FreshRSS

Zobrazení pro čtení

Jsou dostupné nové články, klikněte pro obnovení stránky.

Sabrent announces new Rocket 5 M.2 NVMe Gen 5 SSD

Sabrent announces new Rocket 5 M.2 NVMe Gen 5 SSD


Phison Max14um controller peaking at 14GB/s

Sabrent has announced its latest flagship M.2 NVMe SSD, the Rocket 5. Pairing up the latest Phison with Micron 3D TLC NAND, DRAM cache, the Rocket 5 will peak at 14GB/s and come in capacities of up to 4TB.

As said, for its newest flagship M.2 NVMe SSD, Sabrent is going all in, pairing up the latest Phison PS5026-E26 Max14um controller with Micron B58R 232-layer 3D TLC NAND flash memory and equipping it with LPDDR4-based DRAM cache. While Sabrent did not reveal any precise details, it did say that the Rocket 5 will peak at 14GB/s and 12GB/s sequential read and write transfer speeds, with 1.55M IOPS 4K random read and 1.8M IOPS 4K random write.

The Rocket 5 will replace the Rocket 4 Plus as the flagship SSD, and currently, the Rocket 5 is available on Amazon for $729.99 for the 4TB version, $339.99 for the 2TB one, and $190.17 for the 1TB version.

- SABRENT Rocket 5 4TB Advanced Performance Internal M.2 PCIe GEN 5 14GB/s X4 NVMe SSD - $729.99 at Amazon.com

- SABRENT Rocket 5 2TB Advanced Performance Internal M.2 PCIe GEN 5 14GB/s X4 NVMe SSD - $339.99 at Amazon.com

- SABRENT Rcoket 5 NVMe SSD M.2 1TB Gen 5 - $190.17 at Amazon.com

rocket5 1

 

EU snuffles but might ignore Vole’s EU deal

EU snuffles but might ignore Vole’s EU deal


It's not an acquisition

EU watchdogs might be prepared to let the software King of the World, Microsoft, get away with its €12.15 billion investment in OpenAI.

European Commission bodies keep their cards close to their chest regarding antitrust actions. Still, word on the street, courtesy of Reuters, is that Microsoft's hefty €12.15 billion investment in OpenAI might not set off any EU antitrust alarms. It seems it's not likely to be pegged as an "acquisition" in the legal sense over there.

Thanks to its cash injection into the generative AI LLM provider, Vole could avoid the more stringent investigation procedures and potential regulatory hurdles.

A spokesperson from the European Commission mentioned that for a deal to be "notifiable" to the EC as a merger, it needs to signal a lasting change in the reins of the company in question.

The spokesperson didn't dismiss the possibility of a more thorough and strict regulatory response and mentioned that the probe into the Microsoft-OpenAI deal is still on the go.

While this transaction hasn't been officially flagged up, the Commission has been keeping a beady eye on the control situation over OpenAI even before the recent kerfuffle involving its management, including Microsoft's spot on the OpenAI board and the investment agreements between Microsoft and OpenAI, the spokesperson said.

However, the EC hasn't decided whether the ties between the two firms amount to a "change of control" following Microsoft's investments.

Antitrust watchdogs in the UK and the US are still snuffling around, giving the nod to the deal. The UK's Competition and Markets Authority and the US Department of Justice and FTC are rumoured to be pondering their next moves regarding formal reviews and investigations.

Under EU rules, a "concentration," which would get the antitrust magnifying glass, happens when one company's control changes hands. According to the Consolidated Jurisdictional Notice, this can occur by snagging "sole control" of a company, meaning the controlling party can call the shots over the other.

However, sole control can also be based purely on legal or actual grounds, reflecting the complex web of board, shareholder, and voting rights setups available to companies operating in the EU. For instance, having the lion's share of voting rights could mean effective sole control. At the same time, a minority shareholder likely to win majorities at shareholder meetings could be seen as having absolute power.

UK and EU regulators had warned Microsoft in January that its investments in OpenAI might be scrutinised despite the firm's insistence that its board position is non-voting and, thus, it doesn't really "own" OpenAI.

Microsoft's Volish lips are sealed on the matter.

 

Chipzilla builds its first High NA EUV lithography tool

Chipzilla builds its first High NA EUV lithography tool


Should have it running by the end of the year

Intel's Oregon team has accomplished a significant feat, assembling its first High NA EUV lithography tool. This cutting-edge technology is set to be operational within the year.

The machine, costing a hefty €348 million ($373 million), is as big as a double-decker bus and tips the scales at over 150 metric tonnes. It made its way to Oregon packed in over 250 crates, stowed away in 43 freight containers. These containers flew on several cargo planes to Seattle before being trucked down to Oregon on 20 lorries. Each new system weighs in at more than 150 metric tonnes.

Intel Fellow and director of Lithography, Hardware, and Solutions for Intel Foundry Logic Technology Development Mark Phillips said that with the addition of High NA EUV, Intel will have the most well-rounded lithography toolbox in the industry, enabling the company to drive future process capabilities beyond Intel 18A into the second half of this decade.

The plan is to put this kit to work on developing Intel's 14A generation of chips come 2025, with the wheels set to start turning on early production in 2026 and full-scale commercial production expected by 2027.

High NA EUV is touted to print features up to 1.7 times tinier than what current EUV tools can manage. This tech is set to enable 2D feature scaling, which could lead to a density increase of up to 2.9 times.

When compared to 0.33NA EUV, High NA EUV (or 0.55NA EUV) promises higher imaging contrast for similar features. This means it needs less light per exposure, cutting down the time needed to print each layer and potentially boosting wafer output.

Chipzilla’s game plan includes using 0.33NA EUV and 0.55NA EUV, along with other lithography processes, in developing and manufacturing ICs. This will start with product proof points on Intel 18A in 2025 and continue into the production phase of Intel 14A.

Intel's strategy aims to fine-tune process technology for both cost-effectiveness and performance.

Despite Intel co-founder Gordon Moore being an early backer of the EUV concept and Intel pouring several billion dollars into its development at ASML, the company dragged its feet on adopting EUV tools in manufacturing, opting instead for multi-patterning. This hesitation led to Chipzilla falling behind the frontrunners in process technology.

Apple’s RAM scam

Apple’s RAM scam


8GB just doesn't cut It

The fruity cargo cult Apple is at it again. They're pushing 8GB of RAM on their entry-level Macs, telling us it's "suitable for many tasks," like surfing the net, watching videos, and even "light" video and image editing.

To make sure that its Apple fanboys stick to this 8GB limit all MacBook Airs, the Mac Mini, and the MacBook Pro 14 are stuck with this measly 8GB RAM that's soldered onto the motherboard, so no chance of a boost later on.

Apple users, when they come to the realization that 8GB is insufficient, are met with an exorbitant $ 200 charge from Jobs' Mob to upgrade their gadgets to 16 GB. This is a stark contrast to the experience of PC owners, who know that an 8GB upgrade is a mere pittance.

At PC Gamer, Jeremy Laird is not swayed by Apple's tactics. He points out, "That might have been acceptable if not for the fact that Apple charges a ludicrous $200 to upgrade any of those machines from 8GB to 16GB."

Laird's stance is clear-even if an 8GB Mac can handle the basics, it's still not up to par.

Why? Because a Mac with just 8GB can struggle even with basic tasks like web browsing. For instance, Chrome users are well aware of its memory consumption. Laird, with about 15 tabs open, is already seeing three tabs using over 500MB each. While a normal PC would have no problem with this, a person I am working with has 153 tabs open on her PC, Macs tend to clutch their chest and stagger about like an asthmatic ant with a heavy load of shopping.

What is ironic is that Laird's currently on a 16GB MacBook Air, using a whopping 12.5GB of memory with nothing but Chrome open. He reminisces about the bad old days with an 8GB Apple silicon Air, calling it a "nightmare" that was always gasping for memory just doing basic web browsing.

The big miss, according to Laird, is that while everyone's nattering that 8GB isn't enough for the heavy lifting, they're glossing over the fact that it's not even up to snuff for a spot of web surfing.

Jobs' Mob's 8GB RAM might as well be a paperweight for modern computing needs.

AI successfully fights against a human pilot

AI successfully fights against a human pilot


So where the ‘ell was Biggles when we needed him last Saturday

The Yanks are taking to the skies with AI at the helm. The boffins at DARPA, the Defense Advanced Research Projects Agency, have let slip that they've had an AI-piloted jet go toe-to-toe with a human pilot up in the wild blue yonder.

In December 2022, DARPA started fiddling with AI for their Air Combat Evolution (ACE) programme. They've been busy bees, working on an AI system that can fly a fighter jet on its own, all the while making sure it sticks to the Air Force's safety rules.

After a bit of a play with dogfighting simulations using the AI pilot, DARPA decided to put their tech to the actual test. They popped the AI system into their experimental X-62A aircraft. This bit of kit took to the skies at Edwards Air Force Base in California, where, in September 2023, it had its first successful scrap with a human in the cockpit.

There were flesh-and-blood pilots aboard the X-62A, ready to hit the brakes on the AI if needed, but DARPA reckons there wasn't a moment they had to use the "off" switch.

The X-62A had a bit of a showdown with an F-16, with a human flying solo. Both birds showed off some fancy flying, getting within 2,000 feet of each other while zipping along at 1,200 miles per hour. DARPA's keeping mum on who the top gun was.

"Dogfighting was the problem to solve so we could start testing autonomous artificial intelligence systems in the air," said Bill Gray, the top gun at the Air Force's Test Pilot School. "Every lesson we’re learning applies to every task you could give to an autonomous system."

So far, they've completed 21 test flights and plan to continue until 2024. With AI getting smarter by the minute, there's a bit of a kerfuffle about how the military might use these systems.

Meta shows off new tarted-up Llama Model

Meta shows off new tarted-up Llama Model


Needs to see off competition from OpenAI

Meta has just unveiled the latest iteration of its mighty Llama AI model to stay in the race with rival tech from the likes of OpenAI, and Google.

The firm is touting its new Llama 3 8B and Llama 3 70B models, which pack 8 billion and 70 billion parameters, respectively, as a significant improvement over their forerunners.

Meta confidently asserts that its Llama 3 models, refined on a custom 24,000 GPU cluster, are among the top-tier generative AI models in their size category. To substantiate this claim, Meta highlights the models' exceptional performance on renowned AI benchmarks like MMLU, ARC, and DROP. These tests, despite some scepticism, remain a reliable method to evaluate AI models.

The Llama 3 8B model is proving its worth, surpassing other open-source models such as Mistral's Mistral 7B and Google's Gemma 7B in at least nine benchmarks. Its proficiency in diverse areas, from biology and physics to chemistry, maths, and even commonsense reasoning, is demonstrated.

So Mistral 7B and Gemma 7B might not be cutting-edge (Mistral 7B hit the scene last September), and in some of the benchmarks, Meta's banging on about Llama 3 8B only nudges ahead by a smidge. But Meta's also claiming that its beefier Llama 3 model, the Llama 3 70B, can hold its own against the big guns of generative AI, including Google's latest gem, the Gemini 1.5 Pro.

Apple told to yank chat apps

Apple told to yank chat apps


Making its Walled Garden even less attractive

The Chinese government has ordered Jobs' Mob to remove several top chat apps from its App Store in China, marking the latest in a series of censorship demands on the tech giant in its second-largest market.

Meta Platforms' WhatsApp and Threads, along with Signal, Telegram, and Line, were removed from the Chinese App Store last Friday. Jobs' Mob explained that it was instructed to ditch certain apps due to national security concerns, though it didn't disclose which ones.

A spokesperson for the fruity cargo cult said that it would not be standing up to the Chinese with the same vigour that it might with the EU or the US.

"We are obligated to follow the laws in the countries where we operate, even when we disagree," said a spokesperson for Jobs' Mob said.

These chat apps, which let users chatter and share files one-on-one and in large groups, boast a global user base north of three billion. They're only reachable in China through virtual private networks that skirt around China's Great Firewall, yet they're still widely used.

Beijing's always been a bit twitchy about such platforms, fretting that its citizens might use them to spread iffy content and rally for protests or social movements. A fair bit of the news China keeps under wraps at home tends to slip past the Great Firewall via these apps.

The Cyberspace Administration of China reportedly pressed Jobs' Mob to scrub WhatsApp and Threads from the App Store because they've got political chit-chat that's a bit dodgy, including some less-than-flattering mentions of the Chinese president. The spokesperson from Jobs' Mob insisted that wasn't why they were removed.

The cull reduces the number of foreign chat apps available to Chinese netizens for chinwags with folks outside the country, tightening Beijing's grip on the internet, which is already pretty leery about unfiltered info doing the rounds.

Over the last ten years, Sensor Tower reckons, Instagram, X, Facebook, YouTube, and WhatsApp have racked up more than 170 million downloads from Jobs' Mob's App Store in China. Apps like X played a blinder in spreading the word and footage of the anti-Covid rule protests in China towards the end of 2022.

Jobs' Mob might have to axe even more foreign and local apps in China, following a government decree last year that told mobile app developers to register their apps with the powers-that-be by this March. Beijing said it's gunning for apps that are up to no good, like those involved in phone scams.

These app axings are just the latest in a series of moves Jobs' Mob has made in its biggest market outside the US to stay on the right side of China's ever-tightening censorship and data security laws. Last year, it was the top dog in China's smartphone market, bagging more than 17 per cent of it, says International Data Corporation.

More than 14,000 apps are on the no-go list in China, according to AppleCensorship.com, a site run by the anti-censorship crew GreatFire, who've been keeping tabs on this for over ten years. Among the banned are tools for getting around internet blocks, chat apps, and apps linked to the LGBT community, GreatFire says.

China's not just a massive sales market for Jobs' Mob; it's also where they make most of their gear. But it's not all smooth sailing. Weak spending's been a drag on sales, and local rivals are giving them a run for their money. Beijing has even put the brakes on state employees who use iPhones.

Just this Thursday, Huawei kicked off sales of its swanky new Pura 70 smartphone range, which industry watchers reckon will nick more of Jobs' Mob's market share in China. Counterpoint Research is betting on Huawei's smartphone sales in China jumping 39 per cent this year.

The fresh set of rules over apps applies to both foreign and local app shops and doesn't single out Jobs' Mob. However, the services bit, which includes App Store deals, is a big deal for Jobs' Mob's bottom line, and any kerfuffle with the Chinese App Store could take a bite out of their profits.

Meta's been trying to worm its way back into China's hefty consumer market after Facebook got the boot in 2009, when China clamped down on its internet following some anti-government ruckus. But it's not getting anywhere. Mark Zuckerberg, the big cheese at Meta, has been having a pop at China and the short-video app TikTok lately, which hasn't done him any favours in Beijing.

Meta's still raking it in from China by flogging ads on its platforms to Chinese firms. It's also been sniffing around a deal with Chinese social media and video game giant Tencent to hawk its virtual-reality headsets in the country.

Google sacks workers involved in anti-Israel protest

Google sacks workers involved in anti-Israel protest


So they have been arrested and are now jobless

Search outfit Google has listened to the 28 workers who protested against a deal the technology giant has with the Israeli government and has decided to open a series of rational talks about the matter – no, not really it has fired the lot of them as a warning to other staff who get in the way of it making millions on the back of military contracts.

Google and Amazon have teamed up on a $1.2 billion deal called Project Nimbus, which equips Israel's government and military with advanced cloud computing and AI tools. Amidst this, employees linked to the group No Tech For Apartheid have been protesting at Google's offices in New York and Sunnyvale, California.

The tech giant maintains that Project Nimbus "is not directed at highly sensitive, classified, or military workloads relevant to weapons or intelligence services." However, the company has expressed concerns about protesters "physically impeding other employees' work and preventing them from accessing facilities."

The BBC has received confirmation from Google that recent firings have intensified the already strained relations with staff who are against providing technology to Israel's government. Protests, which began last year after the outbreak of conflict in Gaza in October, saw demonstrators carrying signs stating, “No More Genocide For Profit.”

The latest wave of protests escalated to the point where employees occupied offices and wouldn't leave, leading Google to involve law enforcement. The company has denounced the actions as "completely unacceptable."

Qualcomm Snapdragon X Elite laptops pictured

Qualcomm Snapdragon X Elite laptops pictured


Could be officially unveiled on April 20

Some Lenovo laptops with the Qualcomm Snapdragon Elite X SoC have been spotted online, including the Yoga Slim 7 14 and Thinkpad T14s. According to Qualcomm, these could be unveiled at the Lenovo Tech World '24 keynote on April 20th.

Leaked pictures coming from Twitter Walkingcat user shows both the Yoga Slim 7 14 and the Thinkpad 14s "Snapdragon Edition", both pretty high-end devices from Lenovo. Unfortunately, the provided details only include a couple of pictures and names of these two laptops, as there are no specifications, so there are no details about the actual version of the Snapdragon Elite X SoC.

Yoga Slim 7 14(.5) 2024 Snapdragon Edition pic.twitter.com/k29LupeWk6

— WalkingCat (@_h0x0d_) April 17, 2024

Recent Geekbench benchmark listings leaked some information, showing the Snapdragon X Elite-X1E78100 12-core SoC running at 3.42GHz. Of course, we'll have to wait for the official launch unless we get further leaks as we draw closer to the April 20th event. As said, Qualcomm already teased the announcement on its official Weibo channel for that same Lenovo Tech World '24 event.

 lenovo sdxelitelaptop 1

lenovo sdxelitelaptop 2

 

Samsung and Qualcomm team up on 3GPP Release 17

Samsung and Qualcomm team up on 3GPP Release 17


Claim to increase downlink speeds by 20 per cent

Samsung Networks and Qualcomm have used advanced modulation, as outlined in 3GPP Release 17, to increase downlink speeds by approximately 20 per cent compared to 256 QAM.

The laboratory test, conducted at Samsung’s research and development laboratory in Korea, signifies an industry first for the deployment of 1024 QAM in an FDD band (2.1 GHz) and 1024 QAM across both FDD and TDD (3.5 GHz) spectrums, as per the partners.

Samsung said 256 QAM is widely used commercially within mobile networks to transmit data more efficiently. With the potential of 1024 QAM, operators can extract even greater performance and capacity from their spectrum assets, paving the way for a more efficient and faster mobile network. The test utilised 20 megahertz of spectrum and achieved downlink speeds of 485 Mbps, which the companies have noted is close to the theoretical gain of 2,024 QAM.

The tests depended on Samsung’s 5G virtualised RAN software and radios and a Qualcomm smartphone test device incorporating the Snapdragon X75 5G modem-RF system.

Samsung has also stated that it is examining 1024 QAM in traditional RAN scenarios. The technology is expected to be commercially available this year.

Samsung’s Vice President and head of Networks product strategy, Ji-Yun Seol, said: “We believe that enhancing spectral efficiency is crucial in developing next-generation solutions. This accomplishment highlights Samsung’s continuous commitment to advancing the boundaries of mobile technology.

“Our joint efforts with Qualcomm Technologies have once again produced remarkable outcomes, and we are eager to use our collective expertise to propel network innovation.”

❌