Vue normale

Il y a de nouveaux articles disponibles, cliquez pour rafraîchir la page.
Aujourd’hui — 1 mai 2024Actualités numériques

The BASIC Programming Language Turns 60

Par : msmash
1 mai 2024 à 18:41
ArsTechnica: Sixty years ago, on May 1, 1964, at 4 am in the morning, a quiet revolution in computing began at Dartmouth College. That's when mathematicians John G. Kemeny and Thomas E. Kurtz successfully ran the first program written in their newly developed BASIC (Beginner's All-Purpose Symbolic Instruction Code) programming language on the college's General Electric GE-225 mainframe. Little did they know that their creation would go on to democratize computing and inspire generations of programmers over the next six decades.

Read more of this story at Slashdot.

Proposal Raised To Deprecate "-Ofast" For The LLVM/Clang Compiler

1 mai 2024 à 19:00
Some that crave the absolute best possible performance sometimes build their software with the "-Ofast" optimization level that is a step above "-O3" but comes with the risk of potentially unsafe math. LLVM developers are now weighing whether to deprecate -Ofast to either remove it or have it just be an alias for the -O3 optimizations...

Google Urges US To Update Immigration Rules To Attract More AI Talent

Par : msmash
1 mai 2024 à 18:01
The US could lose out on valuable AI and tech talent if some of its immigration policies are not modernized, Google says in a letter sent to the Department of Labor. From a report: Google says policies like Schedule A, a list of occupations the government "pre-certified" as not having enough American workers, have to be more flexible and move faster to meet demand in technologies like AI and cybersecurity. The company says the government must update Schedule A to include AI and cybersecurity and do so more regularly. "There's wide recognition that there is a global shortage of talent in AI, but the fact remains that the US is one of the harder places to bring talent from abroad, and we risk losing out on some of the most highly sought-after people in the world," Karan Bhatia, head of government affairs and public policy at Google, tells The Verge. He noted that the occupations in Schedule A have not been updated in 20 years. Companies can apply for permanent residencies, colloquially known as green cards, for employees. The Department of Labor requires companies to get a permanent labor certification (PERM) proving there is a shortage of workers in that role. That process may take time, so the government "pre-certified" some jobs through Schedule A. The US Citizenship and Immigration Services lists Schedule A occupations as physical therapists, professional nurses, or "immigrants of exceptional ability in the sciences or arts." While the wait time for a green card isn't reduced, Google says Schedule A cuts down the processing time by about a year.

Read more of this story at Slashdot.

Windows 10 Reaches 70% Market Share as Windows 11 Keeps Declining

Par : msmash
1 mai 2024 à 17:20
Windows 11's market share dropped in April 2024, falling below 26% after reaching an all-time high of 28.16% in February. According to Statcounter, Windows 11 lost 0.97 points, while Windows 10 gained 0.96 points, crossing the 70% mark for the first time since September 2023. Neowin adds: Some argue that Windows 11 still offers little to no benefits for upgrading, especially in light of Microsoft killing some of the system's unique features, such as Windows Subsystem for Android. Add to that the ever-increasing number of ads, some of which are quite shameless, and you get an operating system that has a hard time winning hearts and minds, and retaining its customers.

Read more of this story at Slashdot.

LastPass Separates From GoTo

Par : msmash
1 mai 2024 à 16:41
LastPass, the password manager company, has officially separated from its parent company, GoTo, following a series of high-profile hacks in recent years. The company will now operate under a shareholder holding company called LMI Parent. LastPass -- owned by private equity firms Francisco Partners and Elliott Management -- has faced criticism for its handling of the breaches, which resulted in the theft of customer data and encryption keys. The company has since enforced a 12-character minimum for master passwords to improve security.

Read more of this story at Slashdot.

Les spécifications de consommation d'Intel, cela donne des comportements étonnants sous Linux !

Suite aux soucis de stabilité des processeurs Raptor Lake sous forte charge, ASUS est allé sortir un nouveau BIOS sur ses cartes mères pour offrir l’expérience d’auteur de ces CPU — comprendre, respecter à la lettre les spécifications (parfois floues !) du géant bleu. Or, lorsque l’option incriminée...

Microsoft Concern Over Google's Lead Drove OpenAI Investment

Par : msmash
1 mai 2024 à 16:02
Microsoft's motivation for investing heavily and partnering with OpenAI came from a sense of falling badly behind Google, according to an internal email released Tuesday as part of the Justice Department's antitrust case against the search giant. Bloomberg: The Windows software maker's chief technology officer, Kevin Scott, was "very, very worried" when he looked at the AI model-training capability gap between Alphabet's efforts and Microsoft's, he wrote in a 2019 message to Chief Executive Officer Satya Nadella and co-founder Bill Gates. The exchange shows how the company's top executives privately acknowledged they lacked the infrastructure and development speed to catch up to the likes of OpenAI and Google's DeepMind. [...] Scott, who also serves as executive vice president of artificial intelligence at Microsoft, observed that Google's search product had improved on competitive metrics because of the Alphabet company's advancements in AI. The Microsoft executive wrote that he made a mistake by dismissing some of the earlier AI efforts of its competitors. "We are multiple years behind the competition in terms of machine learning scale," Scott said in the email. Significant portions of the message, titled 'Thoughts on OpenAI,' remain redacted. Nadella endorsed Scott's email, forwarding it to Chief Financial Officer Amy Hood and saying it explains "why I want us to do this."

Read more of this story at Slashdot.

Global Debt Hasn't Been This Bad Since the Napoleonic Wars, Says WEF President

Par : msmash
1 mai 2024 à 15:20
The massive volumes of debt piling up around the globe forced the president of the World Economic Forum to reach back more than 200 years for a comparable period. Fortune: In an interview Sunday with CNBC at a WEF conference in Saudi Arabia, Borge Brende warned overall debt is approaching the world's total economic output. "We haven't seen this kind of debt since the Napoleonic Wars," he said. "We're getting close to 100% of global GDP in debt." According to the International Monetary Fund last year, global public debt hit $91 trillion, or 92% of GDP, by the end of 2022. That was actually a dip from pandemic-era debt levels but remained in line with a decades-long trend higher. Data on global debt during the Napoleonic Wars, which took place in the early 1800s, is harder to come by. But for comparison, some estimates put British government debt at more than 200% of GDP by 1815. Brende also told CNBC that governments need to take fiscal measures to reduce their debts without triggering a recession. For now, global growth is about 3.2% annually, which isn't bad, but it's also below the 4% trend growth the world had seen for decades, he said earlier in the interview. That risks a repeat of the 1970s, when growth was low for a decade, Brende added. But the world can avoid such an outcome if it continues to trade and doesn't engage in more trade wars. "Trade was the engine of growth for decades," he said.

Read more of this story at Slashdot.

Firebat A8 : un MiniPC Ryzen 7 8845HS qui sort du lot

1 mai 2024 à 15:59

Minimachines.net en partenariat avec TopAchat.com

Le Firebat A8 ne ressemble pas aux précédents modèles de cette marque. Toutes les machines commercialisées par le passé étaient des clones purs et simples des MiniPC classiques. Ce modèle change de formule.

Avec l’affaire des virus qui a touché ces fabricants qui reprenaient tous les mêmes designs, les marques veulent se sortir de cette impasse marketing. Il est devenu important de se démarquer, de proposer un design alternatif. Différent de ce que mettent en avant d’autres constructeurs. Histoire qu’on ne confonde pas « leur » machine avec celle des autres. 

Ce Firebat T8 change donc de format même si je suis quasiment sûr que nous retrouveront des machines totalement identique. Un signe qui ne trompe pas avec ces MiniPC en général c’est la connectique employée. Si la coque change avec ici une partie supérieure découpée de grille en forme de croix. Les ports des différents connecteurs devraient se retrouver chez d’autres constructeurs. Firebat étant une de ces marque sans usine s’appuyant sur un constructeur différent pour produire ses machines il a simplement du commander un boitier différent pour se protéger.

Le châssis mesure 14.8 cm de large pour 12.7 cm de profondeur et 5 bons cm d’épaisseur. A l’intérieur on retrouve un Ryzen 7 8845HS et ses 8 cœurs AMD Zen 4 accompagnés des 12 cœurs RDNA3 de son circuit graphique Radeon 780M. Un petit NPU Ryzen AI est également de la partie et délivre 16 TOPS de performances. L’engin propose deux ports SODIMM double canal de DDR5-5600 et un port M.2 2280 NVM PCIe 4.0… Pas de mention d’un espace de stockage 3.5″ supplémentaire malgré l’épaisseur de l’engin. A la place un système comprenant pas moins de trois caloducs pour gérer la chaleur de l’engin avec un ventilateur processeur et un second au niveau des composants mémoire et stockage.

On retrouve un module Wi-Fi6 et Bluetooth 5.2 classique et en façade deux USB 3.2 Type-A, un USB4, un jack audio combo 3.5 mm et un lecteur de cartes MicroSDXC. A noter que le système pourra démarrer sur ce lecteur et donc sera potentiellement apte a piloter des systèmes alternatifs à son  Windows 11 installé par défaut.

A l’arrière on retrouve 4 ports USB Type-A séparés en paires USB 2.0 et USB 3.2. Deux sorties vidéo en HDMI 2.1 et DisplayPort 1.4, deux port Ethernet 2.5 Gigabit, un second jack audio combo 3.5 mm et un autre port USB4. Un port Antivol type Kensington Lock est également visible ainsi qu’une alimentation Jack. Le boitier est compatible VESA et ouvert largement sur les côtés comme sur l’arrière.

Un ensemble très complet pour le moment vendu à plus de 600€ sur AliExpress sans possibilité de livraison vers la France dans une version 16 Go / 1 To. Un élément qui signifie en général que l’engin est toujours en cours de production et non pas en stock. Le prix devrait baisser au moment de la disponibilité de la machine avec plus de possibilités de se faire livrer et, peut être, plus d’infos sur les entrailles de l’engin.

Firebat A8 : un MiniPC Ryzen 7 8845HS qui sort du lot © MiniMachines.net. 2024.

NVIDIA GeForce RTX - IA Décodé - Répétez-le : ChatRTX ajoute de nouveaux modèles d'IA et de nouvelles fonctionnalités dans sa dernière mise à jour

1 mai 2024 à 13:30

Cette semaine sur IA Décodé, une nouvelle mise à jour de NVIDIA ChatRTX élargissant les capacités de la démo technique ainsi que les modèles LLM pris en charge est désormais disponible au téléchargement. Annoncée et présentée pour la première fois à la GTC au début du mois de mars, cette mise à jour apporte plusieurs nouvelles fonctionnalités : - ChatRTX ajoute à sa liste croissante de LLM pris en charge, dont Gemma, le dernier LLM de Google, et ChatGLM3, un LLM ouvert et bilingue (anglais et chinois), offrant aux utilisateurs une flexibilité supplémentaire. - La nouvelle prise en charge des photos permet aux utilisateurs de ChatRTX de rechercher facilement et d'interagir localement avec leurs propres données photographiques sans avoir à étiqueter des métadonnées complexes, grâce au préapprentissage contrastif langage-image (CLIP) d'OpenAI. - Les utilisateurs de ChatRTX peuvent désormais parler avec leurs propres données, grâce à la prise en charge de Whisper, un système de reconnaissance automatique de la parole par l'IA qui permet désormais à ChatRTX de comprendre la parole verbale. […]

Lire la suite

LinkedIn Now Has Wordle-style Games You Can Play Every Day

Par : msmash
1 mai 2024 à 14:40
LinkedIn, the professional network known for job listings and unsolicited career advice, is jumping into gaming. From a report: The platform is officially introducing a set of Wordle-style puzzle games, weeks after they were first spotted in the app. The company is starting with three games: Pinpoint, a word game where players must guess the theme that ties a series of words together; Queens, a puzzle game that's a bit like a cross between Sudoku and Minesweeper; and Crossclimb, a trivia game that involves guessing a series of four-letter words and placing them in the correct order. LinkedIn describes them as "thinking-oriented games," though the format will likely look familiar to fans of The New York Times Games app. Each game can only be played once a day, and players can share their score with friends in cute emoji-filled messages reminiscent of the "Wordle grid." The service will also keep track of "streaks," to encourage players to come back every day. Given the similarities, it shouldn't be surprising that games were developed by LinkedIn's news team, which recently hired a dedicated games editor.

Read more of this story at Slashdot.

Razer se fait fouetter par la FTC pour tromperie avec son masque RGB anti-COVID "Zephyr"

En octobre 2021, en pleine pandémie, Razer, le fabricant singapourien, connu pour ses périphériques gaming, avait eu une idée de génie pour profiter au mieux d’une mauvaise situation (comme beaucoup) : un masque de protection baptisé Zephyr ! Et pas n’importe quoi comme masque, mais un objet de mode...

PCI-SIG Completes CopprLink Cabling Standard: PCIe 5.0 & 6.0 Get Wired

1 mai 2024 à 14:00

The PCI-SIG sends word over this morning that the special interest group has completed their development efforts on the group’s new PCI-Express cabling standard, CopprLink. Designed to go hand-in-hand with PCIe 5.0 and PCIe 6.0, CopprLink defines both internal and external copper cabling for the latest PCIe standards, giving system vendors and assemblers the ability to use wires to connect devices within a system, or even whole systems.

The CopprLink standard is, in practice, a pair of standards sharing the same brand-name under the PCI-SIG umbrella. The internal standard, “CopprLink Internal Cable”, is designed to allow for a new generation of PCIe cables up to 1 meter in length that are capable of sustaining PCIe 5.0 and PCIe 6.0 signaling. Internal CopprLink effectively supplants a host of older internal PCIe cabling standards (including the abandoned OCuLink), which were originally designed for earlier generations of PCIe signaling.

At a high level, internal CopprLink is intended to provide not only host-to-device connectivity, but even more transparent backhaul applications such as motherboard-to-backplane connectivity, and unique applications such as chip-to-chip PCIe connections. In other words, CopprLink allows for cabled PCIe to be used in almost any situation where a PCIe connection needs to be established within a system. Strictly speaking, CopprLink doesn't replace the PCIe CEM connector in any way – but the relatively thick copper cables have less signal loss than PCB traces, making a cabled standard extremely useful even for internal connections. PCI-SIG sees CopprLink cables taking hold in the storage and data center markets, product categories where we already see PCIe cabling in use today.

The companion connector standard for internal CopprLink is the SNIA-developed SFF-TA-1016 connector, which bears more than a passing resemblance to the widely-used SFF-8654 (SlimSAS) connector. SFF-TA-1016 is available in x4, x8, and x16 configurations, and while the PCI-SIG doesn’t go so far as to defining widths within their own standard, the connectors available paint a clear picture of the options at hand. Internal CopprLink x4 should be especially popular with storage, as we already see today.


Top: SFF-TA-1016 Family of Connectors (Figure 4-1, Image Courtesy SNIA)
Bottom: Sample SFF-TA-1016 x4 Contact Plug and Recepticle (Figure 4-2, Image Courtesy SNIA)

Meanwhile, the group has also developed an external cabling standard to cover those same PCIe 5.0/6.0 data rates. External CopprLink cables can go up to 2 meters, allowing for board-to-board connections within a rack, and even short rack-to-rack PCIe connections.

The external version of CopprLink also uses a more robust connector, relying on SNIA’s SFF-TA-1032 standard. Like internal/1016, this is available with x4, x8, and x16 configurations, using 44, 68, and 120 positions/pins respectively. The PCI-SIG is expecting this version of the standard to be primarily adopted by the AI/Machine Learning markets, which need to move heaps of data between systems. Notably, however, they don’t really expect the storage market to make use of this spec – instead, they’ll be served by an updated version of the classic PCI Express External Cabling standard.


SFF-TA-1032 x16 Plug and Connector (Figure 4-1, Image Courtesy SNIA)

Finally, a bit farther out on the group’s roadmap, PIG-SIG is also reiterating that they’re working on a new optical cabling standard as well. The workgroup for this project was established in 2023, so the project is still in its early days. Notably, the forthcoming optical standard is intended to be optical technology-agnostic, allowing for PCIe to be paired with a variety of optical technologies.

In the meantime, with the internal and external CopprLink standards completed, the PCI-SIG is hoping to quickly move this cabling into production. Since these are solely cabling standards – and thus don’t require intensive development efforts such as new controllers or the like – the group is hoping that their members will have something to show off in time for the group’s developer conference this summer, or the Flash Memory Summit in August. After which, hardware vendors should be able to deploy the new cables relatively quickly.

Star Scientist's Claim of 'Reverse Aging' Draws Hail of Criticism

Par : msmash
1 mai 2024 à 14:00
An anonymous reader shares a report: Harvard geneticist David Sinclair, who has said his "biological age" is roughly a decade younger than his actual one, has put forward his largely unlined face as a spokesman for the longevity movement. The 54-year-old has built his brand on the idea that aging is a treatable disease. The notion has proven so seductive that legions of acolytes follow his online postings about his research and the cocktails of supplements he consumes to stave off the inevitable. His social-media accounts are a platform for assertions that his work is pushing nearer to a fountain of youth. He claimed last year that a gene therapy invented in his Harvard lab and being developed by a company he co-founded, Life Biosciences, had reversed aging and restored vision in monkeys. "Next up: age reversal in humans," he wrote on X and Instagram. On Feb. 29, in the eyes of many other scientists working to unlock the mysteries of aging, he went too far. Another company he co-founded, Animal Biosciences, quoted him in a press release saying that a supplement it had developed had reversed aging in dogs. Scientists who study aging can't even agree on what it means to "reverse" aging, much less how to measure it. The response was swift and harsh. The Academy for Health and Lifespan Research, a group of about 60 scientists that Sinclair co-founded and led, was hit with a cascade of resignations by members outraged by his claims. One scientist who quit referred to Sinclair on X as a "snake oil salesman." Days later, in a tense video meeting, the academy's five other board members pressed Sinclair to resign as president. He contended that the press release contained an inaccurate quote, according to people who were in the meeting, but he later stepped down. Sinclair's work is published regularly in top-tier scientific journals and has brought attention to an emerging field vying for credibility and funding. He has parlayed his research into hundreds of millions of dollars of investment in various companies, more than 50 patents and prominence as a longevity influencer. Along the way, his claims -- especially in his social-media posts, interviews and his book -- have drawn criticism from scientists who have accused him of hyping his research and extolling unproven products, including some from companies in which he had a financial interest. "My lab's ideas and findings are typically ahead of the curve, which is why some peers might feel the research is overstated at the time," Sinclair said to The Wall Street Journal in an email. "I stand behind my track record as a trusted scientist in one of the most competitive professions of all." He said he doesn't engage with social-media critics, including those calling him a snake oil salesman, and that many such comments are "nothing more than mischaracterizations."

Read more of this story at Slashdot.

Mysterious 'gpt2-chatbot' AI Model Appears Suddenly, Confuses Experts

Par : BeauHD
1 mai 2024 à 13:00
An anonymous reader quotes a report from Ars Technica: On Sunday, word began to spread on social media about a new mystery chatbot named "gpt2-chatbot" that appeared in the LMSYS Chatbot Arena. Some people speculate that it may be a secret test version of OpenAI's upcoming GPT-4.5 or GPT-5 large language model (LLM). The paid version of ChatGPT is currently powered by GPT-4 Turbo. Currently, the new model is only available for use through the Chatbot Arena website, although in a limited way. In the site's "side-by-side" arena mode where users can purposely select the model, gpt2-chatbot has a rate limit of eight queries per day -- dramatically limiting people's ability to test it in detail. [...] On Monday evening, OpenAI CEO Sam Altman seemingly dropped a hint by tweeting, "i do have a soft spot for gpt2." [...] OpenAI's fingerprints seem to be all over the new bot. "I think it may well be an OpenAI stealth preview of something," AI researcher Simon Willison told Ars Technica. But what "gpt2" is exactly, he doesn't know. After surveying online speculation, it seems that no one apart from its creator knows precisely what the model is, either. Willison has uncovered the system prompt for the AI model, which claims it is based on GPT-4 and made by OpenAI. But as Willison noted in a tweet, that's no guarantee of provenance because "the goal of a system prompt is to influence the model to behave in certain ways, not to give it truthful information about itself."

Read more of this story at Slashdot.

❌
❌