Technology

Nvidia hooks TSMC and friends on GPU accelerated chip design • The Register

GTC Nvidia’s latest gambit? Entrenching itself as a key part of the semiconductor manufacturing supply chain.

At GTC this week, the chipmaker unveiled cuLitho, a software library designed to accelerate computational lithography workloads used by the likes of TSMC, ASML, and Synopsys, using its GPUs.

The idea behind the platform is to offload and parallelize the complex and computationally expensive process of generating photomasks used by lithography machines to etch nanoscale features, like transistors or wires, into silicon wafers.

“Each chip design is made up of about 100 layers and in total contains trillions of polygons or patterns. Each of these 100 layers are encoded separately into a photomask — a stencil for the design if you will — and, using a rather expensive camera, are successively printed onto silicon,” Vivek Singh, VP of Nvidia’s advanced technology group, explained during a press conference on Monday.

Originally, photomasks were just a negative of the shape engineers were trying to etch into the silicon, but as transistors have gotten smaller these photomasks became more complex to counteract the effects of optical distortion. If unchecked, this distortion can blur these features beyond recognition. This process is called optical proximity correction (OPC) and more recently has evolved into inverse lithography technology (ILT). In the case of the latter, the photomasks look nothing like the feature they’re designed to print.

And the more ornate these photomasks get, the more computational horsepower is required to produce them. However, using GPUs, Nvidia believes it can not only speed up this process, but reduce the power consumption required. The company claims that cuLitho running on its GPUs is roughly 40x faster than existing computational lithography platforms running on general purpose CPUs.

“It’ll help the semiconductor industry continue the pace of innovation that we’ve all come to rely on, and it’ll improve the time to market for all kinds of chips in the future,” Singh claimed.

However, at least in the near term, Nvidia’s expectations seem to be a little more grounded. The company expects fabs using cuLitho could produce 3-5x more photomasks a day while using 9 percent less power, which if true, should help to boost foundries’ already thin margins

And with the likes of ASML, Synopsys, and TSMC lining up to integrate Nvidia’s GPUs and libraries into their software platforms and fabs, we won’t have to wait long to see these claims put to the test.

TSMC is already investigating Nvidia’s GPUs and cuLitho to accelerate ILT photomasks, while ASML and Synopsys are working to integrate support for GPU acceleration using cuLitho in their computational lithography software platforms.

And while Nvidia execs would love to sell its latest and most expensive GPU architectures to these companies, Singh notes that the library is compatible with GPUs going back to the Volta generation, which made its debut in 2017.

While Nvidia is using GPUs to accelerate these workloads, it’s worth noting that cuLitho isn’t using machine learning or AI to optimize semiconductor design just yet. But it’s no secret that Nvidia is also working on that particular problem.

“Much of this has to do with accelerating the underlying primitive operations of computational lithography,” Singh said. “But I will say that AI is very much in the works in cuLitho.”

As our sister site The Next Platform reported last summer, Nvidia has been working on ways to accelerate computational lithography workloads for some time now. In a research paper published in July, engineers at the company used AI to design equivalent circuits 25 percent smaller than those created using traditional EDA platforms.

Nvidia is hardly the only company investigating the use of machine learning to accelerate circuit design. Synopsys and Cadence have both implemented AI technologies into their portfolios, while Google researchers developed a deep-learning model called PRIME to create smaller and faster accelerator designs. And previously, the company used reinforcement learning models to design portions of its tensor processing unit (TPU).

With that said, the addressable market for something like cuLitho isn’t that big, and thanks to efforts by the US Commerce Department to stifle China’s fledgling semiconductor industry, the number is only getting smaller.

cuLitho will almost certainly be subject to US export controls governing the sale of advanced semiconductor manufacturing equipment and software to countries of concern, which for the moment means China. Pressed on this point, Singh said the library would be “available wherever this end-to-end OPC software is available,” but declined to comment further on US trade restrictions. ®

Source link

Technology

Typo blamed for Microsoft Azure DevOps outage in Brazil • The Register

Microsoft Azure DevOps, a suite of application lifecycle services, stopped working in the South Brazil region for about ten hours on Wednesday due to a basic code error.

On Friday Eric Mattingly, principal software engineering manager, offered an apology for the disruption and revealed the cause of the outage: a simple typo that deleted seventeen production databases.

Mattingly explained that Azure DevOps engineers occasionally take snapshots of production databases to look into reported problems or test performance improvements. And they rely on a background system that runs daily and deletes old snapshots after a set period of time.

During a recent sprint – a group project in Agile jargon – Azure DevOps engineers performed a code upgrade, replacing deprecated Microsoft.Azure.Managment.* packages with supported Azure.ResourceManager.* NuGet packages.

The result was a large pull request of changes that swapped API calls in the old packages for those in the newer packages. The typo occurred in the pull request – a code change that has to be reviewed and merged into the applicable project. And it led the background snapshot deletion job to delete the entire server.

“Hidden within this pull request was a typo bug in the snapshot deletion job which swapped out a call to delete the Azure SQL Database to one that deletes the Azure SQL Server that hosts the database,” said Mattingly.

Azure DevOps has tests to catch such issues, but according to Mattingly, the errant code only runs under certain conditions and thus isn’t well covered under existing tests. Those conditions, presumably, require the presence of a database snapshot that is old enough to be caught by the deletion script.

Mattingly said Sprint 222 was deployed internally (Ring 0) without incident due to the absence of any snapshot databases. Several days later, the software changes were deployed to the customer environment (Ring 1) for the South Brazil scale unit (a cluster of servers for a specific role). That environment had a snapshot database old enough to trigger the bug, which led the background job to delete the “entire Azure SQL Server and all seventeen production databases” for the scale unit.

The data has all been recovered, but it took more than ten hours. There are several reasons for that, said Mattingly.

One is that since customers can’t revive Azure SQL Servers themselves, on-call Azure engineers had to handle that, a process that took about an hour for many.

Another reason is that the databases had different backup configurations: some were configured for Zone-redundant backup and others were set up for the more recent Geo-zone-redundant backup. Reconciling this mismatch added many hours to the recovery process.

“Finally,” said Mattingly, “Even after databases began coming back online, the entire scale unit remained inaccessible even to customers whose data was in those databases due to a complex set of issues with our web servers.”

These issues arose from a server warmup task that iterated through the list of available databases with a test call. Databases in the process of being recovered chucked up an error that led the warm-up test “to perform an exponential backoff retry resulting in warmup taking ninety minutes on average, versus sub-second in a normal situation.”

Further complicating matters, this recovery process was staggered and once one or two of the servers started taking customer traffic again, they’d get overloaded, and go down. Ultimately, restoring service required blocking all traffic to the South Brazil scale unit until everything was sufficiently ready to rejoin the load balancer and handle traffic.

Various fixes and reconfigurations have been put in place to prevent the issue from recurring.

“Once again, we apologize to all the customers impacted by this outage,” said Mattingly. ®

Source link

Technology

What are the current trends in Ireland’s pharma sector?

SiliconRepublic.com took a look at PDA Ireland’s Visual Inspection event to learn about Ireland’s pharma sector and its biggest strengths.

Ireland’s pharmaceutical stakeholders gathered in Cork recently to learn the latest developments and regulatory changes in the sector.

The event was hosted by the Irish chapter of the Parenteral Drug Association (PDA), a non-profit trade group that shares science, technology and regulatory information to pharma and biopharma companies.

The association held a Visual Inspection event in Cork last month, where speakers shared their outlooks on the industry, the regulatory landscape and tips on product investigation.

PDA Ireland committee member Deidre Tobin told SiliconRepublic.com that one goal of the event was to get bring the industry together and help SMEs engage with top speakers.

“The mission of PDA is really to bring people together in industry and to have that network sharing, that information gathering so that we’re all consistent, we all have the same message,” Tobin said.

Ireland’s advantages

Ireland has grown to become a hub of leading pharma companies over the years, with many multinational companies setting up sites here. By 2017, 24 of the world’s top biotech and pharma companies had made a home for themselves in Ireland.

The sector also remains active in terms of merger and acquisition deals. A William Fry report claimed Pharma accounted for 12pc of all Irish M&A deals by volume in 2022.

Ruaidhrí O’Brien, head of UK and Ireland sales at Körber Pharma and a PDA Ireland member, said the country has a “wealth of experience” across various types of pharmaceutical production, such as API bulk and solid dosage production.

O’Brien claimed there’s also been growth in the “liquid fill finish area”, which relates to completed pharma products such as vaccines. During the Covid-19 pandemic, Pfizer confirmed its Irish operations were being used to manufacture its vaccine.

O’Brien also said Ireland has “skilled people” that are in senior levels within companies, which he feels is why existing companies continue to invest and why “we have amazing investments from all the global leaders”.

Regulatory changes

One speaker at the PDA Ireland Visual Inspection event was John Shabushnig, the founder of Insight Pharma Consulting LLC. He spoke about current and upcoming regulation impacting the global sector.

Shabushnig said he sees the overall industry understanding of what it can and can’t do “continuing to improve”. He also said there is better alignment between regulators and industry now “than I saw 10 or 20 years ago”.

Shabushnig spoke positively about the regulatory landscape overall and couldn’t think of any “big misses” in terms of industry ignoring regulation. But he did note that some developing areas in the industry are “a bit unknown”.

“Advanced therapies, cell and gene therapies, there are some unique challenges on inspecting those products that we’re kind of learning together at this point,” Shabushnig said.

But Shabushnig said there are also “big opportunities” ahead with new tools that can be taken advantage of. One example he gave was using AI for automated visual inspection, which Shabushnig described as a “very exciting tool”.

10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.

Source link

Current

Explaining AI Black Box

By Prof Saurabh Bagchi

Prof Saurabh Bagchi from Purdue University explains the purpose of AI black boxes and why researchers are moving towards ‘explainable AI’.

For some people, the term ‘black box’ brings to mind the recording devices in airplanes that are valuable for postmortem analyses if the unthinkable happens. For others, it evokes small, minimally outfitted theatres. But ‘black box’ is also an important term in the world of artificial intelligence.

AI black boxes refer to AI systems with internal workings that are invisible to the user. You can feed them input and get output, but you cannot examine the system’s code or the logic that produced the output.

Machine learning is the dominant subset of artificial intelligence. It underlies generative AI systems like ChatGPT and DALL-E 2. There are three components to machine learning: an algorithm or a set of algorithms, training data and a model.

An algorithm is a set of procedures. In machine learning, an algorithm learns to identify patterns after being trained on a large set of examples – the training data. Once a machine-learning algorithm has been trained, the result is a machine-learning model. The model is what people use.

For example, a machine-learning algorithm could be designed to identify patterns in images and the training data could be images of dogs. The resulting machine-learning model would be a dog spotter. You would feed it an image as input and get as output whether and where in the image a set of pixels represents a dog.

Any of the three components of a machine-learning system can be hidden, or in a black box. As is often the case, the algorithm is publicly known, which makes putting it in a black box less effective. So, to protect their intellectual property, AI developers often put the model in a black box. Another approach software developers take is to obscure the data used to train the model – in other words, put the training data in a black box.

The opposite of a black box is sometimes referred to as a glass box. An AI glass box is a system whose algorithms, training data and model are all available for anyone to see. But researchers sometimes characterise aspects of even these as black box.

That’s because researchers don’t fully understand how machine-learning algorithms, particularly deep-learning algorithms, operate. The field of explainable AI is working to develop algorithms that, while not necessarily glass box, can be better understood by humans.

Thinking Outside The Black Box

In many cases, there is good reason to be wary of black box machine-learning algorithms and models. Suppose a machine-learning model has made a diagnosis about your health. Would you want the model to be black box or glass box? What about the physician prescribing your course of treatment? Perhaps she would like to know how the model arrived at its decision.

What if a machine-learning model that determines whether you qualify for a business loan from a bank turns you down? Wouldn’t you like to know why? If you did, you could more effectively appeal the decision, or change your situation to increase your chances of getting a loan the next time.

Black boxes also have important implications for software system security. For years, many people in the computing field thought that keeping software in a black box would prevent hackers from examining it and therefore it would be secure. This assumption has largely been proven wrong because hackers can reverse engineer software – that is, build a facsimile by closely observing how a piece of software works – and discover vulnerabilities to exploit.

If software is in a glass box, software testers and well-intentioned hackers can examine it and inform the creators of weaknesses, thereby minimising cyberattacks.

By Prof Saurabh Bagchi

Saurabh Bagchi is professor of electrical and computer engineering and director of corporate partnerships in the School of Electrical and Computer Engineering at Purdue University in the US. His research interests include dependable computing and distributed systems.