@HPC Podcast Archives - OrionX.net - HPC News Bytes – 20240422

Episode Date: April 22, 2024

- New supercomputers at Los Alamos and France-CEA - Stanford AI Index Report - Bitcoin Halving - Argonne-UIC Crabtree Institute for Discovery and Sustainability [audio mp3="https://orionx.net/wp-cont...ent/uploads/2024/04/HPCNB_20240422.mp3"][/audio] The post HPC News Bytes – 20240422 appeared first on OrionX.net.

Transcript
Discussion (0)
Starting point is 00:00:00 Welcome to HPC News Bites, a weekly show about important news in the world of supercomputing, AI, and other advanced technologies. Hi, everyone. Welcome to HPC News Bites. I'm Doug Black. Hi, Shaheen. Let's start off with two big systems announcements. Los Alamos National Lab announced the installation of an HPE NVIDIA AI supercomputer
Starting point is 00:00:26 called Venado, named after a mountain in New Mexico. It's an HPE Cray-EX system powered by 2,560 Grace Hopper chips, which combine NVIDIA ARM-based CPUs with their Hopper-based GPU technology. NVIDIA's Grace Hopper so-called superchip figures prominently also in a new evident HPC system built for the military division of CEA, the French Alternative Energies and Atomic Energy Commission. It's an evident bull-sequana XH3000 supercomputer called the EXA-1HE for high efficiency with 477 compute nodes that use warm water for cooling. It will have a peak performance of 104 petaflops and can achieve a LIMPAC performance of about 60 petaflops, placing it just about number 15 on the top 500 list. The fifth annual AI Index Report from Stanford came out last week. It's a long and detailed view of the state of AI.
Starting point is 00:01:27 It covers how AI tools can help humans, but also do better than humans in a number of tasks. Challenges of testing AI models for responsible behavior. Increasing regulations as setting AI policy becomes more urgent for governments. We covered EU's leading role in this area in a previous episode. Exploding costs, which partly explains the dominant role of industry in setting the pace versus universities or governments. And even more relevant to HPC, AI's impact on scientific discovery. It also includes a list of top 10 takeaways that ends with the growing global awareness of AI, but also nervousness about AI. A survey by
Starting point is 00:02:06 Pew suggests just over half, 52% of Americans, report feeling more concerned than excited about AI. That's a 15% jump from two years ago. Yeah, the report has many contributors, but the two co-directors are Ray Perreault of the SRI International Research Organization and Jack Clark, co-founder of AI company Anthropic, which has attracted more than $4 billion in investment funding from Amazon. You know, we should invite them as guests on our HPC podcast. I thought an interesting finding is the positive impact of AI on productivity. They say several studies issued last year suggest that AI already enables workers to complete tasks more quickly and with higher quality, and that AI can bridge the gap between low and high-skilled workers.
Starting point is 00:02:55 But they also warn that using AI without proper oversight can lead to diminished performance. Now, Shaheen, we know you're a Bitcoin proponent, and we've discussed its place in the advanced computing landscape of the future in which cryptocurrencies will play a central role in global business transactions. This past week, Bitcoin had a halving. Perhaps you can explain why this is significant. Yeah, Bitcoin was introduced in 2009 with an asymptotically increasing supply capped at 21 million coins. These coins are introduced into the system gradually with each transaction block. There are 144 of those blocks per day,
Starting point is 00:03:32 and they're awarded to so-called miners, those who validate transaction blocks based on an incentive scheme that rewards participation and puts an exponentially high cost on attempts at hijacking the system. Bitcoin started out issuing 50 coins for each block to be halved every 210,000 blocks. Each block takes about 10 minutes, so that works out to about four years. The idea is to gradually move to transaction fees to compensate validators as the system gets better established. The first Bitcoin block with a reward of 50 Bitcoins was mined by Satoshi in January 2009. The first halving block took place a bit less than four years later in November 2012, after which the block reward dropped to 25 coins. The second halving was in July of 2016 and the third in May of 2020. This past week it halved again for the fourth time
Starting point is 00:04:26 and the block reward is now 3.125 coins per block. So far 19.7 million bitcoins have been issued, that's 93.7% of total, so the growth in bitcoin supply is now a bit smaller than it is for gold and the last bitcoin will take decades to mine. Segueing into our next story, AI has now replaced Bitcoin as the main culprit for data center energy use. As an industry, we all understand the importance of finding ways to reduce energy consumption without loss of performance. It's one reason I think quantum computing will be interesting, even if it is not faster. So we had welcome news this week from Argonne National Lab and the University of Illinois at Chicago, forming the George Crabtree Institute for Discovery and Sustainability, named for a pioneering researcher on sustainability energy, energy policy, and material science.
Starting point is 00:05:21 And leading these efforts will be founding co-directors Jordi Cabana and Michael Papka, UIC faculty members who also hold appointments at Argonne. Mike, of course, is helping to lead the installation of the Aurora Exascale supercomputer, working with Rick Stevens and the system group at Argonne. We had the pleasure of them as guests of the podcast last November, a popular episode that we encourage you to listen to. Okay, that's it for this episode. Thanks so much for being with us. HPC News Bites is a production of OrionX in association with Inside HPC. Shaheen Khan and Doug Black host the show. Every episode is featured on InsideHPC.com and posted on OrionX.net.
Starting point is 00:06:04 Thank you for listening.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.