hubris VS firecracker

> The CPUs in Oxide racks are AMD, so, presumably AMD-based compute rather than ARM.

These don’t run Hubris though; based on the chips directory in the repo [0], they’re targeting a mix of NXP and ST parts, which are Arm, and the user isn’t likely to see them or care what firmware they’re running: they’re really pretty “boring”.

[0] : https://github.com/oxidecomputer/hubris/tree/020d014880382d8...

I wouldn't put this comment here. It's not just some detail of this function; it's an invariant of the field that all writers have to respect (maybe this is the only one now but still) and all readers can take advantage of. So I'd add it to the `TaskDesc::regions` docstring. [1]

[1] https://github.com/oxidecomputer/hubris/commit/b44e677fb39cd...

With respect to Hubris, the build badge was, in turns out, pointing to a stale workflow. (That is, the build was succeeding, but the build badge was busted.) This comment has been immortalized in the fix.[0]

With respect to Humility, I am going to resist the temptation of pointing out why one of those directories has a different nomenclature with respect to its delimiter -- and just leave it at this: if you really want to find some filthy code in Humility, you can do much, much better than that!

[0] https://github.com/oxidecomputer/hubris/commit/651a9546b20ce...

A lot of questions in there! Taking these in order:

1. We aren't making standalone servers: the Oxide compute sled comes in the Oxide rack. So are not (and do not intend to be) a drop in replacement for extant rack mounted servers.

2. We have taken a fundamentally different approach to firmware, with a true root of trust that can attest to the service processor -- which can turn attest to the system software. This prompts a lot of questions (e.g., who attests to the root of trust?), and there is a LOT to say about this; look for us to talk a lot more about this

3. In stark contrast (sadly) to nearly everyone else in the server space, the firmware we are developing is entirely open source. More details on that can be found in Cliff Biffle's 2021 OSFC talk and the Hubris and Humility repos.[0][1][2]

4. Definitely not vaporware! We are in the process of shipping to our first customers; you can follow our progress in our Oxide and Friends podcast.[3]

[0] https://www.osfc.io/2021/talks/on-hubris-and-humility-develo...

[1] https://github.com/oxidecomputer/hubris

[2] https://github.com/oxidecomputer/humility

[3] https://oxide-and-friends.transistor.fm/

everywhere, for example in https://github.com/oxidecomputer/hubris/search?q=dyn

Is Box really allocating here? Is the "Rust By Example" text incomplete?

Then I had to stop learning Rust for other reasons, but this doubt really hit me at the time.

I work on an embedded OS in Rust (Hubris) that has a very bespoke build system. As part of the build system, it has to set environmental variables based on (1) the target device and (2) the specific "task"; this is an OS with task-level isolation, so tasks are compiled as individual Rust crates.

Oxide Computer told some storied about the difficulty of bring up of a new motherboard, and mentioned a lot of gotcha details and hack solutions for managing their AMD chip.

They talked about their bring up sequence, boot chain verification on their motherboard, and designing / creating / verifying their hardware root of trust.

I heard mention of this on a podcast recently, trying to find the reference.

I'm pretty sure it was [S3]

- "Tales from the Bringup Lab" https://lnns.co/FBf5oLpyHK3

- or "More Tales from the Bringup Lab" https://lnns.co/LQur_ToJX9m

But I found again these interesting things worth sharing on that search. https://oxide.computer/blog/hubris-and-humility, https://github.com/oxidecomputer/hubris

Search 1 [S1], Trammell Hudson ep mentioning firmware (chromebook related iirc) https://lnns.co/pystdPm0QvG.

Search 2 [S2], Security, Cryptography, Whatever podcast episode mentioning Oxide and roots of trust or similar. https://lnns.co/VnyTvdhBiGC

Search links:

[S1]: https://www.listennotes.com/search/?q=oxide+tpm

[S2]: https://www.listennotes.com/search/?q=oxide%20and%20friends%...

[S3]: https://www.listennotes.com/search/?q=oxide%20and%20friends%...

When we started the company, we knew it would be a three year build -- and indeed, our first product is in the final stages of development (i.e. EMC/safety certification). We have been very transparent about our progress along the way[0][1][2][3][4][5][6][7] -- and our software is essentially all open source, so you can follow along there as well.[8][9][10]

If you are asking "does anyone want a rack-scale computer?" the (short) answer is: yes, they do. The on-prem market has been woefully underserved -- and there are plenty of folks who are sick of Dell/HPE/VMware/Cisco, to say nothing of those who are public cloud borne and wondering if they should perhaps own some of their own compute rather than rent it all.

[0] https://oxide-and-friends.transistor.fm/episodes/holistic-bo...

[1] https://oxide-and-friends.transistor.fm/episodes/the-oxide-s...

[2] https://oxide-and-friends.transistor.fm/episodes/bringup-lab...

[3] https://oxide-and-friends.transistor.fm/episodes/more-tales-...

[4] https://oxide-and-friends.transistor.fm/episodes/another-lpc...

[5] https://oxide-and-friends.transistor.fm/episodes/the-pragmat...

[6] https://oxide-and-friends.transistor.fm/episodes/tales-from-...

[7] https://oxide-and-friends.transistor.fm/episodes/the-sidecar...

[8] https://github.com/oxidecomputer/omicron

[9] https://github.com/oxidecomputer/propolis

[10] https://github.com/oxidecomputer/hubris

This architecture leverages microVMs for rapid scaling and high-density workloads. But does it work for GPU? The answer is no. You can look at the old 2019 GitHub issue and the comments to it to get the bigger picture of why it is so.

Hi, I'm the CEO of the company that built this SDK.

We're a company called E2B [0]. We're building and open-source [1] secure environments for running untrusted AI-generated code and AI agents. We call these environments sandboxes and they are built on top of micro VM called Firecracker [2].

You can think of us as giving small cloud computers to LLMs.

We recently created a dedicated SDK for building custom code interpreters in Python or JS/TS. We saw this need after a lot of our users have been adding code execution capabilities to their AI apps with our core SDK [3]. These use cases were often centered around AI data analysis so code interpreter-like behavior made sense

The way our code interpret SDK works is by spawning an E2B sandbox with Jupyter Server. We then communicate with this Jupyter server through Jupyter Kernel messaging protocol [4].

We don't do any wrapping around LLM, any prompting, or any agent-like framework. We leave all of that on users. We're really just a boring code execution layer that sats at the bottom that we're building specifically for the future software that will be building another software. We work with any LLM. Here's how we added code interpreter to Claude [5].

Our long-term plan is to build an automated AWS for AI apps and agents.

Happy to answer any questions and hear feedback!

[0] https://e2b.dev/

[1] https://github.com/e2b-dev

[2] https://github.com/firecracker-microvm/firecracker

[3] https://e2b.dev/docs

[4] https://jupyter-client.readthedocs.io/en/latest/messaging.ht...

[5] https://github.com/e2b-dev/e2b-cookbook/blob/main/examples/c...

As far as I know, Fly uses Firecracker for their VMs. I've been following Firecracker for a while now (even using it in a project), and they don't support GPUs out of the box (and have no plan to support it [1]).

I'm curious to know how Fly figured their own GPU support with Firecracker. In the past they had some very detailed technical posts on how they achieved certain things, so I'm hoping we'll see one on their GPU support in the future!

[1]: https://github.com/firecracker-microvm/firecracker/issues/11...

I pass through a GPU and USB hub to a VM running on a machine in the garage. An optical video cable and network compatible USB extender brings the interface to a different room making it my primary “desktop” computer (and an outdated laptop as a backup device). Doesn’t get more silent and cool than this. Another VM on the garage machine gets a bunch of hard drives passed through to it.

That said, hardware passthrough/VFIO is likely out of the current realistic scope for this project. VM boot times can be optimized if you never look for hardware to initialize in the first place. Though they are still likely initializing a network interface of some sort.

“MicroVM” seems to be a term used when as much as possible is stripped from a VM, such as with https://github.com/firecracker-microvm/firecracker

According to their own FAQ it is indeed: https://github.com/firecracker-microvm/firecracker/blob/main...

What about microVMs like firecracker?

Dynamic memory management - Firecracker's RAM footprint starts low, but once a workload inside allocates RAM, Firecracker will never return it to the host system. After running several workloads inside, you end up with an idling VM that consumes 32 GB of RAM on the host, even though it doesn't need any of it.

Firecracker has a balloon device you can inflate (ie: acquire as much memory inside the VM as possible) and then deflate... returning the memory to the host.

https://github.com/firecracker-microvm/firecracker/blob/main...

Very few things in those companies are being written in Rust, and half of those projects chose Rust around ideological reasons rather than technical, with plenty of 'unsafe' thrown in for performance reasons

https://github.com/firecracker-microvm/firecracker/search?q=...

The fact that 'unsafe' even exists in Rust means it's no better than C with some macros.

Don't get me wrong, Rust has it's place, like all the other languages that came about for various reasons, but it's not going to gain wide adoption.

Future of programming consists of 2 languages - something like C that has a small instruction set for adopting to new hardware, and something that is very high level, higher than Python with LLM in the background. Everything in the middle is fodder.

https://github.com/firecracker-microvm/firecracker is the one that comes to mind, but most of these are internal.