Hacker News Reader

Comment by skissane | Hacker News Reader

Here is a good high level slide deck on z/OS 2.5 new features: https://github.com/IBM/IBM-Z-zOS/blob/main/zOS-Education/zOS...

It starts out with the marketing buzzwords but then drills down into the technical nitty-gritty. In terms of what the "AI" stuff actually means, one answer is contained on slide 121 – support for running Tensorflow and ONNX inside a Linux Docker container using zCX (which lets you run z/Linux Docker containers under z/OS). The zCX SIMD support referenced on slide 102 is probably highly relevant to this.

View on HN

If running containers on Mainframes is of interest, attending this talk at Defcon might be illuminating https://defcon.org/html/defcon-29/dc-29-speakers.html#coldwa... :)

How does that compare dollar-for-dollar to running Tensorflow on a dedicated Linux cluster with GPUs?

Mainframes also have a lot of PCIe channels you can attach GPUs to. IO capacity and hardware offloading has always been the most notable difference between a mainframe and an x86 box.

I wouldn't be surprised if, with some clever programming, one could pipe data from disk directly to GPUs and back to disk without involving the CPUs that run user programs except for the thinnest slice of time.

Do the specific Nvidia cards you'd ideally want for ML have s390x drivers?

That's something you'll need to ask Nvidia.

But, considering what you paid for the mainframe, you probably have the budget to commission them to write drivers just for you (or to license their source and compile it yourself).

(or to serve them a National Security Letter that says "give us your source" and compile it yourself)

Can’t stop laughing sadly.

Btw given they do linux can they twist so that channel is partition into for that linux container. In the new world just driver through.

IBM has built supercomputers with POWER machines stuffed full of GPUs, in close partnership with Nvidia, for years. I think they might still be the only ones offering systems designed for sustained NVLink bus saturation. So yeah, you could do it - and it would be awesome... but it would be a silly misuse of a mainframe where a cluster would make so much more sense.

Probably much more expensive.

Probably depends on whether you've already got the mainframe or not, and whether you've got the dedicated Linux cluster or or not.

"one answer is contained on slide 121" => right there, I decided not to open the slide deck.

You have to go through 100 slides before getting to technical details?

You have to go through 100 slides before getting to technical details for this narrow specific particular topic that is being discussed here

A new OS is a significant endevour and there is a lot of release notes on any and all topics - technical, usability, timelines, compatibility, hardware, architecture, support, upgrade methodology, etc.

Reply on news.ycombinator.com