Data analysis

Cerebras brings CS-2 system to biz nference data analysis • The Register

AI chip startup Cerebras Systems has deployed one of its CS-2 systems in a well-funded startup that uses natural language processing to analyze massive amounts of biomedical data.

As announced on Monday, nference plans to use this CS-2 to train large transformer models designed to process information from unstructured medical data stacks to provide new insights to physicians and improve recovery and treatment. patients. The CS-2 is powered by Cerebras second-generation Wafer-Scale Engine processor, so called because the chip is wafer-sized.

Cerebras said the deployment marks another significant customer win in healthcare and life sciences after installing similar systems at pharmaceutical giants GlaxoSmithKline and AstraZeneca as well as the department’s Argonne National Laboratory. US Energy for research related to COVID-19.

Andrew Feldman, CEO of Cerebras, said The register this facility at Massachusetts-based nference is another testament to Cerebras’ belief that its wafer-sized AI chips are better suited than traditional chips like Nvidia’s GPUs to analyze large amounts of data so quickly possible, which is increasingly important in areas such as healthcare and life sciences.

“It’s all extremely computationally intensive. It’s well suited to these new computational techniques and artificial intelligence. And those are exactly the techniques that we’re hundreds of times faster [at] than [Nvidia],” he said.

In the case of nference, the Mayo Clinic-funded biomedical startup will use Cerebras’ CS-2 system to train self-supervised learning models on large amounts of unstructured medical data, which can include patient records, scientific articles, medical imaging and genomic data. data base.

The unstructured nature of this data, which varies by file format, can be a huge headache for data scientists and machine learning researchers to process using traditional computational methods, according to Cerebras. The AI ​​chip startup said it can even force researchers to resort to the tedious and inefficient task of sifting through documents by hand.

“These are some of the most fertile areas of research. They have large amounts of data, and the human genome and other genomes are among them. They’re incredibly large,” Feldman said.

What makes Cerebras’ chips so suitable for large data sets like this is their large size. While other semiconductor companies make multiple chips from dies cut from a single wafer, Silicon Valley’s Cerebras makes a chip from an entire wafer. This allows Cerebras to pack a huge amount of processing cores – 850,000 in the case of its latest processor – onto a chip made up of 2.6 trillion transistors.

Feldman said this makes Cerebras Wafer-Scale Engine chip 56 times larger than the previous largest chip ever made, and it crucially allows large datasets to remain on the chip during processing, dramatically reducing the need for information to be repeatedly passed in and out, which takes time. He contrasts this super-die approach with Nvidia’s architecture, which he says is slower because it relies on interconnects to transport information between individual GPUs and CPUs as they attempt to perform the same level of processing.

“When we need to move it, we move it around the chip. And that’s the fastest way to move information on demand,” he said. “So when you have to jump from a chip to a Mellanox switch or to the CPU, which is what GPUs have to do, you’re somewhere between 1,000 and 10,000 times slower than if you could keep the information on a chip and move on your piece of silicon.”

What makes Cerebras’ CS-2 system attractive for inference is the fact that it can accommodate longer sequence lines than traditional systems, according to Feldman.

“With the CS-2, they can train transformer models with longer sequence lines than before, allowing [them] to iterate faster and build better, more insightful models,” he said. ®