windowsxp

windowsxp club

  • Google I/o2025 Points: Chrome Big Update! Deep integration of Gemini Live

    At the Google I/O 2025 conference on Tuesday, Google launched a new feature called “Gemini-in-Chrome”, officially introducing Gemini Live to PC and Mac, naturally with Chrome browser. According to the news announced by Google on that day, the function will be pushed to some beta users on May 21, and the official push time has not been announced.

    According to the previous release rhythm, users should soon be able to see this function in the official version of Chrome.

    To be fair, in the current tense situation of the AI war, Google does not have the advantage, and even lags behind. Although Google tried to reverse the decline by quickly iterating Gemini, it had little effect. In the international market, although ChatGPT maintains its advantage, domestic AI such as DeepSeek also shows no weakness. On the contrary, Google Gemini has almost no bright spots.

    Fortunately, Google still has Chrome. In the browser market, Chrome is still far ahead. Of course, Google does not want users to open Chrome just to use ChatGPT or DeepSeek, which are business AI models, so the integration of Gemini is naturally imperative.

    Picture

    After the integration of Gemini into Chrome, users can quickly open a small window in the upper right corner of Chrome browser to use Gemini or Gemini Live, or select the shortcut key (Alt+G&Ctrl+G) to turn on the function.

    It is worth mentioning that Gemini Live’s natural voice dialogue is more convenient than typing, and it can also be more efficient in presentations. On the other hand, the reputation of Gemini Live is much better than that of the traditional Gemini. At the same time, because it is officially produced by Google and has rich API integration, it is reasonable to believe that users can completely rely on voice to control the browser in the future.

    Picture

    In the actual demonstration, users only need to “tell” Chrome what content needs to be opened, such as comparing two products on the shopping platform, and opening two tabs to compare content, etc. The speed and efficiency of actually opening web pages, the efficiency of natural language seems to be far superior to the traditional mode of UI click and input. At least it can be confirmed that Gemini’s permissions in Chrome browser are very high, and routine operations such as controlling tabs are completely out of the question.

    Picture

    As Chinese users, we must be more concerned about whether browsers using the Chrome kernel can also integrate domestic AI models such as DeepSeek. Of course, the permission opening of API is the top priority. If Google is willing, a large number of plug-ins will appear in the future. Combined with the localized AI model, the experience of browsing the web will be subverted.

    Picture

    The vice president of Google also said, “We hope to turn some tasks that can be completed in 30 minutes into a one-click journey.” There have been a lot of big cakes drawn by Google in the past two years. We will wait and see how Gemini-in-Chrome performs.

  • Microsoft announced the integration of multiple AI models, and Musk unexpectedly appeared.

    On May 19, local time, Microsoft announced at the annual Software Developer Build Conference held in Seattle that it would provide new artificial intelligence models developed by companies including xAI, Meta and Mistral in its own data center. On the same day, Microsoft also released a new artificial intelligence tool that can independently complete software programming tasks.

    Boosted by the demand for artificial intelligence, Microsoft has now become the company with the highest market value in the world, with the latest market value exceeding 3.4 trillion US dollars.

    The initiative to integrate Musk’s xAI and other company AI models into Microsoft cloud services highlights the transformation of Microsoft’s investment strategy in the field of artificial intelligence. As a shareholder of OpenAI, the developer of ChatGPT, Microsoft is spreading its investment to a wider range of participants in the field of artificial intelligence.

    Microsoft CEO Satya Nadella said in her keynote speech that Microsoft’s move aims to enable developers to mix and use all these different models.

    Nadella had a video conversation with Musk in the opening speech. Musk’s unexpected appearance attracted attention – he sued OpenAI together with Microsoft last year to oppose the transformation of OpenAI into a for-profit company.

    Sam Altman, CEO of OpenAI, also talked to Nadella via live video at the conference that day.

    On that day, Microsoft GitHub released a new Copilot programming agent to help developers complete specific programming tasks, such as fixing errors or rewriting code. Last week, OpenAI also released a similar intelligent programming agent preview version called Codex.

    Thomas Dohmke, CEO of GitHub, wrote in a blog: “The agent uses the most advanced models and is good at performing tasks of low to medium complexity in a fully tested code base, such as adding functions, fixing errors, extending tests, refactoring code and improving documents. ” A GitHub spokesman said that Anthropic’s Claude 3.7 Sonnet AI model supports the programming agent.

    Since GitHub was acquired by Microsoft in 2018, GitHub’s annual revenue has exceeded $2 billion by last summer.

    Kevin Scott, Microsoft Chief Technology Officer, said before the Microsoft Developer Conference that the future envisaged by Microsoft is that any company’s artificial intelligence agents can work together with agents of other companies and better remember each other’s interaction.

    According to the vision described by Microsoft, future enterprises will be able to build their own agents based on the artificial intelligence models of their choice, and these intelligent agents may mix different artificial intelligence models. Microsoft also said that intelligent programming agents are not limited to applications that write code, but appear as a programmer in the team.

    Bob O’Donnell, president and chief analyst of TECHnalysis Research, a research institution, said: “The concept of treating intelligent agents, including programming agents, as digital employees, is a breakthrough change.”

  • The latest Netflix movie in May

    , Disney+, Spotify and other streaming accounts, Netflix must-watch movies in May include “Their Spring, Summer, Autumn and Winter”, “Little Vinicius: The Legend of a New Star in Football”, “Love, Death & Robot”, “The Abyss of Gambling”, “The Devil’s Scheme 2”, “Crime Bullet 3”, “Fear Street” : The Queen of the Ball, Swallowing Gold, Motherly Love Canteen, Eternal Love, Your Taste, etc.

    April 2025 Highlights List Review: Netflix Netflix’s Latest Good Drama and Movie List Recommendation [April 2025] Full List, Netflix movies not to be missed include: “Heartbeat”, “Devil May Cry”, “Evil Fate”, “MLB Lounge: A Year of the Boston Red Sox, Strange Case 84’s Funny Homestay, Black Mirror, Battle of the Moon, Lanson Canyon, Shinkansen Countdown, Poisonous Robbery and other movies and dramas.

  • Apple officially closed the iOS 18.4.1 downgrade channel, and iPhone users need to be cautious when upgrading.

    Apple recently announced that it has officially closed the signature verification service of iOS 18.4.1 version one week after the release of the iOS 18.5 system. This means that all iPhone users who have been upgraded to the latest version of iOS 18.5 will no longer be able to retreat to the previous iOS 18.4.1 version through official channels.

    It is understood that iOS 18.4.1 version focuses on fixing key errors and improving system security, especially solving the problem that wireless CarPlay cannot be connected in a few vehicles. This improvement is crucial to user experience.

    Apple’s security update mechanism requires that the installation of any iOS version must first be verified by the server. Once the verification channel is closed, users can no longer install the old version of the system without official authentication. This initiative is part of Apple’s long-term security strategy, which aims to ensure that user devices are running system versions containing the latest security patches.

    Apple officially closed the iOS 18.4.1 downgrade channel, and iPhone users need to be cautious when upgrading.

    Apple officially closed the iOS 18.4.1 downgrade channel, and iPhone users need to be cautious when upgrading.

    Apple’s consistent practice is to stop the signature verification service for the old version about a week after the release of the new system version. This action once again reflects Apple’s great importance to user data security and system stability.

    At the same time, Apple also announced the termination of the signature verification of iPadOS version 17.7.6. Previously, due to the stability problems of iPadOS version 17.7.7, Apple has urgently pushed iPadOS version 17.7.8 as a countermeasure to ensure that iPad users’ devices can run smoothly.

  • The new features of iOS 19 are exposed, and Apple may say goodbye to “painting cake” publicity

    It has been less than a month since Apple’s Global Developers Conference WWDC 25. In the early morning of June 10, Beijing time, Apple will officially unveil the mystery of iOS 19, iPadOS 19, macOS 16 and other operating systems. At that time, the new functions and changes of the new system will be comprehensive for the first time. Presented in front of the public.

    Unlike the continuous leaks of hardware products such as iPhone, system software such as iOS 19 has been relatively scarce because it is completely developed inside Apple.

    However, the well-known journalist Mark Gurman brought new information in his latest Power On column, revealing in advance two high-profile new features in iOS 19 and iPadOS 19, which are closely related to the needs of users in the Middle East and South Asia. .

    On the one hand, it is a two-way language keyboard. As an advanced two-way keyboard technology, it can achieve fluent switching between Arabic and English, greatly optimizing the input experience of multilingual users.

    On the other hand, there is the virtual reed pen function, which is a calligraphy tool specially designed for Apple Pencil users. By simulating the stroke characteristics of the traditional Arabic calligraphy pen – reed pen, it helps users create more cultural calligraphy works on the iPad.

    As Apple continues to expand its retail business in the United Arab Emirates, Saudi Arabia and India, these functions are regarded as important factors in opening up local markets and promoting future growth.

    Looking back on Mark Gurman’s previous report, he has also shared a number of new features that are expected to appear on WWDC 25: AI to optimize battery life with the help of AI technology, while adding the remaining charging time display on the lock screen interface.

    In visionOS 3, the eye movement scrolling function allows users to scroll through the application interface only through eye movements.

    Introduce a unified interface between devices with a more consistent operation interface for iPhone, iPad and Mac.

    Committed to making iPad more like Mac and further narrowing the gap between the two, iPad productivity upgrade;

    AirPods real-time translation function that supports real-time dialogue translation after deep integration of AirPods and iOS 19;

    Realize the convenient synchronization of Wi-Fi login information on multiple devices, improve the public Wi-Fi login synchronization of the user experience, etc.

    Recently, Mark Gurman also disclosed Apple’s latest developments in the field of AI and the development difficulties it faces.

    The Siri upgrade, which is highly anticipated by users, is likely to be unable to meet everyone this year.

    Although Apple may display the new features of Siri in iOS 19 at WWDC 25, Gurman pointed out that these updates are unlikely to be the focus of the launch.

    It is reported that Apple regarded AI as a transformative technology as early as 2014 and expanded its technology reserve by acquiring many start-ups.

    However, Apple’s slow progress in the field of AI is largely due to differences in internal management.

    Craig Federighi, the head of Apple’s software department, is cautious about AI investment, believing that its return is uncertain and prefers to give priority to the annual updates of core products such as iPhone and Mac.

    Since joining Apple in 2018, John Giannandrea, the head of Apple’s AI, has been advocating large-scale capital investment in AI development. Unfortunately, his proposal has been restricted many times due to conflicts with Federighi’s concept.

    In order to reverse the impact of Siri’s poor performance in the field of AI on the brand image, Apple plans to separate Apple Intelligence from the Siri brand in an attempt to weaken the drag of Siri’s short board on consumer confidence.

    In addition, Apple seems to have realized the negative impact of overcommitment of immature functions in the past.

    It is reported that Apple may be more cautious in the future and will no longer announce unperfected technologies several months in advance to avoid disappointing users due to the serious delay in the launch of functions.

    Based on this, the upcoming new iOS 19 function is expected to pay more attention to practical implementation. Perhaps there will no longer be the past situation of relying only on PPT display to “draw cakes”, but will bring users a new experience that is more practical and stable. Let’s wait and see.

  • Intel B60 hard NVIDIA A1000, the battle for the hegemony of AI chips is on the rise again

    When the Intel Ruixuan Pro B60 with 24GB of video memory meets the NVIDIA RTX A1000 with 8GB of video memory, how many unknown industry battles are hidden behind this seemingly disparity duel? Can Intel CEO Chen Liwu’s declaration that “if you want to return to the peak, you need to tell the truth” shake NVIDIA’s dominant position in the AI chip market?

    Performance duel: The technology breakthrough behind the 3x graphics memory gap

    Under the spotlight of Computex 2025, Intel showed industry-shocking data: Ruixuan Pro B60 is equipped with 24GB GDDR6 graphics memory, which is three times that of NVIDIA A1000. This is not only reflected in the hardware parameters, but also directly transformed into practical application advantages – when running large models such as Llama3, the generation speed of B60 is up to 2.7 times ahead of its rivals.

    What is more worthy of attention is the multi-card interconnection scheme. Through the Project Battlematrix platform, eight B60 combinations can provide 192GB of video memory, which is enough to control a medium-scale AI model with 150 billion parameters. This kind of “stacked innovation” just hits the pain point of the current AI reasoning market – the thirsty demand of large models for graphics memory.

    Market chess game: Intel’s two-line battle strategy

    Chen Liwu’s “to tell the truth” strategy is reshaping Intel’s product route. On the one hand, use the B60 series to attack the high-end AI reasoning market held by NVIDIA, and on the other hand, harvest cost-effective users through the $299 B50. This approach of “high-end branding and mid-end grabbing share” is the same as AMD’s strategy against Intel.

    But the real killer lies in the layout of the Chinese market. Due to export controls, it is difficult for NVIDIA’s latest flagship chip to enter China, and the B60/B50 series, which meets the regulations, just fills this vacuum. The relevant person in charge of Intel said bluntly that this will “cater to the market demand for domestic AI reasoning model computing power”.

    Industry changes: from single-point breakthrough to ecological war

    NVIDIA’s moat is never just hardware. The software barriers built by CUDA ecology make challengers break back again and again. But this time, Intel is obviously prepared – AIAssistantBuilder open source software stack, containerized deployment scheme, ISV certification system, these supporting measures directly point to NVIDIA’s life.

    The management reform of “directly going deep into the seventh and eighth floors to listen to the opinions of engineers” revealed by Chen Liwu at the dinner may be the key to Intel’s revitalization of engineering culture. When the chip war enters the three-dimensional confrontation stage of “hardware + software + ecology”, execution will become the X factor that determines victory or defeat.

    The duel is far from over. NVIDIA’s next-generation products are on the way, and Intel needs to prove that the success of the B60 is not a pasy. But in any case, 2025 is destined to be a watershed in the AI chip market – when the challenger takes out three times the memory configuration of the king, the rules of the whole industry are being rewritten. As Chen Liwu said, “Let the results speak”, consumers will eventually vote with their feet, and the balance of the market has begun to tremble slightly.

  • From Macintosh to macOS: Apple’s 40-year-old file sharing protocol AFP is about to end

    It is reported that Apple plans to completely remove the Apple Filing Protocol (AFP) in future macOS updates.

    After Apple launched Macintosh in 1984, it developed and launched the AFP proprietary file sharing protocol in 1985, which together with AppleTalk, AppleShare and LocalTalk hardware, formed Apple’s early network. Network ecology, supports file sharing between Mac devices in the local area network.

    In the 1980s before the popularization of the Internet, AFP solved the problem of cross-platform file corruption caused by resource forks technology in early Mac software to ensure file integrity. IT Home quoted a blog post that Mac software files at that time included data fork and resource fork. If it was transmitted to non-Mac devices without compression, the resource fork was often stripped off, resulting in file damage.

    With the commercialization of the Internet in the mid-1990s, Apple gradually abandoned its own protocols and turned to open standards such as TCP/IP. AFP once supported TCP/IP networks through the AppleShare IP version and was integrated into Mac OS X until OS X 10.9 (Mavericks), after which the default file sharing protocol was switched to Microsoft’s SMB.

    macOS 11 (Big Sur) has removed the AFP server function, while macOS Sequoia 15.5 further announced the abandonment of AFP client support, and future versions will be completely removed. Apple recommends that users switch to SMB or use the NFS protocol for old UNIX servers.

    For users who still need AFP, the third-party open source project Netatalk provides support, but since Apple has removed macOS native AFP core support, its future sustainability is doubtful.

    In addition, third-party NAS manufacturers such as Synology may soon stop supporting AFP. Apple emphasized that although AFP is known for its simplicity and reliability, it has lost its practicality in the network environment dominated by TCP/IP and Windows, and it is time to retire.

  • Warcraft ICC’s strongest divine outfit, a single piece increases the damage of 2000 seconds! Orange Axe has to stand aside.

    After the World of Warcraft “Wrath of the Witch King” nostalgic server opened ICC, a large number of players have experienced the charm of this ultimate group book. Even the experience of ordinary mode is much better than that of bento TOC.

    For Tuanben, of course, what players are most concerned about is the drop of equipment and self-improvement. In the face of all kinds of powerful equipment of ICC, one of them can be called the strongest divine costume. In front of it, any unwilled amulet box is not enough.

    Open NetEase News to view wonderful pictures

    Ranking of major accessories after the opening of ICC

    After the opening of ICC, what players are most concerned about is the drop of several best accessories. At present, the four major DPS accessories are also ranked first on the price list. The will of death is the most expensive, with an average price of about 85,000G, while the best of the two legal products, amulets and foreign objects, are between 6-7 thousand.

    Open NetEase News to view wonderful pictures

    But in terms of the magnitude of improvement, they are not as good as the artifact of [Little Hate in the Bottle].

    Next, let’s learn about the horror improvement brought by this accessory through simulation and actual combat data!

    Actual combat data, the bottle is an absolute divine outfit.

    According to the sharing of the disciplinary rider who got the [Little Hate in the Bottle] at the first time, the number of attacks of the accessory special effect “Anger Gush” is 90 times, bringing 3.81% damage. At the same time, the 90 “Holy Seals of Truth” brought by it can be calculated that the damage ratio is 4.35%. After adding up the two, the actual damage to the bottle exceeded 8%.

    This is without 2T10. If the disciplinary ride is replaced with 2T10, after the special effects and the bottle interact, the chain lock improvement will be even greater!

    Open NetEase News to view wonderful pictures

    Let’s find another graduation discipline rider in the early ICC stage of the provincial service in 2024 to have a look. The proportion of “anger spurt” of 277 bottles has reached 5.33%. The 2T10 effect brought by it and the improvement of the corrosion holy seal are even greater.

    Open NetEase News to view wonderful pictures

    As a melee profession with a high utilization rate of armor-breaking attributes, the biggest reason for its strong rise in the T10 stage is that it relies on the special effects of 2T10 and the jewelry of [little hatred in the bottle]. It can be said that these two have achieved the ultimate form of punishment riding of WLK.

    Analog output: the bottle is increased by 15% in seconds! 264 can be qualitatively changed.

    We simulate the damage through the “WowSimsCN nostalgic suit simulator” that comes with the novice box, and the single DPS output of the disciplinary rider in the P4 full graduation state.

    1. In the state of 277 bottles, the second injury is 15180 (±319):

    2. In the state of 264 bottles, the second damage is 14978 (±316), which is 200 seconds less than 277 bottles:

    3. Replace the bottle with P5 Ultimate Graduation Jewelry 284 [Ruifeng Twilight Dragon Scale], with a second injury of 14064 (±359), 900 seconds less than 264 bottles:

    4. Finally, take down the bottle and empty an ornament. The second damage is 13240 (±337), which is about 2000 seconds less than the 277 bottles:

    It can be seen that the improvement brought by this jewelry is terrible. Even if it is a violent battle, the will to take 277 death will only reduce the second damage by about 1400. And the punishment riding a bottle directly increased by 2,000 points, which is equivalent to a 15% increase!

    Even if the punishment rider replaces the second-ranked BIS weapon with an orange axe, the promotion is not as big as a bottle. It’s better to put the pictures to compare, so as to avoid being told to steal the concept. We can see from the figure below that changing the 284 long handle to an orange axe will increase the damage by about 1000 seconds. But above, we have found that 284 scales have been exchanged for 277 bottles to increase the damage of 1100 seconds. Both of them are the second-ranking equipment, so the improvement of the bottle is greater than that of the orange axe!

    Open NetEase News to view wonderful pictures

    At the same time, we can also find a problem, that is, the improvement brought by the 264 bottle is a super qualitative change, and replacing it with 277 can only be said to be the icing on the cake. This is actually in line with expectations. After all, the most important thing for the improvement of bottle accessories is the change from 0 to 1.

  • Intel preheats Panther Lake processor with 16-core design

    Intel has warmed up its next-generation client processor “Panther Lake”. It will be put into production in the second half of 2025, and the “consumer-available” product will be launched in early 2026.

    Open NetEase News to view wonderful pictures

    The Panther Lake processor adopts a 16-core design, including 4 performance cores (P cores), 8 efficiency cores (E cores) and 4 low-power cores (LPE cores), with a total of 16 threads. Its basic frequency is 2.0GHz, and it is equipped with 24MB L2 cache and 18MB L3 cache. In addition, Panther Lake also has 8MB of system-level cache (SLC). It is worth noting that the lower-positioned Wildcat Lake processor will use a 4MB SLC.

    In terms of process technology, Panther Lake will adopt Intel’s latest 18A process, which is comparable to TSMC’s N3 or N2 process nodes, which can reduce power consumption and improve performance while increasing transistor density.

    Open NetEase News to view wonderful pictures

    Intel announced that the Panther Lake processor will reach a level comparable to the Core Ultra 200V “Lunar Lake” processor in terms of energy efficiency, maintaining a lead in the x86 field. In terms of high performance, the performance of Panther Lake will be similar to that of the Core Ultra 200H “Arrow Lake-H” processor.

    The Panther Lake processor will integrate the Xe3 GPU core, and the graphics performance will be significantly improved compared with the previous generation of Xe2 architecture. At the same time, the AI computing power of the processor has also reached 180 TOPS, of which the CPU contributes 10 TOPS, the GPU contributes 50 TOPS, and the NPU contributes 120 TOPS. In terms of expandability, the Panther Lake processor will support 4 PCIe 5.0 channels and 8 PCIe 4.0 channels, and is equipped with 4 Thunderbolt 4 interfaces.

    According to sources, Intel plans to launch a Panther Lake processor with 6P8E8LPE24Xe3 GPU configuration at the end of 2025. Other SKUs, such as the version configured by 4P8E4LPE12Xe3 GPU, will be launched one after another in early 2026.

  • Microsoft Edge browser was exposed to integrate the Phi-4 mini model to realize the end-side AI function.

    Blogger Leopeva64 announced his latest discovery on X on the 17th. In Edge Canary 138.0.3323.0 or later, several new feature logos related to “Phi mini” have appeared.

    One of the functions called “Prompt API for Phi mini” allows the browser to call the built-in Phi-mini model to perform generative AI-related tasks.

    Microsoft Edge browser was exposed to integrate the Phi-4 mini model to realize the end-side AI function.

    Microsoft Edge browser was exposed to integrate the Phi-4 mini model to realize the end-side AI function.

    Phi-4 is a small language model that requires low computing power and higher operating efficiency, especially suitable for reasoning classes or complex tasks. According to Microsoft, Phi-4 has about 14 billion parameters and excels in text generation.

    According to the foreign media Windows Latest, compared with integrated ChatGPT, Phi-4 can significantly reduce costs. According to Microsoft’s description, Edge’s Prompt API will support the local execution of text summary, classification, rewriting and other tasks. However, the company also made it clear that the version at this stage does not have the ability to verify the facts.

    At present, Edge has provided the “AI rewrite” function, which can complete similar functions but rely on cloud computing. The new API emphasizes the “end-side AI” capability, that is, the processing is completed directly through the local hardware of the device.

    Microsoft has also added several interfaces to support the Phi-4-mini model, including:

    Summarization API for Phi mini can use the built-in small language model (Phi-mini in Microsoft Edge) to summarize the text;

    Writer API for Phi mini, this model can be used to create text;

    Rewriter API for Phi mini, rewrite the text content.

    In addition, there are two auxiliary function signs. One is to bypass the device performance threshold and directly enable the local AI model; the other is the log tool for debugging the operation of local AI functions.

    According to IT Home, Microsoft is making the Edge browser the first AI-based browser on Windows 11. Last year, it was reported that Microsoft may be using a language model called Phi-3 to provide some AI functions for Edge, with The sample adopts the method of running locally on the device side.