We may earn a commission from links on this page.

If you’re looking for a robot vacuum, you’ve likely considered the Dyson 360 Vis Nav, a premium vacuum-only robot. I’d like to suggest that, instead, you consider the Switchbot K10+ Pro, a zippy little vacuum that is on sale for $299.99 (50% off) right now, compared to the Dyson’s $999.99 price tag.

I frequently read Reddit forums for people shopping for robot vacuums, and I’m generally surprised by two things. First, the number of people who are still looking for a robot that will only vacuum, not mop, and second, that Switchbot almost never comes up in the conversation. 

Dyson 360 Vis Nav

$999.95
at Amazon

$999.95
at Amazon

SwitchBot K10+ Pro Robot Vacuum Self Emptying for 90 Days Hands-Free, Super Quiet Mini Robotic Vacuum Cleaner with High Suction Power, Anti-Hair-Tangle, Works with Matter/Alexa/Siri, 2.4Ghz WiFi

$299.99
at Amazon

$599.99
Save $300.00

$299.99
at Amazon

$599.99
Save $300.00

Small size trumps smart design

Switchbot K10+ Pro Tower is tiny
Credit: Amanda Blum

Neither the 360 Vis Nav nor the K10+ Pro are traditional looking robots. The Dyson is U-shaped and sports bright colors. It has a deconstructed look, in gunmetal grey, red, and purple. The Switchbot is a mini robot, only eight inches across. Both designs deviated from the norm to give the robots better access to areas of the home, but despite all the research that I’m sure went into the Dyson’s shape, the Switchbot achieves the goal with more success. Simply, the diminutive size of the Switchbot allows it to make a tighter radius around chair legs and fit into smaller crevices between furniture.

A self-emptying tower is always better than no self-emptying tower

The 360 Vis Nav on the dock.
Credit: Amanda Blum

Personally, I’d need a real good reason to purchase a robot vacuum without a self-emptying tower—if the point of a robot is autonomy, I shouldn’t have to endure the indignity of emptying a dust canister over the trash can. The Switchbot K10+ Pro has such a tower, and like the robot itself, it is adorably small. Still, it is effective, evacuating all the debris from the K10+ Pro. The size has made it easy to park the tower almost anywhere; it is currently servicing my bedroom and closet, where the dock sits next to, and is dwarfed by, a pair of boots. 

The Dyson has no tower, replacing it instead with a charging dock. In my review of the 360 Vis Nav, I wrote about how annoying I found the fact that the dock could not stay put, and was often pushed around by the robot attempting to dock. While the dock is smaller than the Switchbot tower, it is garish, with black and white checks that apparently help the robot navigate to the dock. I’d be remiss if I didn’t explain that the Dyson robot had the best and easiest canister to empty of any vacuum I’d tried: You hold it inside the trash can, press a button and the dust is yeeted out at warp speed into the can. Still, I’d prefer to not bother with it at all.

The Dyson canister is well designed for minimal mess.
Credit: Amanda Blum

Features count, when it comes to apps

Credit: Amanda Blum

One of the charms of the K10+ Pro is that it behaves like a premium vacuum, particularly as you use the app, despite its low price. Similarly, an annoyance of the 360 Vis Nav is that it lacks many features of a premium vacuum, despite costing the same price as one. The Dyson app is, to be sure, beautifully designed with a clean, dark interface. Dyson has set the 360 Vis Nav up to be easy to use, and as such, there simply aren’t a lot of extras. You can set a schedule, decide the power with which the vacuum will work, and create/edit maps. But features like Pin and Go (set a pin and the robot proceeds there to clean) and remote control are absent, as are advanced settings for obstacle avoidance or a camera view from the robot. 

Credit: Amanda Blum

While some of those features, like Pin and Go, are very useful, they’re not necessary. However, I have to point out that most robots in the price range of the 360 Vis Nav would have those features and more. To be fair, Switchbot has the remote control feature (useful when a vacuum is stuck under a couch), and though it lacks Pin and Go, the Switchbot is less than half the price and has some cool features I haven’t seen elsewhere. You can pair NFC tags with various robot functions, which is neat. 

About equal in vacuuming prowress

Neither the 360 Vis Nav or the K10+ Pro are the best vacuums I’ve experienced on a robot. Switchbot’s K10+ Pro has about 20000Pa suction power on board, and Dyson’s 360 Vis Nav has 22000Pa. The K10+ Pro does a perfectly respectable job on micro and medium sized particles and struggles with large particles. What impressed me most was how effective the Switchbot was on carpet, which is where it ended up earning a permanent home in my house. Despite its size, it effectively picks up pet hair, human hair, dust, and most other items left behind and creates vacuum tracks, which soothe my inner clean freak. 

In contrast, the Dyson did great with micro and medium-small debris, and struggled with anything larger than that due to the design of the vacuum. The space between the plastic housing of the robot and roller that moves debris into the vacuum is simply too small to allow anything medium sized or larger. The robot also left nice tracks on the carpet, but struggled to get around as easily. The Dyson did have one advantage, though: the K10+ Pro doesn’t have extending arms of any kind, but the Dyson 360 Vis Nav has a side vent that allows it to get close to the wall when vacuuming. This was most noticeable over carpet, actually. The Dyson got a lot of dog hair right at the edge of the carpet that the Switchbot skipped.

An easy choice

If you were going off name recognition or reputation, Dyson is almost unparalleled in the vacuum industry for expensive but value-heavy models that vacuum better than others. I was disappointed in how little that translated to the 360 Vis Nav. In every single way that counts, the Switchbot K10+ Pro beats the Dyson 360 Vis Nav’s handily. Even better, it does it for $700 less.

A prominent venture capitalist has warned that the technology industry’s massive buildout of AI data centers risks becoming “irrational” and could end in disaster, particularly as companies pursue small nuclear reactors to power the facilities. Josh Wolfe, co-founder and partner at Lux Capital, compared the current infrastructure expansion to previous market bubbles in fiber-optic networking and cloud computing. While individual actions by hyperscale companies to build data center infrastructure remain rational, Wolfe said the collective effort “becomes irrational” and “will not necessarily persist.”

The warning comes as Big Tech companies pour tens of billions into data centers and energy sources, with Meta announcing just this week a deal to purchase power from an operating nuclear station in Illinois that was scheduled to retire in 2027. Wolfe said he is worried that speculative capital is flowing into small modular reactors based on presumed energy demands from data centers. “I think that that whole thing is going to end in disaster, mostly because as cliched as it is, history doesn’t repeat. It rhymes,” he said.

Read more of this story at Slashdot.

Agentic AI is the next big step forward for artificial intelligence: bots that don’t just respond with (often inaccurate) guidance and instruction, but can actually take actions on our behalf. Imagine ChatGPT, if it could research hotel prices and book your vacation, or Gemini being able to order the groceries you need on the web.

This technology is still in its nascent stage, but AI tools tend to progress rather rapidly these days, and we’ve already seen previews of browsers that can do your browsing for you, and bots that will find a new apartment for you and your roommates. These agents are going to be available very soon.

This all raises important questions: Do you want to give AI access to your calendar and credit card? Do you trust your favorite bot to pick the best restaurant in town? The promise is that you’ll save time on these tasks, but AI makes mistakes and can’t really be trusted for anything important right now.

Now we have an early AI agent for the iPhone, built by coder Rounak Jain (as reported by 9to5Google). It’s not available as an app, but you can test it out if you’re prepared to put in a bit of work and spend a few cents on OpenAI credits.

Our phone agent future?

The PhoneAgent assistant, which Jain originally put together as part of an OpenAI hackathon last year, is still rough—the programmer himself says this is experimental software that can “get things wrong sometimes.” However, it does give us a glimpse of where Apple, Google, OpenAI, and the rest may soon take us.

The tool connects to OpenAI models, so you can tap into the same kind of resources you would get with ChatGPT, besides being able to take action on your iPhone. This can be particularly handy when you need to create text in an email, message, or document, which can be generated with a quick prompt.

In his demo, Jain gets the app to take a photo, share it with his friend Ron, and attach a generic haiku to the message. In another request, the agent messages flight details to Ron (that guy again!), before booking an Uber to the airport, figuring out from the flight number that the Delta terminal was needed. System-level operations are also possible, such as opening the Control Center and turning on the flashlight.

Commands can be issued via voice or typed text, and a notification gets displayed when each task is completed—you can actually reply to this notification with follow-up commands (to turn the flashlight off, for example).

In my testing, the PhoneAgent was occasionally buggy and always quite slow, but that’s to be expected when you’re testing early-stage software like this. On the whole, it carried out my commands as intended: I got it to send a message to my friend Beth expressing how much I was looking forward to seeing her, and managed to take a photo with the Camera and open it up in the Photos editing interface with a single command.

I got AI to tell Beth how much I was looking forward to seeing her.
Credit: Lifehacker

App support is (understandably) a bit patchy, and I wasn’t able to use the agent to play the latest Self Esteem album on Spotify, or take a photo and automatically attach it to a note. Third-party integrations are going to be crucial, as devices like the Rabbit R1 have shown, but it’s not clear exactly how much access app makers will want to give to AI bots.

If this is what Siri and Gemini eventually evolve into, I can see it making a significant difference: Imagine being able to check where your next meeting is and get directions there with a single command, or having AI book your car service while you’re busy with something else. There are still a lot of hurdles to overcome, though—Jony Ive and Sam Altman, take note.

How to get it working

The process of getting PhoneAgent working is a little involved, but it’s not too difficult—and it’s free, apart from the OpenAI credits you need to spend to access its models. (I ran a few dozen commands through the software, which ended up costing me around $1 in credits.) First, you’ll need to install Xcode (from the macOS App Store) and GitHub Desktop on your Mac, and a free GitHub account, if you don’t already have one.

From GitHub Desktop, clone the PhoneAgent repository: Choose Clone a Repository from the Internet from the opening screen, or open the File menu and pick Clone Repository. Switch to the URL tab, paste “https://github.com/rounak/PhoneAgent.git” into the URL field, pick a destination folder on your Mac, then click Clone.

You need to get PhoneAgent installed in Xcode on a Mac.
Credit: Lifehacker

You’ve now got the PhoneAgent code on your Mac. Head to the folder you saved it to in Finder, double-click on the PhoneAgent.xcodeproj file, and it should open in Xcode. You’re then ready to transfer it to your iPhone to test it out: Connect your phone to your Mac with a cable, unlock the phone, and choose to trust the computer if prompted. Then head to Privacy & Security from iOS Settings, and turn on Developer Mode.

After your iPhone restarts, open the drop-down menu at the top of Xcode on macOS (to the right of PhoneAgent), and pick your iPhone from the list. Then, select PhoneAgent at the top of the left-hand navigation pane, switch to the Signing & Capabilities tab, and use Add Account to enter your Apple credentials. With that done, choose your name from the Team drop-down under Signing & Capabilities. You also need to change the Bundle identifier to something unique.

Next, select PhoneAgentUITests.swift from the left-hand pane, and look for the “func testLoop()” entry in the code (just a little way down). If you hover over the code line label to the immediate left, it should turn into a play button: Click this to run PhoneAgent on your iPhone.

Choose your iPhone as the app destination.
Credit: Lifehacker

The first screen you see will prompt you for an OpenAI key. In a web browser, sign up for an OpenAI account, then head to this page: Add payment details and buy credits to use the service (the minimum spend is $5). After that, switch to the API keys page, and click Create new secret key to get a key you can give to the PhoneAgent (make sure you keep a copy of this key somewhere safe, too).

You can then start entering commands by tapping the mic button or tapping the text at the bottom of the screen and telling the agent what you want it to do.

We may earn a commission from links on this page.

Since its premiere in 2021, HBO Max’s comedy Hacks has earned near-universal acclaim, an Emmy for Best Comedy Series, and an army of fans. The show follows legendary (read: “older”) stand-up comic Deborah Vance (Jean Smart) as she reluctantly teams up with young, disgraced comedy writer Ava (Hannah Einbinder) in a desperate bid to revive her career.

Hacks has maintained its sharp writing and great performances over four seasons—a rare TV trait. HBO confirmed that a fifth season is on the way, but no release date has been announced, so Hacks fans will have to put on their waiting shoes. In the meantime, here are five shows that can help fill the Hacks-shaped hole in your viewing schedule and your heart.

Broad City (2014-2019)

If you only watch one show while you wait for Hacks, make it Broad City. The two shows share comedy DNA: Hacks show-runner and co-creator Paul W. Downs was a writer and executive producer on Broad City; Hacks co-creator Jen Statsky was a writer and producer on Broad City, and Hack’s third co-creator, Lucia Aniello, served as executive producer, writer, and director on the earlier series. It’s not just the creatives behind the scenes: Both Hacks and Broad City focus on the experiences and relationships of female lead characters, but if Hacks is about surviving in showbiz, Broad City is about surviving your 20s, with a similar blend of messy friendships, self-discovery, and sharp observations. You can stream Broad City on Hulu, Paramount+, Roku, Disney+, and Prime Video.

The Studio (2025-present)

Seth Rogen-created comedy The Studio and Hacks may stream from competing platforms, but they take place in the same fictional universe. The Hacks crew makes a cameo on a Studio episode set at the Golden Globes, and Rogen appears as himself on Deborah’s late-night show on Hacks. It’s like that time George Jetson visited The Flintstones, but with more passive-aggression and fewer dinosaurs who wash dishes. Beyond the shared fictional universe, both shows offer an of-the-moment, meta take on the entertainment industry and feature characters navigating the choppy waters of the unique personalities and practices of the business of show. You can stream The Studio on Apple TV+.

The Marvelous Mrs. Maisel (2017-2023)

The Marvelous Mrs. Maisel is a spiritual prequel to Hacks. Rachel Brosnahan plays Miriam “Midge” Maisel, a comic on the come-up in early 1960s New York. While the timeline doesn’t work out for it to actually be Deborah’s backstory, the parallels are clear: Both women are whip-smart performers fighting for relevance in a male-dominated comedy world. Maisel captures the same spirit of a woman fighting to make it in the male-dominated world of comedy. Maisel’s stylized take on the entertainment business contextualizes the world Deborah Vance and Ava Daniels live in. It’s also an often painful reminder of how little has changed. You can stream Marvelous Mrs. Maisel on Prime Video.

Absolutely Fabulous (1992-1995)

Early 90s British cult-sensation Absolutely Fabulous features a faster pace and broader humor than Hacks, but the relationship between fashion P.R. maven Edina and her no-nonsense daughter Saffron mirrors the push and pull between Deborah and Ava in Hacks. Both Ab Fab and Hacks explore the challenges of aging in industries obsessed with youth, and both feature brash women fighting against irrelevance. Just don’t expect any sentimental heart-to-heart talks on the more biting Absolutely Fabulous. You can watch Absolutely Fabulous on Peacock, BritBox, Hulu, Prime Video, and Disney+.

Barry (2018-2023)

If you like Hacks‘ dark take on the entertainment industry, why not go even darker with HBO’s Barry? Barry’s title character, played by Bill Hader, is an excellent hitman who becomes a terrible actor with the help of a washed-up acting coach played by Henry Winkler. Like Hacks, Barry revolves around the collision between ambition and self-delusion, framed through the relationship between a struggling artist and an older mentor. Like Hacks, Barry excels at finding humor in the often brutal realities of trying to make it in entertainment—though Barry takes that darkness to considerably more extreme places. You can stream Barry on HBO Max.

If you have an old iPad and you’re interested in externalizing some of the thousands of photos in your iCloud account, this guide’s for you. Instead of buying a new digital picture frame, you can turn your retired iPad into a dedicated photo display without much effort. You’ll need to change a few settings, download a photo frame app and decide how you want to prop up your new display — which can be as simple as using a stand or as complicated as getting out the drill and modding an analog frame. Here’s how (and why) to upcycle your old iPad into a digital picture frame.

Upcycled iPad versus a digital picture frame

An actual digital picture frame has a few obvious advantages: the necessary software is already there, it’s display-ready, complete with a frame and matted display, and, as a unitasker, it won’t display Gmail notifications over your photos. But the resolution for most smart frames tops out at 1,280 by 800 pixels. Even the standard iPad (9th gen) from 2021 has a resolution of 2,160 by 1,620. Plus, your old iPad is already yours. A decent digital frame will cost you around $100. And, since your iPad is certainly not a single-purpose device — it can take a FaceTime call or play an episode of Poker Face if the need arises.

Amy Skorheim for Engadget

How to turn your old iPad into a digital picture frame

I figured I’d be able to use Apple’s built-in software to do something as simple as creating a looping slideshow of images, but the experience wasn’t quite what I was looking for. Playing a slideshow within the Photos app does so as a Memory, complete with music, and there’s no way to turn that music off (though you can mute it). You can loop a Memory, but you can’t randomize the images and the interval between images is far too short (anything less than ten minutes feels like yet another screen blinking at you). So, to turn an iPad into a randomized, always-on and silent digital picture frame with a decent display length, you’ll have better luck with a third-party app.

I tried a few and my favorite is Synched Photo Frame by Re-frame ($10, lifetime access). It doesn’t pack a ton of features — no transitions, no image effects, no time and date overlay options and it’ll only access images from your Photos app — but it does the best job of making photos look good on an iPad.

It lets you set the viewing interval from 10 seconds to 24 hours and there’s a programmable sleep function so it doesn’t beam smiling pictures into the 3AM void of your living room. If your iPad is oriented differently than a photo, it shows two side-by-side images (i.e., two portrait images next to each other if your iPad is in landscape or two stacked horizontal pics if your tablet is vertical). A clean white border surrounds each split image; full-screen images go edge to edge. Images are cropped slightly to fit, which means they may lose a few details around the borders, but I think the effect is more attractive than blurred extended edges or black bars. The two-up solution is how our top pick for a digital photo frame, the Aura Carver, handles images with different aspect ratios — but you can’t set that device in portrait mode.

LiveFrame is another option and it’s actually cheaper at $5 for lifetime access. It has more features like transitions and image filters, but there’s no option for having two images up at the same time like in Re-Frame. Instead it can blur the edges, add bars or (hideously) stretch the image. I also experienced a number of glitches with the app and found the navigation a little tougher than it should be. I also tried Digital Photo Frame Slideshow but, at $30 annually, it’s too expensive for my taste and it doesn’t have a sleep function.

How to prepare your iPad for photo frame mode

Step one: Create an album

Most photo display apps will pull from albums you select in Photos. You probably already know how to make a new album, but there are a couple of elements to consider as you create one for a digital photo frame app on an iPad. All instructions are for iPadOS 18. If you’re working with an older version, the steps may be slightly different.

Amy Skorheim for Engadget

1. Open Photos and tap the side bar icon in the upper left corner.

2. Under Albums, tap + New Album at the bottom of that list.

3. Name the album something memorable, as you’ll likely have to search for it within the photo frame app. Tap Create Album, then the blue plus sign.

4. If you want to only (or mostly) add a specific orientation (landscape or portrait), tap the up/down arrow icon at the bottom left of the pop up menu.

5. Tap View Options, then Aspect Ratio Grid. This will show the shape of each image so you can more easily select only the ones with the orientation you want. You can also use the View Options menu to zoom in on the thumbnails to get a better view of each picture.

6. Select the images you want and tap Add in the upper right.

Step two: Adjust your iPad settings

Adjusting a couple of settings will ensure your iPad keeps playing all day, interruption-free. Simply having the digital photo frame app playing should be enough to prevent your iPad from going dark, but your iPad may auto-lock when the app’s sleep mode kicks in, requiring you to unlock it in the morning to start the app back up. You’ll also want to make sure your slideshow isn’t interrupted with notifications from other apps.

1. Turn off Auto-Lock: Open Settings, tap Display & Brightness, tap Auto-Lock, select Never. You may get a warning that this setting will impact battery life, but since your iPad will stay plugged in, it won’t be an issue.

2. Turn off all Notifications: You’ll need to turn off notifications for every app on your iPad. Go to Settings> Notifications > Notification style, toggle off Allow Notifications for every app. Note: Using the Do Not Disturb Focus will enable DND on all your connected Apple devices, unless you toggle off Share Across Devices in Settings > Focus. But turning off notifications is still the better option.

Step three: Don’t let randos get at your stuff

Since you’ll probably want to display your new picture frame somewhere that people can see, there’s a risk of someone attempting to use your iPad. To prevent this, you can use an Accessibility feature called Guided Access. Apple designed it so parents could hand an iPad to their child without also giving them access to additional apps. We’re repurposing it for photo frame security to prevent anyone from navigating to your home screen or anywhere else. People can still tap on the screen to use the photo frame app’s controls or you can disable touch altogether.

Amy Skorheim for Engadget

1. Enable Guided Access: Open Settings, tap Accessibility then Guided Access. Set an unlock code, then set Display Auto-Lock to Never.

2. Turn on Guided Access: While the picture frame app is playing your slideshow, triple click the side button. A grey circle will appear, tap X to close it.

3. Enable or disable touch: If you still want to allow others to use functions within the photo frame app, toggle the Touch option on. If you want no response when the screen is touched, toggle it off. To access the Options menu, triple click the side button and enter your passcode. Tap Options in the lower left corner.

4. Note: Guided Access is turned on and off via the side button, so this option might not be feasible if you decide to put your iPad in a frame. One rather extreme workaround is to reset your iPad and set it up with a brand new Apple ID connected to family sharing that allows access to a shared Photos Library.

Amy Skorheim for Engadget

Put your new digital photo frame on display

This can be as simple as finding a stand you like and calling it a day. An Etsy seller called Event Frame makes custom frames for iPads. They aren’t cheap, going for around $100 each, but it’s one of the only companies I’ve found that actually offers model- and generation-specific sizing. I also considered the Snap Float Folio from Moft, as I’ve been impressed with some of the brand’s other origami-inspired accessories. If you’re crafty, you can take inspiration from this person’s Instructables and make your own frame from a physical media picture frame. Personally, I just got a right-angled USB-C cable and stuck my iPad in a collapsible stand I already had. The pictures are pretty enough.

Amy Skorheim for Engadget

This article originally appeared on Engadget at https://www.engadget.com/mobile/tablets/how-to-turn-your-old-ipad-into-a-digital-picture-frame-130017314.html?src=rss

We may earn a commission from links on this page. Deal pricing and availability subject to change after time of publication.

The Google Pixel Tablet is down to $359 right now for the 256GB version, the lowest price it’s been so far, according to price trackers. Originally priced at $499, this $140 discount puts it in a good spot for anyone who wants a tablet that pulls double duty.

Google Pixel Tablet

$359.00
at Amazon

$499.00
Save $140.00

$359.00
at Amazon

$499.00
Save $140.00

The Pixel Tablet is designed to be a solid Android tablet first, but it also works as a smart display when you’re not using it. If you grab the speaker dock, you get a stand that charges the tablet and gives it a serious upgrade in sound quality.

The tablet itself is clean and straightforward. It has an 11-inch display with 2,560 by 1,600 resolution, a 60Hz refresh rate, and can hit 500 nits of peak brightness—plenty for reading, streaming, or casual browsing even in brighter rooms. Battery life is reasonable for the price range. In PCMag’s tests, it ran about five hours and 25 minutes of HD video playback at full brightness, which could stretch longer under less demanding use. The cameras—both front and rear—are 8MP with a wide 84-degree field of view and can shoot 1080p video at 30fps: fine for video calls and casual snaps, but nothing you’d replace your phone camera with.

The Pixel Tablet runs Android well, with many popular apps optimized for the larger screen. And like other Pixel devices, Google promises at least five years of security updates, which isn’t something every mid-range tablet can guarantee. If you want a tablet that doesn’t gather dust when you’re not actively using it, the Pixel Tablet’s ability to turn into a smart display gives it more reason to stay in sight—and the current price makes it a lot easier to justify.

Google’s self-driving taxi service Waymo has surpassed 10 million total paid rides, marking a significant milestone in the transition of autonomous vehicles from novelty to mainstream transportation option. The company’s growth trajectory, WSJ argues, shows clear signs of exponential scaling, with weekly rides jumping from 10,000 in August 2023 to over 250,000 currently. Waymo is on track to hit 20 million rides by the end of 2025. The story adds: This is not just because Waymo is expanding into new markets. It’s because of the way existing markets have come to embrace self-driving cars.

In California, the most recent batch of quarterly data reported by the company was the most encouraging yet. It showed that Waymo’s number of paid rides inched higher by roughly 2% in both January and February — and then increased 27% in March. In the nearly two years that people in San Francisco have been paying for robot chauffeurs, it was the first time that Waymo’s growth slowed down for several months only to dramatically speed up again. Waymo currently operates in Phoenix, Los Angeles, and San Francisco, with expansion planned for Austin, Atlanta, Miami, and Washington D.C. The service faces incoming competition from Tesla, which plans to launch its own robotaxi service in Austin this month. Waymo remains unprofitable despite raising $5.6 billion in funding last year.

Read more of this story at Slashdot.

For too long, the AI hardware landscape has felt like a one-horse race, with NVIDIA’s H100 GPUs dominating the field. But here in Bend, Oregon, where a healthy dose of competition keeps everyone sharp, AMD is increasingly proving that not only are they in the race, but they’re making a strong bid for the lead. Their recent MLPerf Training submission, highlighting the prowess of the AMD Instinct MI300 and the upcoming MI325X GPUs, signals a significant shift in the AI training arena. In an era of hardware availability constraints, all AMD has to be is “good enough” – and they are clearly outperforming that bar, pushing the boundaries and offering compelling alternatives.

A Major Milestone: AMD’s MLPerf Training Debut

AMD’s recent entry into the MLPerf Training v5.0 benchmarks marks a pivotal moment. This wasn’t just a quiet submission; it was a loud declaration of intent, showcasing the strength of their platform with competitive performance on both the Instinct MI300X and the Instinct MI325X GPUs. The chosen workload, fine-tuning the Llama 2-70B-LoRA model, is no lightweight; it’s a widely adopted and critical task for customizing large language models in today’s generative AI landscape. This highlights AMD’s focus on real-world, highly relevant AI workloads rather than abstract benchmarks.

The debut results are particularly impressive: the Instinct MI325X platform actually outperformed an average of six OEM submissions using NVIDIA’s H200 platform by up to 8% when fine-tuning Llama 2-70B-LoRA, a widely adopted workload for customizing large language models. This isn’t just “good enough”; this is superior performance in a crucial segment. The Instinct MI300X platforms also delivered competitive performance against NVIDIA H100 on the same workload, validating both GPU platforms in the Instinct MI300 Series as strong contenders across a broad range of training needs, from enterprise applications to cloud-scale environments.

Industry-Wide Validation and Ecosystem Strength

What truly amplifies AMD’s MLPerf debut is the industry-wide validation from its ecosystem partners. In addition to AMD’s own submission, six OEM partners submitted MLPerf Training results using AMD Instinct MI300 Series GPUs. This demonstrates that AMD Instinct performance is not just a lab benchmark; it’s reproducible and robust across a wide range of diverse platforms and configurations. These partner results reinforce the consistent performance of the MI300X and MI325X across various infrastructure environments, even setting new industry milestones in several cases.

A standout example comes from Supermicro, which became the first company ever to submit training results to MLPerf using a liquid-cooled AMD Instinct solution (Supermicro liquid-cooled MI325X MLPerf). Their liquid-cooled Instinct MI325X platform achieved an impressive time-to-train score of 21.75 minutes, not only showcasing top-tier performance but also highlighting the thermal efficiency and scaling potential of advanced cooling solutions in dense AI deployments. MangoBoost further raised the bar with groundbreaking multi-node submissions, demonstrating the scalability of the MI300X platform in distributed AI infrastructure. Dell, Oracle, Gigabyte, and QCT also contributed strong results, collectively proving the power, flexibility, and openness of AMD Instinct GPUs, which are clearly being embraced and pushed to new heights by innovators across the ecosystem.

ROCm Software: The Open-Source Secret Sauce

The remarkable performance gains showcased by AMD are significantly powered by the rapid evolution of AMD ROCm v6.5 software, AMD’s open software stack. Over the past year, ROCm has matured into a robust training platform. In the Llama 2-70B-LoRA fine-tuning task, ROCm was key to unlocking top-tier performance on Instinct MI325X and MI300X GPUs, with critical improvements like Flash Attention, Transformer Engine support, and optimizer-level tuning. AMD has even released optimized docker containers used in this work, along with easy-to-use reproduction instructions, empowering developers. These results provide strong evidence that an open software stack, backed by deep engineering optimizations, can indeed rival and, in some cases, outperform proprietary ecosystems. For customers scaling generative AI, ROCm software delivers the performance, flexibility, and developer support needed to move fast and efficiently.

A Significant Generational Leap

In AI infrastructure, staying current is essential. Organizations investing in AI expect meaningful performance gains with each new platform. The AMD Instinct MI325X GPU sets a new standard for what a generational leap can deliver. In platform (8-GPU) fine-tuning workloads like Llama 2-70B-LoRA, the Instinct MI325X offers up to a 30% performance uplift compared to the Instinct MI300X, a significant generational advancement that dramatically accelerates large-scale AI development and shortens training timelines.

Wrapping Up: The Future is Open and Accelerated with AMD

AMD’s debut MLPerf training submission marks a major milestone, signaling that their Instinct MI300 Series GPUs provide customers with compelling new options for scaling AI innovation. With competitive, and in some cases, superior performance on key AI workloads, and a robust, open ROCm software stack, AMD is rapidly closing the gap on NVIDIA’s long-standing dominance. In a market where hardware availability can often be a constraint, AMD’s ability to be “good enough” has now evolved into being a genuine leader in crucial AI training workloads. As models grow larger and AI use cases expand, AMD’s commitment to open, high-performance solutions promises to empower customers to train, fine-tune, and deploy AI faster and more flexibly than ever before. The momentum behind AMD Instinct is real, and it’s just getting started

Team Asobi is releasing five new levels for Astro Bot, and you can get all of them for free. The developer has announced that the brand new levels will drop on July 10 at 9AM Eastern time. Like the speedrun levels and the previous levels the developer released at no cost, the new ones will be part of the Vicious Void Galaxy. Four of the new levels are called Twin-Frog Trouble, Suck It Up, Handhold Havoc and High Inflation, but Team Asobi is keeping the last one a secret. Each one will come with its own Special Bot, including a “revengeful newcomer.” The developer also hinted that finishing all five levels will lead to “something really cool.”

Astro Bot is a 3D platformer, wherein you control a small robot named Astro across levels set in different galaxies and planets. It features unique and creative uses of the DualSense controller, such as having to hold down the analog triggers and physically tilting the device in order to move Astro’s spaceship. Team Asobi has released several free levels before these new ones since the game launched in late 2024, and so far, Astro Bot has over 90 levels that you can play. 

In addition to announcing the freebies at PlayStation State of Play, Team Asobi has also revealed that it’s launching an updated version of the Astro Bot Limited Edition DualSense wireless controller later this year. The developer promised more details in the future, but for now, you can see that the controller will come with Astro’s smiling eyes on its touchpad. 

Team Asobi / Sony

This article originally appeared on Engadget at https://www.engadget.com/gaming/playstation/astro-bot-is-getting-five-new-and-free-levels-this-july-123008291.html?src=rss

I’ve written before about why hill repeats are an excellent thing to add to your run training, and how to get started. But one of the trickier aspects is finding the perfect hill, or even knowing how steep the incline of that hill is. A tool I recently started using makes this super easy. 

The tool is officially called On the Go Map, but you can easily search for it by its nickname, “How Far Did I Run?” To use it, click the start and end points of your hill. (I use the walking icon, which snaps the path to walkable roads and paths, but you can also draw straight lines if you prefer.) 

You can use this tool to measure the distance and elevation of any running route—like a free version of MapMyRun—but when it comes to your favorite hill, all you have to do is create that mini route that goes up your hill, and then click the elevation icon in the bottom left corner of the screen. (It looks like a mountain.) Then a pane opens up at the bottom of the screen, showing the overall grade of the hill as a percentage. 

This percentage is the same as what you’ll see when you select the incline on a treadmill. 1% or 2% would be a very mild hill, a gentle rolling grade. My favorite “I would like to die doing hill repeats today” hill is 8%, and anybody going over 10% on a regular basis is a certified maniac. Most gym treadmills top out around 12%. 

In the example picture at the top of this article, I’ve selected a portion of hill in a local park. On the Go Map helpfully tells me how long that segment is (0.17 miles, so a little less than a lap of a quarter-mile track) and how much elevation I’ll gain going up it (85 feet). To find the percent grade, look in the upper right hand corner of the elevation pane. This hill is a 9.5% grade. 

If you haven’t found the perfect hill to measure yet, tap the menu icon and choose “terrain” as your map style. Any road that crosses multiple contour lines is going to be a steep one. Pan around until you find a promising looking hill, and then measure it. 

Giant Squid has announced that it’s latest game, Sword of the Sea, will come to PlayStation 5 and PC on August 19. It’ll also appear on PlayStation Plus from day one for Extra and Premium Subscribers. The fantasy adventure game marries surfing mechanics with the atmospherics of the developer’s previous games like Abzû and Journey. 

Giant Squid showed off more gameplay in a new trailer as part of PlayStation’s June 4 State of Play. “At its core, Sword of the Sea is all about surfing,” the developer wrote. “You’ll be carving your way across mysterious moving waves of sand on the ancient and powerful hoversword.” Other scenarios include dreamy ice-scapes and oceans full of flying dolphins. 

Giant Squid again teamed with creative director Matt Nava and composer Austin Wintory, who also worked on Abzû and Pathless. “As a surfer and snowboarder myself, I have felt the rush of movement and seen the beauty of nature in a way that only these extreme sports can deliver,” Nava wrote about Sword of the Sea. 

This article originally appeared on Engadget at https://www.engadget.com/gaming/sword-of-the-sea-launches-august-19-120037636.html?src=rss

A bitter fight over alleged corporate espionage involving two of Silicon Valley’s hottest startups took a new twist on Tuesday, after $12 billion HR software company Deel claimed arch-rival Rippling had directed one of its employees to “pilfer” the company’s assets by posing as a customer. From a report: The latest claim comes after Rippling alleged earlier this year that a staff member had been spying on behalf of Deel. The employee locked themselves into a bathroom and smashed their phone with an axe when confronted with allegations, according to their own testimony.

In new legal filings seen by the Financial Times, Deel has countered by arguing that: “Rippling has been actively engaged in a carefully co-ordinated espionage campaign, through which it infiltrated Deel’s customer platform by fraudulent means and pilfered the company’s most valuable proprietary assets.”

Read more of this story at Slashdot.

We may earn a commission from links on this page.

Welcome to “Cookbook of the Week.” This is a series where I highlight cookbooks that are unique, easy to use, or just special to me. While finding a particular recipe online serves a quick purpose, flipping through a truly excellent cookbook has a magic all its own. 

You’ve mastered roast chicken, crushed barbecue pork, perfected your homemade biscuits, and your grandma’s recipe for tortellini is immaculate. Some folks think a good main dish is the ticket to a perfect meal, but they’re not seeing the big picture. One of the most crucial components to a stellar meal is none other than using the right sauce—but sauce from scratch is a skill set unto itself. 

Sauce making takes a keen palate for balancing flavor, consideration for texture, sometimes careful emulsification, and awareness for proportions. It’s no wonder we often reach into the condiment shelf of our fridge door, no one wants to screw up a sauce. I think having a helpful sidekick with reliable sauce recipes might be just what we need. That’s why I chose Saucy for cookbook of the week. 

A bit about the book

Written by Ashley Boyd, Saucy is a tantalizing book of a wide variety of sauce recipes—herby, tangy, sweet, and umami. This cookbook is made to complement, not take over your diet. Even the petite dimensions of the book signal that it’s a team player; the backup dancer that elevates your main book of recipes.

The chapters are organized by flavor profile, texture, and sometimes by ingredient, like creamy or tomato-based. One of my favorite features of this cookbook is the section in the front that gives you sauce pairings based on what your main dish is. When I set out to make lunch or cook dinner, I’m thinking of my main meal first. Am I eating dumplings tonight or salad? How should I finish off that roast chicken? Once I have that all figured out, I can flip through this cookbook and mosey over to the Chicken and Fish section of pairings and decide which of the 15 suggested sauces sounds good to me.

After you get through the multiple chapters of flavor-packed sauces and drizzles, Boyd goes a step further and supplies you with a few classic recipes in case you don’t actually have that main dish part figured out. There are recipes for pancakes, bread pudding, roast chicken, steak, and a few others. 

The sauce (and dish) I made this week

I fully believe a well-matched sauce can take a meal to the next level. However, I rarely make them from scratch. (Unless it’s gravy for mashed potatoes, obviously.) So that leaves me chronically buying sauces and tending to my library of condiments in the fridge. I’ll be honest, homemade sauce always tastes better but I settle for the bottled stuff for ease. 

I took this cookbook review as an opportunity to break out of my comfort zone and make myself some fresh sauce. It wasn’t easy to start; I was interested in almost every sauce I saw in this book. I had leftover rotisserie chicken to eat so I decided a sandwich would be nice. I landed on the Lebanese Garlic Sauce, and thought of all the delicious wraps and pitas I’ve had with that sort of punchy garlic sauce. Lunch was decided.

Part of the deciding factor was that this garlic sauce only required four ingredients—garlic cloves, salt, canola oil, and lemon juice—and I already had them all. To make life even easier, I used my Vitamix Ascent X5 to take care of all that finicky emulsification business. This sauce is a true aioli and meant to be bright white, thick, and spreadable like mayo, but a vampire’s nightmare. An aioli can take patience lest it break, like all emulsions. If you don’t have a high-powered blender, I recommend using an immersion blender to make life easier.

Credit: Allie Chanthorn Reinmann

I tossed all of the garlic cloves into the blender along with the salt and pulsed it together until it was finely minced. Then I started the process of drizzling in oil and alternating with lemon juice, scraping the sides, and repeating as the cookbook directions indicated. 

I may not have picked the funnest nor the fastest sauce, but I do believe I picked the most complementary sauce for my rotisserie chicken wrap. Alone, this garlic sauce is way too strong. It’s almost harsh with raw garlic. A bad recipe? No, friends. This is exactly why sauce-making is an art. Sauce is meant to support the main dish. It needs to be strong in small doses. You wouldn’t eat hot sauce as a bowl of soup. I hope.

Credit: Allie Chanthorn Reinmann

I smeared thin swipes of the creamy garlic sauce along one side of my Lebanese pita bread and piled chicken, tomato, lettuce, pickles, yogurt, and harissa onto the other. I was a little worried the garlic would be too sharp, too raw, and ruin my lunch, but instead every bite was a victory lap. It’s the best version of that pita that I could have made. The Lebanese Garlic Sauce saved my lunch from being just OK. The only reason I don’t have post-lunch-sadness right now is because that recipe made about 12 ounces of sauce and I can’t wait to use it again.

A great cookbook to elevate everything you make

Saucy is a great cookbook for any home cook that’s ever had the feeling like something’s missing. That feeling that your burger, chicken wing, or salad is on the brink of perfection but it’s falling flat for indescribable reasons. You need a sauce. 

The recipes in this book are largely straightforward with four to 10 ingredients and most of them are easy to find in major grocery stores. Each one gives you short procedural steps, followed by instructions on how to store that particular sauce. Sauces are powerful, so you often only need a little bit. The rest of the recipe can be sealed up in a jar to live in the fridge for at least five days and some for up to three months. 

If you see yourself becoming a sauce maven, I suggest saving some jam jars so you can keep your collection organized in the fridge. Actually, a nice jarred sauce might make a great gift for Father’s Day. And why not throw in a copy of Saucy too?

How to buy it

Saucy is available as an ebook, but I insist on grabbing the hardcover if you can. It’s wonderfully compact with some really mesmerizing photography. If you have a bookstore nearby, leave the computer behind and see if they have Saucy in the cookbook section. If not, maybe they can order it for you.

Saucy: 50 Recipes for Drizzly, Dunk-able, Go-To Sauces to Elevate Everyday Meals

$9.99
at Amazon

$9.99
at Amazon

Sony has just unveiled the first wireless fight stick designed by the company itself. It doesn’t have an official product name yet, but it’s currently codenamed Project Defiant. The arcade-style controller uses the PlayStation Link wireless technology to connect to the PS5 and the PC. Sony says the technology provides an ultra-low latency connection that enables “precise in-game response” to every press on its buttons and every movement on its joystick. Players can also plug the device into their PC or their console via a USB-C connection if they want to ensure that there’s no lag 

The company said it custom designed its controls and used “toolless interchangeable restrictor gates (square, circle, and octagon),” as well as mechanical switches for its buttons. Sony also used the same touch pad on the fight stick as the one found on the DualSense wireless controller. In addition to being able to control the action on screen, the stick has the ability to wirelessly wake up the PS5 by pressing the PS button in the same way a DualSense controller can. 

The fight stick has a storage compartment for its restrictor gates and the PS Link USB adapter. It also sounds like the device will ship with a sturdy carrying case, so you can take it safely with you when needed. Sony has yet to announce Project Defiant’s pricing and release date, but it’s planning to make the fight stick available for purchase sometime in 2026, perhaps in time for players to be able to use it to play Marvel Tōkon: Fighting Souls. The tag fighting game, which is also coming out next year, was developed by the studio behind the Guilty Gear franchise.

This article originally appeared on Engadget at https://www.engadget.com/gaming/playstation/project-defiant-is-sonys-first-wireless-fight-stick-for-ps5-and-pc-113009444.html?src=rss

Under the Trump administration, the Food and Drug Administration is eagerly embracing artificial intelligence tools that staff members are reportedly calling rushed, buggy, overhyped, and inaccurate.

On Monday, the FDA publicly announced the agency-wide rollout of a large language model (LLM) called Elsa, which is intended to help FDA employees—”from scientific reviewers to investigators.” The FDA said the generative AI is already being used to “accelerate clinical protocol reviews, shorten the time needed for scientific evaluations, and identify high-priority inspection targets.”

“It can summarize adverse events to support safety profile assessments, perform faster label comparisons, and generate code to help develop databases for nonclinical applications,” the announcement promised.

Read full article

Comments

Volvo has introduced a new seatbelt technology that can customize the protection it provides in real time. The “multi-adaptive safety belt” system, as the automaker is calling it, uses data input from both interior and exterior sensors to change protection settings based on various factors. It can take a person’s height, weight, body shape and seating position into account, as well as the direction and speed of the vehicle. The system can communicate all those information to the seatbelt “in the blink of an eye” so that it can optimize protection for the passenger. 

If the passenger is on the larger side, for instance, they will receive a higher belt load setting to reduce the risk of a head injury in the event of a serious crash. For milder crashes, someone with a smaller frame will receive a lower belt load setting to prevent rib injuries. Volvo didn’t specifically say if the system also takes the position of a seatbelt on women into account, since it doesn’t always fit right over a woman’s chest. However, the automaker explained that the system expands the number of load-limiting profiles to 11. Load limiters control how much force a seatbelt applies on the body during a crash. Typically, seatbelts only have three load-limiting profiles, but Volvo expanding them to 11 means the system can better optimize the protection a passenger gets. 

Volvo used information from five decades of safety research and from a database of over 80,000 people involved in real-life accidents to design the new safety belt. The system was also created to incorporate improvements rolled out via over the-air software updates, which the company expects to release as it gets more data and insights. 

“The world first multi-adaptive safety belt is another milestone for automotive safety and a great example of how we leverage real-time data with the ambition to help save millions of more lives,” said Åsa Haglund, head of Volvo Cars Safety Centre. “This marks a major upgrade to the modern three-point safety belt, a Volvo invention introduced in 1959, estimated to have saved over a million lives.”

Volvo engineer Nils Bohlin designed the modern three-point seatbelt and made its patent available for use by other automakers. The company didn’t say whether it’ll be as generous with the multi-adaptive safety belt, but the new system will debut in the all-electric Volvo EX60 midsize SUV sometime next year.

This article originally appeared on Engadget at https://www.engadget.com/transportation/volvo-is-introducing-the-first-multi-adaptive-seatbelt-technology-on-the-ex60-ev-070017403.html?src=rss