TSMC To Charge Up To 30% More For Chips Made In the US

According to DigiTimes, TSMC will charge an extra 30% more for chips made in American than for chips made in Taiwan. Tom’s Hardware reports: TSMC has started discussions with customers about orders and pricing for both overseas plants, which are set to begin commercial production in late 2024. Industry insiders believe that prices of chips produced on TSMC’s N4 and N5 process technologies in the U.S. will be 20% — 30% higher than those in Taiwan, while older process chips produced in Japan’s Kumamoto facility on N28/N22 as well as N16/N12 nodes may cost 10% – 15% more than similar chips fabbed in Taiwan.

While American chip designers certainly won’t appreciate higher costs on chip production in the U.S., it is likely that they will make chips aimed at government and less price-sensitive applications in Arizona. Therefore, they should be able to pass those extra costs on to their customers without risking their competitive positions. Given the high construction and operational costs of fabs in Japan and the U.S., TSMC is going to pass those extra expenses on to customers to maintain its gross margin target of 53%.

Read more of this story at Slashdot.

China’s AI Industry Barely Slowed By US Chip Export Rules

Export controls imposed by the U.S. on microchips, aiming to hinder China’s technological advancements, have had minimal effects on the country’s tech sector. While the restrictions have slowed down variants of Nvidia’s chips for the Chinese market, it has not halted China’s progress in areas like AI, as the reduced performance is still an improvement for Chinese firms, and researchers are finding ways to overcome the limitations. Reuters reports: Nvidia has created variants of its chips for the Chinese market that are slowed down to meet U.S. rules. Industry experts told Reuters the newest one – the Nvidia H800, announced in March – will likely take 10% to 30% longer to carry out some AI tasks and could double some costs compared with Nvidia’s fastest U.S. chips. Even the slowed Nvidia chips represent an improvement for Chinese firms. Tencent Holdings, one of China’s largest tech companies, in April estimated that systems using Nvidia’s H800 will cut the time it takes to train its largest AI system by more than half, from 11 days to four days. “The AI companies that we talk to seem to see the handicap as relatively small and manageable,” said Charlie Chai, a Shanghai-based analyst with 86Research.

Part of the U.S. strategy in setting the rules was to avoid such a shock that the Chinese would ditch U.S. chips altogether and redouble their own chip-development efforts. “They had to draw the line somewhere, and wherever they drew it, they were going to run into the challenge of how to not be immediately disruptive, but how to also over time degrade China’s capability,” said one chip industry executive who requested anonymity to talk about private discussions with regulators. The export restrictions have two parts. The first puts a ceiling on a chip’s ability to calculate extremely precise numbers, a measure designed to limit supercomputers that can be used in military research. Chip industry sources said that was an effective action. But calculating extremely precise numbers is less relevant in AI work like large language models where the amount of data the chip can chew through is more important. […] The second U.S. limit is on chip-to-chip transfer speeds, which does affect AI. The models behind technologies such as ChatGPT are too large to fit onto a single chip. Instead, they must be spread over many chips – often thousands at a time — which all need to communicate with one another.

Nvidia has not disclosed the China-only H800 chip’s performance details, but a specification sheet seen by Reuters shows a chip-to-chip speed of 400 gigabytes per second, less than half the peak speed of 900 gigabytes per second for Nvidia’s flagship H100 chip available outside China. Some in the AI industry believe that is still plenty of speed. Naveen Rao, chief executive of a startup called MosaicML that specializes in helping AI models to run better on limited hardware, estimated a 10-30% system slowdown. “There are ways to get around all this algorithmically,” he said. “I don’t see this being a boundary for a very long time — like 10 years.” Moreover, AI researchers are trying to slim down the massive systems they have built to cut the cost of training products similar to ChatGPT and other processes. Those will require fewer chips, reducing chip-to-chip communications and lessening the impact of the U.S. speed limits.

Read more of this story at Slashdot.

Intel To Drop the ‘i’ Moniker In Upcoming CPU Rebrand

When Intel debuts its forthcoming Meteor Lake client processors, the company may drop its iconic “i” CPU branding and add a new moniker. Chipzilla today told The Register “We are making brand changes as we’re at an inflection point in our client roadmap in preparation for the upcoming launch of our Meteor Lake processors. We will provide more details regarding these exciting changes in the coming weeks.” From the report: The Register asked Intel about branding after semiconductor analyst Dylan Patel on Monday tweeted “Imagine you’re losing market share when you’ve been monopoly for decades, and your bright idea is to burn all brand recognition to the ground!” “That’s Intel’s plan by removing the ‘i’ in i7 i5 i3. All the decades brand recognition being lit on fire for no reason!”

Patel labelled the rebranding a “horrible very short sighted move” that won’t fix Intel’s woes and “will cause more harm than good, as many buyers know + recognize the i7 i5 branding, they won’t once it’s changed.” “The new branding sounds bad with ultra strewn about + confusing scheme.”

Patel’s mention of “Ultra” branding appears to be a reference to this benchmark result for game Ashes of the Singularity: Escalation which lists a processor called “Intel Core Ultra 5 1003H”.

Read more of this story at Slashdot.

Microsoft To Take On Apple Silicon With Custom ARM Chips

According to Windows Latest, Microsoft is working on new ARM chips to compete against Apple Silicon. “I have also spotted some job listings that suggest the company is building its own Silicon-based ARM chips for client devices” writes Mayank Parmar. “Additionally, I understand that Microsoft is optimizing Windows 12 for Silicon-ARM architecture.” From the report: These developments coincide with the upcoming launch of Windows 12, which has a special version optimized for silicon and designed to leverage AI capabilities. The job listings (most of them have now been taken down) describe positions related to custom silicon accelerators, System on Chips (SoCs), and high-performance, high-bandwidth designs. This suggests that Microsoft is building its own ARM-based chips, aiming to compete with Apple’s M chips lineup in terms of performance and efficiency.

Read more of this story at Slashdot.

New Biocomputing Method Uses Enzymes As Catalysts For DNA-Based Molecular Computing

Researchers at the University of Minnesota report via Phys.Org: Biocomputing is typically done either with live cells or with non-living, enzyme-free molecules. Live cells can feed themselves and can heal, but it can be difficult to redirect cells from their ordinary functions toward computation. Non-living molecules solve some of the problems of live cells, but have weak output signals and are difficult to fine-tune and regulate. In new research published in Nature Communications, a team of researchers at the University of Minnesota has developed a platform for a third method of biocomputing: Trumpet, or Transcriptional RNA Universal Multi-Purpose GatE PlaTform.

Trumpet uses biological enzymes as catalysts for DNA-based molecular computing. Researchers performed logic gate operations, similar to operations done by all computers, in test tubes using DNA molecules. A positive gate connection resulted in a phosphorescent glow. The DNA creates a circuit, and a fluorescent RNA compound lights up when the circuit is completed, just like a lightbulb when a circuit board is tested.

The research team demonstrated that:

– The Trumpet platform has the simplicity of molecular biocomputing with added signal amplification and programmability.
– The platform is reliable for encoding all universal Boolean logic gates (NAND, NOT, NOR, AND, and OR), which are fundamental to programming languages.
– The logic gates can be stacked to build more complex circuits.

The team also developed a web-based tool facilitating the design of sequences for the Trumpet platform. “Trumpet is a non-living molecular platform, so we don’t have most of the problems of live cell engineering,” said co-author Kate Adamala, assistant professor in the College of Biological Sciences. “We don’t have to overcome evolutionary limitations against forcing cells to do things they don’t want to do. This also gives Trumpet more stability and reliability, with our logic gates avoiding the leakage problems of live cell operations.”

“It could make a lot of long-term neural implants possible. The applications could range from strictly medical, like healing damaged nerve connections or controlling prosthetics, to more sci-fi applications like entertainment or learning and augmented memory,” added Adamala.

Read more of this story at Slashdot.

Ukraine Is Now Using Steam Decks To Control Machine Gun Turrets

Thanks to a crowdfunding campaign dating back to 2014, soldiers in Ukraine are now using Steam Decks to remotely operate a high-caliber machine gun turret. The weapon is called the “Sabre” and is unique to Ukraine. Motherboard reports: Ukrainian news outlet TPO Media recently reported on the deployment of a new model of the Sabre on its Facebook page. Photos and videos of the system show soldiers operating a Steam Deck connected to a large machine gun via a heavy piece of cable. According to the TPO Media post, the Sabre system allows soldiers to fight the enemy from a great distance and can handle a range of calibers, from light machine guns firing anti-tank rounds to an AK-47.

In the TPO footage, the Sabre is firing what appears to be a PKT belt-fed machine gun. The PKT is a heavy barrelled machine that doesn’t have a stock and is typically mounted on vehicles like armored personnel carriers. It uses a solenoid trigger so it can be fired remotely, which is the cable running out of the back of the gun and into the complex of metal and wires on the side of the turret.

The Sabre system wasn’t always controlled with a Steam Deck […]. The first instances of the weapon appeared in 2014. The U.S. and the rest of NATO is giving Ukraine a lot of money for defense now, but that wasn’t the case when Russia first invaded in 2014. To fill its funding gaps, Ukrainians ran a variety of crowdfunding campaigns. Over the years, Ukraine has used crowdfunding to pay for everything from drones to hospitals. One of the most popular websites is The People’s Project, and it’s there that the Sabre was born. The People’s Project launched the crowdfunding campaign for Sabre in 2015 and collected more than $12,000 for the project over the next two years. It’s initial goal was to deploy 10 of these systems.

Read more of this story at Slashdot.

Spotify Tries To Win Indie Authors By Cutting Audiobook Fees

In an effort to appeal to indie authors, Spotify’s Findaway audiobook seller “will no longer take a 20 percent cut of royalties for titles sold on its DIY Voices platform — so long as the sales are made on Spotify,” reports The Verge. From the report: In a company blog post published on Monday, Findaway said that it would “pass on cost-saving efficiencies” from its integration with the streaming service. While it’s free for authors to upload their audiobooks onto Findaway’s Voices platform, the company normally uses an 80/20 pricing structure — where Findaway takes a 20 percent fee on all royalties earned. But that fee comes after sales platforms take their own 50 percent cut on the list price. So under the old revenue split, an author who sold a $10 audiobook would have to give $5 to Spotify and $1 to Findaway. But moving forward, that same author will no longer have to pay the $1 distribution fee to Findaway when a sale is made through Spotify.

The margins on audiobooks are exceptionally high, much to the chagrin of the authors. For example, Audible takes 75 percent of retail sales (though it’ll only take 60 percent with an exclusivity contract). Many authors share royalties with their narrators and have to pay production fees — meaning they get an even smaller share of royalties. The move by Spotify and Findaway is likely a bid to draw more indie authors from Audible, which is currently its biggest competitor. But Spotify’s audiobooks business — which it launched last fall — still has a long way to go. Unlike music or podcasts, most audiobooks on Spotify must be purchased individually, and sales are restricted to its web version. Even CEO Daniel Ek admitted that the current process of buying an audiobook through Spotify is “pretty horrible.” “We at Spotify are just at the beginning of our journey supporting independent authors — we have many plans for how to help authors expand their reach, maximize revenue, and ultimately build a strong audiobooks business,” said Audiobook’s communications chief, Laura Pezzini.

Read more of this story at Slashdot.

OpenAI Threatens Popular GitHub Project With Lawsuit Over API Use

A GitHub project called GPT4free has received a letter from OpenAI demanding that the repo be shut down within five days or face a lawsuit. Tom’s Hardware reports: Anyone can use ChatGPT for free, but if you want to use GPT4, the latest language model, you have to either pay for ChatGPT Plus, pay for access to OpenAI’s API, or find another site that has incorporated GPT4 into its own free chatbot. There are sites that use OpenAI such as Forefront and You.com, but what if you want to make your own bot and don’t want to pay for the API? A GitHub project called GPT4free allows you to get free access to the GPT4 and GPT3.5 models by funneling those queries through sites like You.com, Quora and CoCalc and giving you back the answers. The project is GitHub’s most popular new repo, getting 14,000 stars this week.

Now, according to Xtekky, the European computer science student who runs the repo, OpenAI has sent a letter demanding that he take the whole thing down within five days or face a lawsuit. I interviewed Xtekky via Telegram, and he said he doesn’t think OpenAI should be targeting him since he isn’t connecting directly to the company’s API, but is instead getting data from other sites that are paying for their own API licenses. If the owners of those sites have a problem with his scripts querying them, they should approach him directly, he posited. […] Even if the original repo is taken down, there’s a great chance that the code — and this method of accessing GPT4 and GPT3.5 — will be published elsewhere by members of the community. Even if GPT4Free had never existed anyone can find ways to use these sites’ APIs if they continue to be unsecured. “Users are sharing and hosting this project everywhere,” he said. “Deletion of my repo will be insignificant.”

Read more of this story at Slashdot.

‘sudo’ and ‘su’ Are Being Rewritten In Rust For Memory Safety

Phoronix reports:

With the financial backing of Amazon Web Services, sudo and su are being rewritten in the Rust programming language in order to increase the memory safety for the widely relied upon software… to further enhance Linux/open-source security.
“[B]ecause it’s written in C, sudo has experienced many vulnerabilities related to memory safety issues,” according to a blog post announcing the project:
It’s important that we secure our most critical software, particularly from memory safety vulnerabilities. It’s hard to imagine software that’s much more critical than sudo and su.
This work is being done by a joint team from Ferrous Systems and Tweede Golf with generous support from Amazon Web Services. The work plan is viewable here. The GitHub repository is here.

Read more of this story at Slashdot.

Russian Forces Suffer Radiation Sickness After Digging Trenches and Fishing in Chernobyl

The Independent reports:

Russian troops who dug trenches in Chernobyl forest during their occupation of the area have been struck down with radiation sickness, authorities have confirmed.

Ukrainians living near the nuclear power station that exploded 37 years ago, and choked the surrounding area in radioactive contaminants, warned the Russians when they arrived against setting up camp in the forest. But the occupiers who, as one resident put it to The Times, “understood the risks” but were “just thick”, installed themselves in the forest, reportedly carved out trenches, fished in the reactor’s cooling channel — flush with catfish — and shot animals, leaving them dead on the roads…

In the years after the incident, teams of men were sent to dig up the contaminated topsoil and bury it below ground in the Red Forest — named after the colour the trees turned as a result of the catastrophe… Vladimir Putin’s men reportedly set up camp within a six-mile radius of reactor No 4, and dug defensive positions into the poisonous ground below the surface.

On 1 April, as Ukrainian troops mounted counterattacks from Kyiv, the last of the occupiers withdrew, leaving behind piles of rubbish. Russian soldiers stationed in the forest have since been struck down with radiation sickness, diplomats have confirmed. Symptoms can start within an hour of exposure and can last for several months, often resulting in death.

Read more of this story at Slashdot.