interesting to see if Apple adds AV1 support to its new M2. Even if the chip lacks the necessary hardware, it will be great to see some form of software accelerated decoding.

Apple may finally be adding AV1 codec support to multiple products