Hacker News Reader

Hacker News

Comment by dfex | Hacker News Reader

Full thread

dfex

Am I mistaken or has Intel pretty much shelved the Tofino switching hardware that supports P4 in the first place?

I seem to recall Oxide having to switch suppliers over this?

View on HN

Yes, Intel canceled Tofino two years ago.

I'm not complaining but it's weird that they're open sourcing the SDK now. Maybe it's to support Mount Evans.

Releasing dead software that is used only to support dead hardware?

That's fair enough.

At least you can still make use of that hardware, many companies should take note of this. You can make hardware and software that doesn't die once someone pulls the plug on the other side.

Still, of course it would've been better to have released it sooner.

I think half of Tofino's complexity was in their compiler. So it may inspire new hardware vendors to reuse it in some contexts.

This was the post by Bryan Cantrill of Oxide on the Tofino saga with Intel

https://bcantrill.dtrace.org/2024/12/08/why-gelsinger-was-wr...

I also was told tofino was looking EOL. Like NUC, dropped from some C suites KPI set and longterm roadmap.

I'd love to be wrong, this is just what people said.

However, NUC form factor is more lively than ever before. Maybe someone can grab the language and run for something? Miktrotik guys come into my mind.

It’s complicated. The NUCs had a lot of inevitable catastrophic hardware failures, like NIC ports that would break. Their problem is they were not good.

How is a ethernet port breaking "inevitable"?

100% of the Broadwell NUCs, AFAIK, eventually have the problem, but many were trashed before they did. These issues persisted for years, such as with 10th and 12th generation NUCs. Search "proxmox nuc NIC not working".

Broadwell is gonna broadwell. I won't ever trust their NICs ever again. They've been way too much pain with them both personally and professionally.

Their firmware is crap quality, and their bugs are just absolutely astoundingly bad.

At one job we literally had to fix their firmware for them after several months of back-and-forth, engineers spotted the absurdly obvious bug in minutes of seeing their code.

What's the best "NUC-like" product on the market now? I know ASUS has their lineup of "NUC" that was spun off from Intel. I have used MSI's "Cubi" which was a pretty nice kit. And I know everyone drools over Minisforum stuff (which is expensive but very nicely designed). Any other notable ones?

Depends on what you want. Currently my focus is N100 based systems to use as lightweight personal computers and small servers (if you can find fanless versions).

I have BeeLink S12 Pro, which I post this comment from, and I like it. However I/O wise, it's not the fastest (x1 NVMe + 2.5" SATA SSD). I also have a GMKTec G3, which has a x2 NVMe and M.2 SATA Slot. I have recently seen a "Minix Z100" fanless, which has x4 NVMe but no SATA, however it has 2.5gbE. Lastly, my parents are using a similar N95 system on Windows 11, and it works pretty nicely for them. All of these systems support 32GB of single channel RAM, but they're plenty fast for simple jobs or simple desktop systems which you won't try to conquer the world.

If you have no power/heat/noise/money upper limit, you can get them in more packed configurations, but that's not what I need, actually.

NUC somewhat avoided the google^Wintel graveyard - they sold at least the branding to Asus.

NUCs are shifting completely to ASUS who is going to continue working on them and there are some long term commitments (there are industrial variants for example)

Apropos:

https://github.com/oxidecomputer/tofino

https://www.infoq.com/presentations/tofino-2/

They didnt switch supplier iirc they are still using Tofino since it is is still a capable hardware and they see it being useful for years to come

They spoke on their podcast (I think it was there?) about ditching Tofino for the next generation of the Oxide computer. So it sounds like the current model will always ship with Tofino, but due to no future product development they won't use it again in a new machine. It sounded like they had just secured a replacement for the future but I can't remember who it was.

We've talked about it a bunch, most recently when talking about Intel after Gelsinger.[0] I went into more detail on Intel's total mishandling of Tofino in my blog entry describing why Gelsinger was the wrong choice to lead Intel in 2021.[1]

As you might imagine, this move from Intel is something that we at Oxide have advocated for strenuously -- and it is a tremendous tribute to the former Tofino team at Intel that this got done. As I hope I made clear in my blog entry: the folks working on Tofino at Intel have been great to work with; they deserved much better than their (former) executive leadership.

[0] https://oxide-and-friends.transistor.fm/episodes/intel-after...

[1] https://bcantrill.dtrace.org/2024/12/08/why-gelsinger-was-wr...

Thanks for the context, your respect for the Tofino team certainly came through in the podcast.

Thanks Brian [0] was a great podcast and got me through a large lawn mowing!

Does this move to open source P4 have any impact on Oxide at all, or is it too little too late?

We definitely advocated for it (at all stages of our interaction with Tofino -- but especially strenuously after the part was killed), and it does have impact on us in that it was one of the few parts of our stack that we couldn't open (and now we can -- stay tuned there). We remain big believers in P4 and want to see P4 compilers for other switching silicon, so getting an open is a big step in that direction. I would also say that what Intel did here beat our expectations, and includes many aspects of their software that we didn't think they would make available. So even though we will be moving beyond Tofino, getting their software open source is great and we're very supportive -- and if anything, it just sharpens for us what a poor strategic decision it was to kill Tofino.

P4 has more or less gone nowhere. Tofino was a full generation behind and didn’t make sense. P4 was compelling because people thought they’d solve the Elephant flow problem with traffic engineering in P4 but the resources to actually do this at scale never materialized for many reasons.

ehh scream SDN 5 times. kinda miss the 2010’s now.

cisco silicon one uses p4 fwiw. internal development though, but the language makes sense for what the things are.

do they expose the p4 functionality externally? ive heard this from them but never actually seen the proof - it seems like vaporware

I don't think they would lie but they just don't release the compiler. Maybe they give it to Meta.

yes,they might provide it to hyperscalers or allows them to customize some of the packet parsing pipeline for these companies.

P4 as a language is still used on some programmable NICs for describing packet parsing dataplanes and exists as an HDL. Unfortunately the SDKs are not always exposed to small scale companies or universities like Intel did with Tofino.

Reply on news.ycombinator.com