FreshRSS

Zobrazení pro čtení

Jsou dostupné nové články, klikněte pro obnovení stránky.

Qualcomm teases Snapdragon X announcement

Qualcomm teases Snapdragon X announcement


Is it the Apple busting Snapdragon X Elite?

In a rather cryptic teaser, Qualcomm has hinted at a significant announcement regarding the Snapdragon X chipset, expected early this week. 

On Friday, Qualcomm released a teaser video on the official Snapdragon X social media, intriguing us with a " Keep your eyes peeled for 4/24."

The video showcases renders of a CPU chipset, which then morph into 3D representations of an "X."

There is not a lot more information to go on. It ispossible that the much-anticipated Snapdragon X Elite is on the verge of being unveiled.

In short, not a great deal. From what we can gather, Qualcomm's announcement slated for April 24, 2024, is likely about a new chipset.

Qualcomm has a couple of Snapdragon X chips up its sleeve. It could be the eagerly awaited debut of the Snapdragon X Elite or perhaps the formal acknowledgement of the Snapdragon X Plus. There's even a chance it might be both.

Considering the leaked images of the Lenovo Yoga Slim 7 Snapdragon Edition that surfaced online earlier this week, this announcement might just be the grand reveal of the X Elite chipset.

Qualcomm's Snapdragon X Elite is poised to be the most formidable ARM chipset thus far, boasting specialised AI PC capabilities and up to 45 TOPs of NPU performance. For those watching the pennies, an entry-level variant of the chipset will be marketed as the Snapdragon X Plus.

The X Elite outperforms the Apple M3 chip by 21 per cent on Geekbench. Moreover, an imminent Windows 11 AI feature might be exclusively available on Snapdragon X Elite PCs.

Zen 5 APU rumours are cranking up

Zen 5 APU rumours are cranking up


Strix Halo has more grunt than the PS5

The Dark Satanic Rumour Mill, specifically the Chiphell forums, has manufactured a hell-on-earth yarn that the new Zen 5 APUs, dubbed Strix Halo, are a bit of a powerhouse.

We're talking a whopping 40 Compute Units of AMD's RDNA 3.5 graphics, which is more grunt than the PlayStation 5's got under the hood.

Speculations suggest that AMD might be brewing two versions of this technology. One is a robust 16-core variant with the full 40 CUs for the GPU, and the other is a slightly more restrained 12-core with 32 CUs. The buzz is that the top-tier chip could rival the RTX 4060 Ti desktop GPU in 3DMark Time Spy.

However, it's important to remember that these are just rumours, not confirmed facts. While AMD has hinted at the existence of Strix Halo through discussions about their open-source ROCm platform, the actual specifications are still up in the air.

Now, if we're to believe the buzz, these chips aren't just about core counts. They reckon the GPU can clock up to a nippy 3GHz, a smidge faster than the PS5 can brag about. Of course, there's more to graphics oomph than just speed—things like how much memory the thing can juggle and how quickly it can do it are essential. But if this rumour's on the money, it's a bit of a corker.

The plot thickens with a supposed peek at the chip layout, showing a GPU that's practically a giant compared to the CPU dies AMD usually plays with. We're talking about 200 square millimetres of silicon real estate, while the CPU dies are more like 80 to 85 each. The word is that the 16-core beast will pair up two of those dies and could hit speeds up to 5.8GHz.

Memory-wise, it's all shared between the CPU and GPU, much like the PS5 or Xbox Series X tech. The grapevine has it that these chips will come with a 256-bit memory interface, packing LPDDR5X-8533, which should give you a hefty 500GB per second of bandwidth. That's not too shabby, and it's what you'd expect in the latest consoles.

.AMD knows its onions when it comes to APUs. Its kit is under the bonnets of consoles and handhelds like the Steam Deck and ROG Ally. These gadgets all have their own performance goals and power needs, but there's no denying AMD's knack for whipping up a top-notch APU with solid integrated graphics.

Its biggest problem, though, is cooling. While the PS5 and Xbox Series X have some fancy cooling tricks up their sleeves, they've got room to spare. A laptop's a different kettle of fish—space is tight, and these rumoured Strix Halo chips might struggle to flex their entire muscle in such snug quarters. Still, they're said to be aimed at the big boys of gaming laptops, which usually have a bit more breathing room and beefier cooling setups.

Torvalds thinks AI is a huge laugh

Torvalds thinks AI is a huge laugh


That is, until they replace him

At The Linux Foundation's Open Source Summit North America, Linus Torvalds dismissed the latest AI software as hilarious.

He said that: “Maybe I'll be replaced by an AI model!"

Torvalds summed up his attitude as, "Let's wait ten years and see where it goes before we make all these crazy announcements."

Contrary to dismissing AI, Torvalds acknowledges its potential, stating, "That's not to say I don't think AI will be helpful in the future." He even points out a positive outcome already: "NVIDIA has gotten better at talking to Linux kernel developers and working with Linux memory management" due to its necessity for running AI's large language models (LLMs) efficiently. This recognition of AI technology progress will likely instil confidence in AI and Linux development professionals.

Torvalds is also "looking forward to the tools actually to find bugs. We have a lot of tools, and we use them religiously, but making the tools smarter is not a bad thing. Using smarter tools is just the next inevitable step. We have kernel rewriting tools with very complicated scripts and pattern recognition. AI can be a huge help here because some of these tools are very hard to use, and you have to specify things at a low enough level."

Just be careful, Torvalds warns of "AI BS.".

MediaTek announces Dimensity 6300 SoC

MediaTek announces Dimensity 6300 SoC


6nm update of the Dimensity 6100+

MediaTek has announced its newest Dimensity 6300 SoC that will be coming to mid-range smartphones this year. Built on a 6nm manufacturing at TSMC, the Dimensity 6300 SoC is an octa-core SoC with two Cortex-A76 cores and six Cortex-A55 cores, paired up with Mali-G57 MC2 GPU.

According to MediaTek, the new Dimensity 6300 SoC should offer a 10 percent improvement in CPU performance and up to 50 percent GPU performance improvement, compared to the Dimensity 6100+ SoC introduced last year. As said, the Dimensity 6300 is an octa-core design with two Cortex-A76 clocked at 2.4GHz and six Cortex-A55 cores clocked at 2.0GHz. The CPU is paired up with the Mali-G57 MC2 GPU, and MediaTek claims we'll see 11 percent higher power efficiency and up to 13 percent higher FPS in games.

The new GPU supports 10-bit HDR AMOLED screens with up to 120Hz refresh rate, up to 108MP camera system, and the latest 5G modem technologies with a boost in downlink performance up to 20 percent, reaching 3.3Gb/s. It also comes with MediTek's UltraSave 3.0+ technology.

The first phone that should come with the new Dimensity 6300 SoC is the Realme C65 5G, launching next month.

mediatek dimensity6300 1

 

Z80 chip finally retired

Z80 chip finally retired


Been available for 48 years

After a venerable 48-year tenure, Zilog is bidding farewell to the Z80. Initially an offshoot from the Intel 8080 project, this chip ascended to fame as one of the most cherished and extensively utilised 8-bit CPUs in gaming and general computing.

The legendary integrated circuit, crafted by Federico Faggin, will only be available for a few months for enthusiasts to place their orders before Zilog's manufacturing ally withdraws support for this venerable technology.

According to Techspot, Zilog's announcement states that the Wafer Foundry Manufacturer (WFM) will halt the acceptance of "last time buy" (LTB) orders for the residual Z80 products come mid-June.

Zilog will then orchestrate and timetable LTB orders for the Z80 based on customer demand, while WFM will subsequently confirm the actual delivery dates. Depending on the aggregate LTB demand, the company might enforce more stringent stipulations on the minimum and maximum order quantities.

Federico Faggin, erstwhile Intel engineer and Zilog's founder embarked on this venture in 1974 following his involvement with the Intel 4004, the pioneering 4-bit CPU. The Zilog Z80 debuted in July 1976, envisioned as a software-compatible "extension" and refinement of the Intel 8080 processor.

Conceived by a small team of 12, the Z80 achieved extraordinary success, propelling Zilog to inaugurate its own chip fabrication facilities and swell its workforce to over a thousand within a mere two years. Initially intended for embedded systems, much like its Intel progenitor, the Z80 evolved into a pivotal component in gaming hardware from the 1970s through to the mid-1980s.

The Z80's influence reverberated through the realms of home computers and gaming consoles, leaving an indelible mark on the industry. It powered Sega's Master System and SG-1000, as well as Nintendo's Game Boy and Game Boy Color, bringing joy to countless gamers. It also played a pivotal role in the creation of many iconic arcade games, including the original Pac-Man. Beyond gaming, this 8-bit processor found utility in military hardware, musical synthesizers such as the Roland Jupiter-8, and a plethora of other electronic gadgets, underscoring its versatility and impact.

Zilog disseminated its Z80 blueprint to American firms Synertek and Mostek, bolstering Faggin's production capabilities, and to European giant SGS/STMicroelectronics. The CPU's design was subsequently emulated by manufacturers in Japan, Eastern Europe, and the Soviet Union, while conglomerates like NEC, Toshiba, Sharp, and Hitachi churned out their compatible chip variants.

Zilog has pivoted Z80 production back towards the embedded device sector, delivering advanced microcontroller offerings that maintain compatibility with the original Z80 and Z180 architectures.

AI can exploit security by reading a book

AI can exploit security by reading a book


If only humans still did that

In the latest "AI is taking over the world" news, a gaggle of computer boffins from the University of Illinois Urbana-Champaign discovered that AI agents can read security advisories and exploit real-world security vulnerabilities.

UIUC’s Richard Fang, Rohan Bindu, Akul Gupta, and Daniel Kang have penned a paper claiming that OpenAI's GPT-4 can, without any human hand-holding, exploit vulnerabilities in real-world systems if you just hand it a CVE advisory on a silver platter.

To prove their point, they gathered 15 one-day vulnerabilities, some of which are so critical they'd make your hair stand on end, and GPT-4 could exploit 87 per cent of them.

Expanding on the alarming revelation that large language models can be weaponized to automate cyber attacks in a controlled environment.

 Daniel Kang, an assistant professor at UIUC, said: "GPT-4 can execute certain exploits independently, surpassing the capabilities of open-source vulnerability scanners. This capability of GPT -4 could potentially lead to severe security breaches."

The team noted they had a smorgasbord of vulnerabilities, from websites to containers to Python packages, and more than half of them are deemed 'high' or 'critical' risk by the CVE folks."

Kang and his merry men crunched the numbers and figured out that pulling off a successful LLM agent attack would set you back a mere $8.80 per exploit.

Rivos set to churn out AI server chip

Rivos set to churn out AI server chip


Something RISCy

Rivos is smirking behind closed doors after receiving a $250 million cash injection.

The dark satanic rumour mill claims that the outfit is set to churn out its first AI server chip, giving Nvidia a run for its money. Nvidia's been hogging the AI chip market like it's their birthright, but Rivos is ready to crash the party.

Rivos is playing coy about their new toy but let's just say it's built on the RISC-V architecture. That's right, the open-source darling that's got Arm, Intel, and AMD shaking in their boots because it doesn't cost a penny in license fees.

Lip-Bu Tan, the big cheese at Walden Catalyst and one of Rivos' backers is all about building hardware that's as flexible as a contortionist, just like Nvidia's CUDA.

There's a rumour that Allen Wu, the ex-bigwig of Arm China, is cooking up his own RISC-V chip firm. With a crew of Arm's old hands, they're not just dipping their toes; they're diving headfirst into the chip pool. The plot's a bit murky, though.

Will they be doing their own thing, or playing middleman for Tenstorrent in China? Either way, they're not just playing for peanuts – they're in it to win it.

Musk’s cyber truck recalled

Musk’s cyber truck recalled


Just as  well he did not sell many of them and it was not one of his rockets

In a turn of events that shocked absolutely nobody, Elon Musk's preposterous 6,000 lb. bulletproof ugly Cybertruck has been recalled due to a rather serious hiccup with the accelerator pedal.

The Tesla Cybertruck, Musk's brainchild, was trotted out in 2019, and by the tail end of 2023, Tesla was already dispatching these behemoths despite a litany of unresolved safety qualms. The stainless steel, box-like turkey, tipping the scales at an absurd 6,000 lbs and allegedly impervious to bullets, has been the butt of jokes since its inception. Musk's attempt to showcase its "bulletproof" windows ended in a farce when a couple of them shattered at the toss of a metal ball.

As the deliveries commenced, many issues remained unaddressed: the stainless steel's contribution to a minuscule crumple zone, the hood's length and truck's height causing visibility woes, the absence of a rear window, and a dubious steer-by-wire system.

To top it off, Musk's penchant for sidelining safety measures in his rush to market is well-documented. The necessity for a 6,000 lb bulletproof truck raised eyebrows, leaving many to ponder the intended use of such a vehicle. Thus, it was hardly a shocker when every Cybertruck was summoned back shortly after delivery.

Tesla was compelled to recall a staggering 3,878 Cybertrucks produced between November 13, 2023, and April 4, 2024, which presumably accounts for every Cybertruck in existence.

The National Highway Traffic Safety Administration's recall notice pinpointed a defect where the accelerator pedal could become wedged when depressed.

The report cited an "unauthorised alteration" during production, where "lubricant (soap)" was applied to the pedal, diminishing its "retention." A Cybertruck owner highlighted the issue on TikTok, showing how the excessively long pedal could slip into a gap and get stuck.

While the brake pedal remained operational and could override a jammed accelerator, lifting one's foot off the pedal could catapult the vehicle to perilous velocities. This was particularly alarming given the Cybertruck's ability to bolt from 0 to 60 mph in a mere 2.6 seconds.

Consequently, all delivered Cybertrucks were recalled for pedal replacements. Yet, one must query why the myriad other flaws remain unrectified. For instance, the vehicle's dismal visibility was so pronounced that Musk was caught on film nonchalantly mowing down a road sign with his Cybertruck.

The recall sparked a flurry of attention on social media, with many curious about the number of Cybertrucks sold. The paltry figure is quite telling, especially against numerous recorded malfunctions. One driver recounted how a routine car wash incapacitated their Cybertruck, leaving it inoperative and emitting odd sounds from the dashboard for five hours, much to the bewilderment of social media onlookers.

Judging by the scant number of Cybertrucks sold and the frequency of failures showcased on social media, one might deduce a relatively high failure rate. Moreover, this isn't Tesla's sole recall debacle. Last year, all self-driving Teslas were recalled due to an autopilot glitch. With such a litany of blunders, speculation abounds about when Tesla might crumble or face a class-action lawsuit.

The silver lining, if one can call it that, is the mere 3,878 Cybertrucks in circulation, reducing the likelihood of encountering one and reassuring us that not every member of the incel brigade has managed to arm themselves with a Cybertruck just yet.

Now, Musk's Cybertruck's colossal flop is laid bare for all to see, with a complete recall and a laughably low delivery count, despite Musk's grandiose proclamations that his truck is the next revolutionary vehicular marvel.

 

Galax rolls out low-profile RTX 4060 graphics card

Galax rolls out low-profile RTX 4060 graphics card


Dual-slot, triple-fan cooler

Galax has unveiled a rather interesting low-profile graphics card based on AD107 GPU, the Galax RTX 4060 low-profile. Although currently only available in Japan, the new low-profile RTX 4060 could be coming to the rest of the world, as it might be quite popular in the SFF market.

Due to the smaller PCB needed for AD107 and four memory chips, the Galax RTX 4060 low-profile is 18.2cm long and 6.9cm tall. In order to keep the GPU well-cooled, Galax went for a dual-slot cooler with three 40mm fans. The RTX 4060 is perfect for low-profile design, as both Asus and Gigabyte have a similar graphics card.

In case you missed it, the AD107 GPU packs 3,072 CUDA cores, 8GB of GDDR6 memory on a 128-bit memory interface. Galax also implemented a minor factory overclock to the GPU, but it still needs a single 8-pin PCIe power connector. It also comes with four display outputs, two DisplayPort 1.4a and two HDMI 2.1a ports.

Hopefully, we'll see the new Galax RTX 4060 low-profile on retail/e-tail shelves soon. (via Videocardz.com)

galax rtx406lp 1

 

 

GPU-Z updated to version v2.59.0

GPU-Z updated to version v2.59.0


Fixes issues with Resizable BAR

TechPowerUp has released the newest version of its GPU-Z, a comprehensive graphics sub-system information, diagnostic, and monitoring utility. The newest version, v2.59.0, fixes several issues with Resizable BAR, as well as adds support for new Nvidia Ada GPUs.

According to the release notes, the new GPU-Z v2.59.0 fixes several issues seen with the earlier GPU-Z version, including the "unsigned driver cannot load" error on Windows 7/8 and several issues with Resizable BAR status. In addition, it also adds support for Nvidia RTX 2000 Ada and RTX 1000 Ada Laptop GPUs.

You can find the full release notes and download the latest GPU-Z v2.59.0 over at Techpowerup.

 

❌