AMD Engineer Confirms RDNA3 GPU Power Consumption Will Increase

For the past few months there’s been a constant drip-drip-drip of rumors about the upcoming GPUs from AMD and Nvidia. The most surprising element of those rumors, at least on the Nvidia side of the fence, has been a dramatic increase in power consumption. Nvidia’s next-gen Ada Lovelace RTX 40-series were originally reported to require almost double the power of its current Ampere cards. Those numbers have since come down a bit, but the latest rumors still hint at a 600W flagship GPU.

That’s hard to even fathom. On the AMD side of the rumor mill though, things have been much more quiet. We know the company is moving to a chiplet-based GPU design for the first time, but will AMD be cranking up power levels too? According to an AMD engineer in a recent interview, yes, it will. By how much, he doesn’t say, only that this is the new trend.

The insight was revealed by AMD’s Sam Naffziger in an interview with Tom’s Hardware. He’s been with the company 16 years and is a Senior Vice President, Corporate Fellow, and Product Technology Architect. He has been working on the company’s chiplet designs for years now, which were formerly reserved for its Zen, Epyc, and Threadripper CPUs. With RDNA3, that design will now be applied to GPUs for the first time. Naffziger explains that it’s not a binary question of whether to make a chiplet power efficient or more powerful. Though those two goals are at odds with each other, Naffziger explains that improving efficiency gives AMD more headroom to maximize performance too.

This slide depicts the steadily decreasing rate of node advancement and silicon area scaling. (Image: AMD). The limited scaling depicted for analog and SRAM is part of an overall trend of diminishing marginal returns. AMD and Intel have adopted chiplet-based designs partly in response to this trend. 

When Naffziger was asked whether RDNA3 would follow Nvidia’s rumored tactic of increasing power levels, he replied that AMD really has no choice but to follow the same path. “Performance is king, but even if our designs are more power-efficient, that doesn’t mean you don’t push power levels up if the competition is doing the same thing. It’s just that they’ll have to push them a lot higher than we will.” He was also asked what is the main impetus for graphics cards needing a lot more power all of a sudden, and he said it’s due to twin forces of soaring compute demand and shrinking gains in node advances.

“It’s really the fundamentals of physics that are driving this,” he said. “The demand for gaming and compute performance is, if anything, just accelerating, and at the same time, the underlying process technology is slowing down pretty dramatically — and the improvement rate. So the power levels are just going to keep going up. Now, we’ve got a multi-year roadmap of very significant efficiency improvements to offset that curve, but the trend is there.”

AMD hasn’t revealed much about RDNA3 just yet, but it has put a few numbers in writing. One of those is that it expects a greater than 50 percent improvement in performance-per-watt. That’s a huge leap, but it’s what it achieved in previous generational transitions. That was going from Polaris/Vega to RDNA, then to RDNA2. However, those were back in the heady days of GCN 14nm for Polaris, and TSMC 7nm for RDNA. AMD will be tapping TSMC’s 5nm node for RDNA3, but little else is known about power requirements just yet.

Twitter user AMDGPU’s mockup of the Navi 31 package. (Image: @AMDGPU_)

All that’s been alleged so far is that the company’s RDNA3 GPUs will feature a seven-chiplet design. The company has announced it’s “rearchitected” its compute unit. We also know it will offer a “next-gen” version of its Infinity Cache. Still, Naffziger was clear that although it will be increasing power levels, it probably won’t be going to same extremes as Nvidia. This tracks with its recent bragging about its superior efficiency compared to the company’s Ampere GPUs. Obviously, this is a mantle AMD is keen to hold onto in the future.

Now Read:

• Deathloop: Putting AMD’s FidelityFX Super Resolution 2.0 to The Test Against Nvidia’s DLSS
• MSI Yanks Video Showing Off AMD’s New AM5 Socket (Updated)
• AMD’s RDNA3 7900XT Flagship to Offer 384-bit Memory Bus, 24GB of VRAM

Microsoft Adds a Ton of New Gaming Features to Edge

Most gamers don’t really think about a web browser as part of their gaming setup. Heck, most gamers don’t even use a browser for gaming. Microsoft is looking to change that though by making its Edge browser the go-to choice for gamers. Yes, it sounds kind of weird, but Microsoft announced today it’s added a lot of new gaming features to its browser. We know, just stick with us here. Also, if you’re an Edge user who doesn’t care about browser gaming, Microsoft has added a new performance-boosting feature for PC games as well.

The addition of the new gaming features were announced in the company’s latest blog post. It’s part of Microsoft being “all in” on gaming, despite what you might have heard. Microsoft says it’s focusing on a 20-year roadmap to deliver the best gaming experience possible, on any platform, meaning PC, Xbox, phone, and tablet. Today, it’s starting with Edge.

Microsoft has added a new gaming-centric homepage to the browser. The gist is as soon as you launch your browser, you’re shown all your games so you start gaming immediately. This also includes gaming news and info, guides, livestreams, and so forth. You can also customize what’s shown on this page as well. If you’re logged into your Xbox account, you can have access to your Xbox Cloud Gaming library too. If you are an Xbox Game Pass Ultimate member, you can launch streaming games right from the homepage.

Microsoft Edge’s new gaming homepage. (Image: Microsoft)

Once you’re actually streaming a game via your browser, you can benefit from this next new feature. It’s called Clarity Boost and it does exactly what it says. According to Microsoft, it’s a “spatial upscaling enhancement that makes streaming games look clearer and sharper while playing in Microsoft Edge on Windows 10 and 11.” The blog says it uses client-side scaling to improve the quality of the video stream. Microsoft didn’t get into any nerdy details on what it’s doing here. Still, the sample it posted shows a clear improvement in image quality. As a way to tempt people to try it out, it’s offering people the chance to play Fortnite without any subscription required. To enable or disable Clarity Boost, you just click the (…) in the left-hand corner of the game.

Microsoft Clarity Boost. (Image: Microsoft)

For those of you who play on PC with locally installed games, there’s something for you too. Microsoft has added a new power-saving feature to Edge for PC gamers. It essentially puts Edge into snooze mode when it detects a game being started. This prevents the browser from using precious resources while you’re gaming. When you quit the game, the feature detects it and automatically shuts it off. This works in both Windows 10 and 11, using Edge. To enable it just go to the Settings tab, then System and Performance.

Finally, Microsoft has also added a bunch of free-to-play casual and arcade games (top). When you select “games” from the dropdown under the (…) in the browser they appear on the right-side. It features a roster of old and new games that open right in the browser. Some of the titles include Microsoft Jewel, Microsoft Mahjong and Microsoft Edge Surf. Personally, we clicked on Solitaire just because we’re old school, and we were reminded of how much fun it is, deadlines be damned.

All of these features are available now. Just update Edge and you’ll see them. For us, after we updated we were using version 103.0.1264.37 (Official build) (64-bit).

Now Read:

• Microsoft to Add VPN Service to Its Edge Browser
• Microsoft Edge Gets New Anti-Chrome Pop-Ups
• Microsoft Will Put Internet Explorer Out of Its Misery Next Year

Amazon’s Alexa Will Soon be Able to Use a Dead Person’s Voice

(Photo: Lazar Gugleta/Unsplash) Alexa, Amazon’s widely-critiqued yet still widely-loved smart speaker, will soon be able to use the voice of a deceased loved one. All it will need is a one-minute recording of the person you lost.

Amazon introduced the technology at Amazon re:MARS 2022, its annual AI event centered around machine learning, automation, robotics, and space. Alexa AI head scientist Rohit Prasad referred to the upcoming feature as a way to remember friends and family members who have passed away. 

“While AI can’t eliminate the pain of loss, it can definitely make their memories last,” Prasad said. 

Prasad demonstrated the feature using a video of a child asking Alexa if his grandmother could finish reading him a story. In its regular Alexa voice, the smart speaker obliged; then the grandmother’s voice took over as the child flipped through his own copy of The Wizard of Oz. Though of course there’s no way for the viewer to know what the woman’s real voice actually sounds like, the grandmother’s synthesized voice admittedly sounded quite natural, speaking with the cadence of your average bedtime story reader. 

It’s important to note the context in which this technology was introduced. Amazon led into its announcement of the voice synthesizing feature by emphasizing the “companionship relationship” many people have with Alexa. “In this companionship role, human attributes like empathy and affect are key for building trust,” Prasad said. He went on to explain that this has become increasingly important throughout the COVID-19 pandemic, the byproduct of which has been significant physical and social isolation. But why would Amazon be so invested in building companionship and empathy between Alexa and its users?

(Photo: Andres Urena/Unsplash)

The technology is reminiscent of that used to bring the likeness of deceased movie stars into new media. James Dean—or a CGI version of him—is expected to “star” in an upcoming film despite having died decades ago. A fashion magazine recently sparked confusion and outrage after featuring a CGI “photoshoot” of Marilyn Monroe. Some argue that because the actual person being imitated never agreed to participate in such media, it isn’t right to use and manipulate their likeness; others argue that because the stars are deceased, their likenesses are essentially fair game. 

It’s also worth questioning whether Alexa’s upcoming voice synthesizing feature could be used to imitate the voices of living people. It’s doubtful that Alexa would know whether the recording you’ve offered is of a living person or someone who’s passed away. Beyond using the feature to prank someone who’s still around (as, I’ll admit, I would probably do), the imitation of a living person’s voice could be used to thwart all sorts of security measures or otherwise “hack” someone’s smart home. Given the feature’s recent announcement, however, these concerns have yet to be addressed. 

As of now, it’s unclear whether this feature is still in the research phase or if it will roll out to everyday Alexa users soon.

Now Read:

• Analysis Shows Amazon’s Alexa Collects More Data Than Any Other Smart Assistant
• Leaked Amazon Research Says It Might Run Out of People to Hire by 2024
• Amazon is Giving Away Over 30 PC Games on Prime Day

Microsoft Halts Sales of Emotion-Reading Tech, Limits Facial Recognition Access

(Photo: Turag Photography/Unsplash) Microsoft has decided to retire its emotional assessment technology, as well as place restrictions on the use of its facial recognition systems. 

In a blog post published Tuesday, the software giant announced that it would be sunsetting facial analysis tools that claim to identify emotional states and personal attributes, like a person’s gender or age. Before, these capabilities were freely available within Azure Face API, Computer Vision, and Video Indexer. Now those who already have access to Microsoft’s emotional-reading features will have one year of use before their access is revoked.

“We collaborated with internal and external researchers to understand the limitations and potential benefits of this technology and navigate the tradeoffs,” wrote Sarah Bird, Azure AI’s principal group product manager, in the post. “In the case of emotion classification specifically, these efforts raised important questions about privacy, the lack of consensus on a definition of ‘emotions,’ and the inability to generalize the linkage between facial expression and emotional state across use cases, regions, and demographics. API access to capabilities that predict sensitive attributes also opens up a wide range of ways they can be misused—including subjecting people to stereotyping, discrimination, or unfair denial of services.”

Microsoft says it will continue the use of its emotion and facial recognition software for Seeing AI, which “narrates the world” for those with little to no seeing ability. (Photo: Microsoft)

Microsoft has also chosen to place restrictions on who will be able to use its facial recognition technology in the future. Going forward, anyone interested in using these tools will need to submit an application detailing their project, after which Microsoft will approve or deny access. Microsoft said it will independently assess the benefits and risks of continuing to use both emotion and facial recognition tools for “controlled accessibility scenarios,” such as its own Seeing AI. 

The company’s decision follows a widespread effort to investigate the social implications of emotion and facial recognition tech. Well-known human rights groups have recently called out Zoom for its mood recognition AI, which ranks unsuspecting sales call recipients on an “emotional scorecard.” Similarly, some are concerned that Intel’s emotion-reading e-learning software will incorrectly target or alienate students who are deemed “distracted” or “confused.” While these calls for action often go ignored, they sometimes work: last month the IRS dropped its ID.me program, which required users to upload a video selfie to access government services, after receiving near-unanimous backlash. Now Microsoft seems to have joined the ranks of those willing to backpedal—to do the right thing or for the sake of keeping customers, who knows. 

Now Read:

• Ukraine is Using Facial Recognition to Contact Kin of Dead Russian Soldiers
• Clearview Plans Facial Recognition Database That Knows Every Person on Earth
• Google Bans Deepfake Training on Colab

Samsung Rumored to Begin 3nm Mass Production Next Week

A news agency in Korea is reporting Samsung is about to make a major announcement. The world’s second-largest silicon foundry will reportedly enter mass production of its 3nm process next week. In doing so, Samsung becomes the first global foundry to reach this milestone in advanced node production. It notably beats its main rival TSMC to the punch, as the Taiwanese powerhouse isn’t expected to ramp 3nm until later this year. Intel will reach 3nm sometime in 2023 with its Intel 3 process.

The move by Samsung marks its transition from FinFet to Gate-All-Around (GAA) transistors. It’s the first major foundry to make the switch. The company says it will allow for a 45 percent reduction in chip area compared with FinFet. It will also allow for 30 percent more performance and will require 50 percent less power. Samsung has been developing its 3nm process for some time now and even showed it to President Biden recently. However, reports indicated it was suffering from poor yields. This is expected when developing a new process node. Still, it was reported it was only getting between 10 and 20 percent yields at 3nm previously. If it’s moving into mass production, also called high volume manufacturing (HVM), it has probably made strides on this front.

The transition from FinFET to GAA transistors is just over the horizon. (Image: Samsung)

Samsung first revealed its 3nm plans way back in 2019 when it said it was in the Alpha stage. The company announced it had chosen Nanosheets as its preferred design. This is one of the two GAA designs that are possible; the other is Nanowires. Samsung’s design is called MBCFET, which stands for Multi-Bridge Channel Field Effect Transistor. Interestingly, when it announced its plans it trumpeted a 35 percent boost in performance. However, the latest reporting has lowered that number by five percent. Still, this reporting it not an official Samsung announcement, so that might change if and when Samsung officially announces it.

One thing to note here is just because Samsung is the first foundry to reach HVM at 3nm doesn’t mean it necessarily has a leg up on its competition. As we reported previously on an article about TSMC, some analysts believe customers might not want to be first in line for a brand new, radical change such as GAA transistors. Instead, they might be more inclined to stick with “old” 3nm technology from TSMC, which is sticking with FinFet for its 3nm process. As we reported earlier this week, it’ll be using a new configurable design called FinFlex. This lets customers customize the design on-die for various benefits such as power consumption, performance, and chip size.

Since Samsung builds chips that its customers must integrate into their own products, it’ll be a little while before we see actual 3nm products for sale. Still, it’s an exciting development to see a company the size of Samsung finally move beyond FinFet for the first time. The agency report also states Samsung is in the early stages of its developing its 2nm process as well. That won’t be coming online until 2025, however.

Now Read:

• TSMC Founder Says US Silicon Fab Expansion Plans ‘an Exercise in Futility’
• Apple’s M2 Pro Could Be First to Use TSMC 3nm Silicon
• Silicon Wafer Supplier Says Its 300mm Capacity is Sold Out Through 2026

Meta CEO Says It Expects One Billion People to Be in The Metaverse by 20230

In a recent interview with Jim Cramer of Mad Money fame, Meta CEO Mark Zuckerberg opined on various metaverse-related topics. As we all know, the metaverse is seen by the company as the next big thing in computing. So big, in fact, that it changed its name from Facebook to reflect its pivot towards virtual worlds. In the discussion Zuckerberg  addressed how he envisions many aspects of the technology’s future.

In the interview, Zuck revealed Meta’s lofty ambitions. He said he hopes there will be at least a billion people in the metaverse by the end of the decade. In addition, he thinks each of those people will be spending hundreds of dollars a year on digital bobbles. He estimated people will be “buying digital goods, digital content, different things to express themselves,” he said. These digital goods could also include avatar clothes, virtual home decorations, or items to adorn a virtual conference room. Doing the math, this means Mark thinks his company will be pulling in hundreds of billions of dollars in metaverse revenue eventually. That’s several factors beyond its current annual revenue, which is around $117 billion. Still, 2030 is a long ways away, so anything is possible.

Overall, Zuckerberg thinks the metaverse will eventually have a “massive economy,” especially for creators. This is a rich comment coming from him, as the company recently came under fire for its surprisingly high commissions on sales of user-generated content in Horizon Worlds. The company announced it would be taking a 47.5 percent cut of all sales in the game. The fee is so egregious Tim Cook from Apple called Zuckerberg a hypocrite in a rare public rebuke. As far as when this economy will arrive, it seems like it might take awhile. Zuckerberg said the metaverse will follow the same trajectory as Facebook and Instagram. By that he means it wants there to be a billion or two users on the platform before it “scales monetization.”

Elsewhere in the discussion, Zuckerberg said the company might need to build an operating system for the metaverse. According to BusinessInsider, he said, “I think over the long term, we will need this level of integration between hardware and software and an operating system, just to deliver on what we want to build.” Facebook/Meta has never built an operating system, though he’s probably right that it will need to figure out how to do that. Apple is also reportedly working on a similar venture dubbed rOS for Reality Operating System. The two companies are anticipated to become head-to-head competitors, pardon the pun, when Apple eventually releases its virtual reality (VR) headset.

This is just the latest in PR blitz by Zuckerberg to create some hype for the metaverse. Earlier this week he demonstrated some of the futuristic designs it’s working on for next-gen VR hardware. None of the prototypes he displayed are anywhere close to coming to market, much like the metaverse itself. Still, it’s interesting to see what the future could hold for a head-mounted display. Also, if Zuckerberg’s ambitions are realized, it would mark a massive turnaround for its VR division. Known as Reality Labs, it’s been a money pit the company keeps shoveling ad dollars into. The division reportedly lost $10 billion last year, with $3 billion of that in the fourth quarter alone.

Now Read:

• New Study Shows Working in The Metaverse Sucks
• iPod Inventor Tony Fadell: ‘F*ck The Metaverse’
• Analyst: Apple’s Upcoming Headset Will not be Metaverse-Friendly

First Intel Arc A380 Desktop Benchmarks Are Disappointing

We’ve all been waiting with bated breath to see some real, not-from-Intel benchmarks for the company’s new Arc desktop GPUs. Its first entry-level desktop card just launched in China, and now a review has been posted online. Finally, we have some real benchmarks, including both synthetic and in games. Overall, the results do not look good for Intel.

To recap, the A380 is Intel’s entry-level desktop GPU in the Arc Alchemist family. It’s priced at roughly $150 or so, and was designed to compete against the Radeon RX 6400 and Nvidia GTX 1650 GPUs. All of these GPUs are made for 1080p gaming. They exist in a niche where a person is on a tight budget but needs more oomph than integrated graphics. Intel’s GPU actually has very decent specs for this price range. It also offers some technology usually only found on more expensive GPUs too. This includes its version of super sampling that’s like DLSS and FSR. It also supports hardware-based ray tracing and fields 6GB of RAM, which is rare in this class of GPUs.

Despite these advantages, the A380 was slower than the AMD and Nvidia GPUs in every gaming test. According to Wccftech, the results were posted by Bilibili content maker, Shenmedounengce. The games used for testing included: League of Legends, GTA 5, PUBG, Shadow of The Tomb Raider, Forza Horizon 5, and Red Dead Redemption 2. The card being tested is the Gunnir Arc A380 6GB Photon.

(Image: Bilibili, Shenmedounengce)

What’s interesting here is the AMD card (red bar, above) is a single-slot GPU with very modest specs. It has just 4GB of VRAM, a 64-bit memory bus, and is rated to consume 20W less power than the A380 too. Yet, as you can see it tops Intel’s fledgling effort in every benchmark. What’s even stranger is the Gunnir card being tested is actually a boosted version of Intel’s A380 too. It has a higher maximum clock rate of 2,450Mhz compared to 2,000MHz for the reference design. It also has a higher TBP of 92W versus 75W for Intel’s design.

The A380 wasn’t all bad though. Despite its losses in gaming tests, it did notch some victories in synthetic benchmarks. For example, in 3DMark Timespy it trounced the competition, and even beat the more expensive RX 6500 XT. It also naturally faired quite well in the Port Royal test, which is a benchmark for ray tracing. Generally speaking though, we tend to put more weight into real-world gaming tests versus synthetic tests like this, for obvious reasons.

Given the fact that Intel’s GPU beats its competitors on paper, but not in gaming tests, it’s a bit of a head scratcher. As we’ve written before, it’s quite possible Intel is having issues getting its drivers up to snuff. That’s obviously a huge factor in the GPU’s performance. Though it doesn’t look good now, Intel’s strategy of a “staggered” global rollout is starting to make more sense given the issues that are evident. Hopefully there’s still a lot more room for improvement, and Intel can get its drivers polished before it releases the more powerful Arc cards later this summer.

Now Read:

• Intel Freezes All PC Hiring For at Least Two Weeks
• Intel Flashes Arc Desktop Graphics Card at Texas Gaming Tournament
• Intel Shows Off Tile-Based Meteor Lake, Ponte Vecchio CPUs at Vision Event