Hacker News Reader

Hacker News

Comment by jmclnx | Hacker News Reader

Parent Full thread

jmclnx

Interesting, I wonder why, I can only guess AI ? But I was unaware Mainframes were used for AI.

View on HN

Nothing do to with AI. Its the upgrade to the new Z Generation.

https://www.redbooks.ibm.com/abstracts/sg248579.html https://www.redbooks.ibm.com/redbooks/pdfs/sg248579.pdf

Nobody dropping Mainframes.

The last 2 generations of IBM Z CPUs have "AI" inference acceleration built in. One of the use-cases was real-time credit-card fraud detection.

IBM also has a PCIe add-in card for AI called Spyre, that's also available for POWER11 systems.

https://www.ibm.com/new/announcements/telum-ii

https://research.ibm.com/blog/spyre-for-z

> But I was unaware Mainframes were used for AI.

IBM was front and center with AI long before the AI bubble.

- Watson won jeopardy in 2011. And IBM launched several successful AI products using the tech.

- Deep Blue beat Kasparov in chess in '97. They also had other NN-based systems for playing games.

Was, IBM hired me to teach watson law, what I saw as a mess, more management than developers. I was laid off 5 weeks after starting, the project was cancelled within the year

IBM is too dysfunctional to innovate like Big Tech has been

And yet Big Tech depends on technology owned by IBM, and also has luck the company isn't one that routinely does lawsuits due to patents.

Anything Red-Hat touches, like GNOME, GCC, Linux kernel, postman, anything Java is mostly done by Oracle, Red-Hat and IBM as the main ecosystem corporate drivers, PS3 used Cell,....

IBM is more often the subject of lawsuits from their employees

I'm 2nd gen, I remember the country clubs and IBM Santa. The company no longer cares about its people the way Watson father & son did.

> And yet Big Tech depends on technology owned by IBM

The opposite is true too. We live in a highly interdependent and interconnected world.

Neither were mainframes though, Watson and Deep Blue were both POWER systems.

Wasn't Watson basically a parlor trick though?

It wasn't a parlor trick and could have evolved into a useful product doing a small, basic subset of what LLMs do today. The problem was IBM's leadership didn't have the slightest understanding of the technology, thought they'd invented ChatGPT and pushed it into applications far beyond its potential, e.g. diagnosing cancer.

Deep Blue was also a bit of a parlor trick. It relied on a ton of special-purpose hardware - literally a rack of custom-made chess ICs. It's neat that it worked, but it didn't have any wider applicability.

Yeah it kind of was. Watson was essentially an NLP search engine.

They should have made an MLP search engine. Those people will buy anything.

Reply on news.ycombinator.com