Month: July 2024
AMD just made big improvements with AFMF 2, a new version of its tech to make all PC games run more smoothly
Want a free frame rate boost for all your PC games? You’re in luck if you’ve got an AMD GPU…
AMD has released a new version of its Fluid Motion Frames feature, a frame generation tech to speed up frame rates with PC games, and the sequel sounds like a promising step forward – albeit it’s only in preview for now.
AMD Fluid Motion Frames 2, or AFMF 2 for short, works with RX 6000 and RX 7000 GPUs, and is described as a ‘significant upgrade’ benefiting from substantial optimization work for better quality and performance.
Specifically, one of the key upgrades here is lower latency for the frame generation process (which is the insertion of extra frames to artificially boost the frame rate) – less latency means a greater overall smoothness with the results of AFMF 2.
AMD also observes there’s now better consistency during fast motion, when the action is hotting up in a shooter, for example.
This is tied up in a new setting called ‘Search Mode’ which, as Team Red notes, enhances the overall image quality, and also improves how ‘fallback’ works in AFMF 2. Fallback is when frame generation is temporarily turned off during fast on-screen movement, to avoid the jitter that might otherwise be visible when things are moving very quickly in a game.
We’re told AI has been employed to fine-tune AMD’s algorithm for frame generation, and there’s an ‘auto’ setting in Search Mode that automatically reduces the occurrence of fallback to improve smoothness at 1440p or 4K resolutions. (This measure isn’t applied when running at 1080p, as it won’t produce the best results here).
There’s a second new mode, namely ‘Performance Mode,’ which offers a new ‘performance’ setting as an alternative to ‘quality,’ the latter of which was the default for AFMF 1. Performance is designed to reduce overheads and it’s the default setting for an integrated GPU (on supported Ryzen CPUs with integrated Radeon graphics), helping to keep frame rates higher on these lesser GPUs – whereas quality remains the default choice for those with a discrete graphics card.
Also new is support for Vulkan and OpenGL games (alongside existing DX11 and DX12 support), as well as support for borderless full-screen gameplay.
Finally, you can now enable AMD Radeon Chill and use it with AFMF 2 to prevent the frame rate from exceeding the maximum refresh rate of your monitor (and therefore any danger of image tearing or other glitches).
AMD recommends AFMF 2 is used with Radeon Anti-lag when you don’t expect frame rates to exceed the monitor’s refresh rate (after they’ve been boosted, that is), but to use Chill instead if the frame rate is going to leap higher than the refresh rate. It’s a choice of one or the other, and Anti-lag helps to reduce latency, and as noted, AFMF 2 already does some good work on this front.
(Image credit: Future / John Loeffler)
Analysis: AFMF pain killers
Note that right now, this is still a technical preview, and AFMF 2 is only available in a preview driver which, as ever with pre-release software, could be a bit wonky. It shouldn’t be too long before it progresses to the finished version of AMD’s Adrenalin graphics driver, though. (It took four months for the original AFMF to go from preview to release, but we expect the sequel to progress more quickly).
As we mentioned above, there are some key improvements here. Reduced input lag is huge, as too much in the way of lag pretty much spoils the smooth feel of gameplay when using frame generation, and has been a major complaint about the tech in the past. According to AMD, the improvements here are in the order of 28% lower latency, on average, based on testing at 4K resolution in Cyberpunk 2077 (with the ‘ray tracing ultra’ setting).
In-house benchmarks, of course, need to be taken with some caution, and we’ll have to see how AFMF 2 performs ourselves – when it’s fully finished, of course.
Remember that AFMF is a standalone take on frame generation, which is applied at a driver-level (as part of the Adrenalin driver). It is also a part of FSR, but it’s not to be confused with FSR 3.1 – it’s just one element of the latter, and AFMF can also be used on its own. The key difference is FSR must have support coded for it in games, but standalone AFMF (or AFMF 2) is baked into the driver, and so works for all PC games (with newer AMD graphics cards, that is – cross-platform support for the best rival GPUs is lost, obviously, as this is a driver tech).
There are caveats, as you might imagine – don’t expect the results to be as good as FSR 3, as they won’t be. But for a cover-everything solution, AFMF is a nifty piece of tech, and version 2 helps to address some of the bigger stumbling blocks with the original feature. Namely those worries about increased input lag affecting some games in particular, such as fast-paced shooters, and the tweak to address fallback and motion-related concerns, too.
Via VideoCardz
You might also like
The best cheap graphics cards of 2024: top GPUs on a budgetTop gaming PCs: great rigs for serious PC gamingBest PC games of 2024: must-play titles you don’t want to miss
Microsoft wants Congress to outlaw AI-generated deepfake fraud
Illustration: Alex Castro / The Verge
Microsoft is calling on members of Congress to regulate the use of AI-generated deepfakes to protect against fraud, abuse, and manipulation. Microsoft vice chair and president Brad Smith is calling for urgent action from policymakers to protect elections and guard seniors from fraud and children from abuse.
“While the tech sector and non-profit groups have taken recent steps to address this problem, it has become apparent that our laws will also need to evolve to combat deepfake fraud,” says Smith in a blog post. “One of the most important things the US can do is pass a comprehensive deepfake fraud statute to prevent cybercriminals from using this technology to steal from everyday Americans.”
Microsoft wants a “deepfake fraud statute” that will give law enforcement officials a legal framework to prosecute AI-generated scams and fraud. Smith is also calling on lawmakers to “ensure that our federal and state laws on child sexual exploitation and abuse and non-consensual intimate imagery are updated to include AI-generated content.”
The Senate recently passed a bill cracking down on sexually explicit deepfakes, allowing victims of nonconsensual sexually explicit AI deepfakes to sue their creators for damages. The bill was passed months after middle and high school students were found to be fabricating explicit images of female classmates, and trolls flooded X with graphic Taylor Swift AI-generated fakes.
Microsoft has had to implement more safety controls for its own AI products, after a loophole in the company’s Designer AI image creator allowed people to create explicit images of celebrities like Taylor Swift. “The private sector has a responsibility to innovate and implement safeguards that prevent the misuse of AI,” says Smith.
While the FCC has already banned robocalls with AI-generated voices, generative AI makes it easy to create fake audio, images, and video — something we’re already seeing during the run up to the 2024 presidential election. Elon Musk shared a deepfake video spoofing Vice President Kamala Harris on X earlier this week, in a post that appears to violate X’s own policies against synthetic and manipulated media.
Microsoft wants posts like Musk’s to be clearly labeled as a deepfake. “Congress should require AI system providers to use state-of-the-art provenance tooling to label synthetic content,” says Smith. “This is essential to build trust in the information ecosystem and will help the public better understand whether content is AI-generated or manipulated.”
Illustration: Alex Castro / The Verge
Microsoft is calling on members of Congress to regulate the use of AI-generated deepfakes to protect against fraud, abuse, and manipulation. Microsoft vice chair and president Brad Smith is calling for urgent action from policymakers to protect elections and guard seniors from fraud and children from abuse.
“While the tech sector and non-profit groups have taken recent steps to address this problem, it has become apparent that our laws will also need to evolve to combat deepfake fraud,” says Smith in a blog post. “One of the most important things the US can do is pass a comprehensive deepfake fraud statute to prevent cybercriminals from using this technology to steal from everyday Americans.”
Microsoft wants a “deepfake fraud statute” that will give law enforcement officials a legal framework to prosecute AI-generated scams and fraud. Smith is also calling on lawmakers to “ensure that our federal and state laws on child sexual exploitation and abuse and non-consensual intimate imagery are updated to include AI-generated content.”
The Senate recently passed a bill cracking down on sexually explicit deepfakes, allowing victims of nonconsensual sexually explicit AI deepfakes to sue their creators for damages. The bill was passed months after middle and high school students were found to be fabricating explicit images of female classmates, and trolls flooded X with graphic Taylor Swift AI-generated fakes.
Microsoft has had to implement more safety controls for its own AI products, after a loophole in the company’s Designer AI image creator allowed people to create explicit images of celebrities like Taylor Swift. “The private sector has a responsibility to innovate and implement safeguards that prevent the misuse of AI,” says Smith.
While the FCC has already banned robocalls with AI-generated voices, generative AI makes it easy to create fake audio, images, and video — something we’re already seeing during the run up to the 2024 presidential election. Elon Musk shared a deepfake video spoofing Vice President Kamala Harris on X earlier this week, in a post that appears to violate X’s own policies against synthetic and manipulated media.
Microsoft wants posts like Musk’s to be clearly labeled as a deepfake. “Congress should require AI system providers to use state-of-the-art provenance tooling to label synthetic content,” says Smith. “This is essential to build trust in the information ecosystem and will help the public better understand whether content is AI-generated or manipulated.”
Mortgage Refinance Rates Fall: Today’s Refinance Rates, July 30, 2024
Several benchmark refinance rates ticked downward. Refinance rates change daily, but experts say rates will stabilize in the coming months.
Several benchmark refinance rates ticked downward. Refinance rates change daily, but experts say rates will stabilize in the coming months.
Mortgage Rates Increase for Prospective Buyers: Mortgage Rates on July 30, 2024
A handful of important mortgage rates inched up. Will the housing market ever recover?
A handful of important mortgage rates inched up. Will the housing market ever recover?
TCL’s Samsung Frame TV rival is called Nxtframe, and it’s even got a B&O ace up its sleeve
TCL’s brought in Bang & Olufsen to give its frame TV some serious audio oomph.
Stop me if you’ve heard this one before: it’s a TV, but when you’re not using it as a TV it’s an art gallery. Nope, it’s not Samsung’s The Frame TV (or even Samsung’s Music Frame speaker). It’s TCL’s Nxtframe, aka the A300, and it’s reportedly making its debut at IFA in Berlin in September 2024.
The most interesting thing about this TV might not be how it looks, however, but how it sounds. That’s because TCL has swapped audio suppliers: instead of its usual Onkyo audio speaker system, it’s going to be using a Bang & Olufsen-designed soundbar (see our Beosound Theatre review for a brief taster of what the Danish audio specialist can do in this area) and subwoofer that you can tune with B&O’s BeoSonic software.
As FlatpanelsHD reports, the new TV is the first product from a six-year agreement that will bring Audio By Bang & Olufsen to TCL’s high-end TVs sold worldwide. According to Bang & Olufsen CEO Kristian Teär, “We’ll collaborate to bring premium sound experiences to TCL’s customers and enhance the overall product experience” – and more partnerships may be imminent. Teär says that B&O is committed to “expanding our licensing partnerships business”.
You’ve been Framed: the Art TVs to watch in 2024
The Nxtframe has just been unveiled in China and joins a veritable gallery of frame-style TVs: we’ve also seen models from Hisense and Roku in the form of Hisense’s $999 CanvasTV and Roku’s Backdrops mode, which is coming via software update to its Pro Series QLED TV this summer. Detailed specifications for TCL’s model haven’t yet been announced and we don’t as yet know what content will be available for you to gaze at.
That could be important, because Samsung’s secret weapon with The Frame TVs isn’t so much the tech (as impressive as it is), it’s the Art Store, which is so exclusive that even Samsung’s other TVs can’t have it. While it costs extra, it’s not a huge expense – and it means you don’t have to create your art collection manually.
But in terms of the TVs themselves Samsung is still the market leader here, and The Frame continues to evolve: adding anti-glare matte screens in 2023 was a big improvement, and the 2024 models also have Pantone Validated certification for color accuracy and a 60Hz mode to reduce power consumption when in Art Mode. The CanvasTV is anti-glare too but doesn’t currently have the Pantone seal of approval.
You may also like
See our roundup of the best TVs of 2024Not sure about a whole new TV, just need better audio? See our pick of the best soundbarsOr go bigger (and more immersive) with one of the best Dolby Atmos soundbars
HPE-Juniper deal comes a step closer with EU approval
Europe could grant regulatory permission for HPE’s Juniper Networks takeover despite UK CMA investigation.
Hewlett Packard Enterprise (HPE) is expected to receive unconditional EU antitrust approval for its acquisition of Juniper Networks, according to sources familiar with the matter (via Reuters).
The European Commission, which is set to announce its decision on August 1, has not commented on the potential approval. Neither have HPE or Juniper.
That being said, Britain’s Competition and Markets Authority (CMA) has launched an initial investigation into the takeover, and the second phase of its investigation could come into play on August 14.
EU expected to approve HPE-Juniper acquisition
HPE hopes that the acquisition, first announced in January 2024, will strengthen its position in the market for AI and hybrid cloud applications by incorporating Juniper’s networking solutions.
The company could cite Cisco’s dominant market position as a counterargument against any competition concerns.
In the UK, the CMA’s initial review will determine if the acquisition warrants a more in-depth investigation. It’s unclear whether Europe’s decision could influence the CMA’s.
If HPE’s deal is approved and goes ahead, it would significantly expand the Texas-headquartered tech giant’s networking division. HPE also purchased Californian Aruba Networks for $3 billion in 2015.
The Juniper deal has not been without scrutiny, though. Some have expressed concerns about reduced market options post-merger. The potential elimination of overlapping products between Juniper and Aruba could reduce options further.
With the right regulatory approval, HPE and Juniper’s merger could close by late 2024 or early 2025. HPE, which saw a modest 3% increase in quarterly revenue to $7.2 billion in the three months that ended April 30, could get a healthy boost following the takeover, while Juniper is expected to benefit from HPE’s broader reach.
More from TechRadar Pro
UK regulator set to probe HPE’s $14bn Juniper Networks acquisitionThese are the best small business serversHere are the best cloud hosting providers
The growth of Confidential Computing
How confidential AI dwells at the intersection of AI and Confidential Computing, by bridging the gap between AI and Zero Trust policies.
Over the last five years, Confidential Computing has evolved and matured. Today, it’s used by organizations all around the world that are concerned with protecting their systems as well as sensitive, confidential or regulated data. In fact, there’s so much confidence in the technology that some researchers expect the U.S. market to hit $5.5B this year. In this article, we’re going to review Confidential Computing technology, explore how it’s being used and examine what future innovations might look like.
Confidential Computing protects data in use during processing, with sensitive data isolated in the CPU and encrypted in memory while it’s processed. The mechanism for doing that is a secure enclave in the hardware called a Trusted Execution Environment (TEE). The goal is for sensitive data and trusted code to be loaded into the TEE, which protects it from tampering. This isolated and secure environment helps prevent unauthorized access and modification of in-memory applications and data, thereby increasing assurances that the data remains secure.
The concept of TEEs dates to the early 2000’s with a standard developed by GlobalPlatform. Today, many Confidential Computing standards are driven by the Confidential Computing Consortium (CCC), which is a Linux Foundation project. Intel is a founding member, along with Microsoft, Google, Red Hat and others.
Use cases
Confidential Computing has a variety of use cases. First, with increasing reliance on the cloud computing, Confidential Computing allows organizations to maintain control and better secure their data in the cloud, protecting it from access by malware, other cloud tenants and even the cloud provider. The attestation function provides cryptographic evidence or measurements of the TEE’s authenticity and current state. Any stakeholder relying on the TEE to protect their workloads can receive these measurements and decide whether to trust the code running in the TEE.
Second, it allows multiple parties to collaborate and share data while maintaining privacy. Each organization can be assured that the data they contribute for collaborative analysis is kept confidential from the other parties, and that the environment in which they are sharing hasn’t been compromised. This has broad applications, but a good example is in healthcare where individuals’ health data has become dispersed across a wide and expanding array of data silos. However, providers must collaborate to deliver quality care. Confidential Computing helps protect connected clinical workloads and data in use.
Third, it helps strengthen compliance and data sovereignty programs, which are all about maintaining control of data and making sure it’s used in the jurisdiction it was meant for. Compliance often relies exclusively on processes and procedures, and sovereignty on geo-location. However, data is a liquid asset and can inadvertently “escape” into other data centers (even when the best procedures are in place). Confidential Computing provides an additional technological safeguard to a data sovereignty strategy. The data in use is protected inside a TEE, and since the workload owner holds the keys to decrypt the data, it cannot be collected, viewed or accessed without the owner’s knowledge and consent. Combined with cloud storage and network encryption, Confidential Computing empowers workload owners to control access to their data.
And finally, it gives organizations hardware-based isolation and access controls for sensitive workloads. That could be protecting proprietary business logic, analytics functions, machine learning algorithms or entire applications. Confidential Computing “armors-up” workloads, helping protect sensitive data, content and software IP from advanced attack, tampering and theft. One area of growth and innovation is around Confidential AI, which is the deployment of AI systems inside TEEs to protect sensitive data and valuable AI models while they are actively in-use. It takes modern AI techniques, including Machine Learning and Deep Learning, and overlays them with traditional Confidential Computing technology.
Advancements in Confidential Computing
Confidential Computing has come a long way. In the past two years alone, there have been several advancements. For example, trust services that deliver uniform, independent attestation of trustworthy environments. Application isolation that creates small trust boundaries for data protection. Code integrity and virtual machine (VM) isolation that enhances compliance and control for legacy applications.
However, there are still challenges to be faced. Attestation services are just rolling out to the market, and making these services more understandable, digestible and automated is underway. In addition, the entire computing industry is working toward quantum resistant computing. This transition may require that we encrypt data with a change of key length or that new algorithms need to be invented and standardized. Furthermore, while chip makers are heavily invested in eliminating side-channel and physical attacks against CPUs, this is a constant community effort.
The future of confidential computing is bright. More and more organizations are understanding the technology’s value for protecting data in use through isolation, encryption and control, and verification capabilities. This will help organizations unlock new opportunities for business collaboration and insight.
We list the best Zero Trust Network Access solutions.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro