rb2k_

There's differences. e.g. MOSFET vs TRIAC

Zooz (popular zwave manufacturer) has some interesting tidbits on their homepage.

https://www.support.getzooz.com/kb/article/1103-zen72-vs-zen...

"The ZEN77 Dimmer is recommended for 3-way and 4-way installations since you won't need to rewire your other switches in the set-up, you can simply replace the main switch with direct connection to power with the ZEN77 dimmer.

This model can control up to 100 Watts of LED bulbs but we don't recommend using it in installations with chandeliers or large groups of lights over 6 bulbs. Version 1.0 and 2.0 of the ZEN77 (700 series Z-Wave chip) were MOSFET dimmers so if your bulbs work better with trailing-edge (or reverse-phase) drivers, those versions of the model worked best. Version 3.0 of the ZEN77 (700 series Z-Wave chip) is now a TRIAC dimmer so if your bulbs work better with leading-edge drivers, this model will work better. The 800 series version of the ZEN77 is also a TRIAC dimmer. Why We Changed to all TRIAC: We found that newer LED bulbs dimmed better with TRIAC dimmers, and considering limited availability for MOSFETS, we decided to transfer the ZEN77 model to TRIAC as well."

The readme mentioned H20 GPUs. Nvidia's "China compatible" card (41% Fewer Cores & 28% Lower Performance Versus Top Hopper H100 Config)

Some Llama3 400B numbers for the April 15th checkpoint are listed at https://ai.meta.com/blog/meta-llama-3/

Direct image link: https://scontent-atl3-2.xx.fbcdn.net/v/t39.2365-6/439015366_...)

In the past (with h265 / h264 at least), hardware encoding always ended up with visibly worse quality (and often even bigger file sizes) compared to a software encoder like x264/x265.

Do you happen to know if that's still the case?

(I guess for use-cases such as live streaming it doesn't matter that much, but for video that ends up in some archive, it's probably less acceptable)

Last I talked to people about Apple's internal tooling, it was still pretty fragmented between different organizations.

I was told about e.g. multiple internal CI/build systems because one org didn't want to rely on the other, so they both built their own little software stacks.

That's a bit different than what I've heard from meta/google/amazon who seem to be more open to running centralized internal services

Did you base your router config on some existing template or did you just built it up from scratch?

There should be a difference between applying critical thinking and being borderline hostile though.

I think one of the problems with large parts of academia is that the system thrives by finding flaws within things other people have created.

Criticism is certainly helpful and necessary for advancing the state of the art, but without something to balance it, it turns into a pretty bleak place to work. I guess this is one example that feels more like a personal vendetta than a constructive criticism of the work.

Nah, a dedicated "code review team" is not a thing.

Often the suggestion featured does suggest a whole team as a review based on code ownership.

> Some companies hire dedicated tech people whose only job is to be oncall, handle alerts, and improve the oncall infrastructure. This role is called ‘DevOps Engineer’ at some companies, SRE (Site Reliability Engineer) at others, and may also be called ‘Operations Engineer.’

Any company that I've seen put SRE/"DevOps"/... as the sole primary on-call rotation basically just created a glorified operations team.

Unless you have shared pain for botched releases, you will never get rid of these problems.

One thing that this article is also kinda missing is the base compensation. Some companies (e.g. FB) are in the top 90th percentile or so of the industry but then don't pay on-call compensation..

But that means that in one case you make 200k base, have an on-call and don't get any extra on-call money. In the other case you make 140k base, have an on-call and get 10k extra on-call money.

Ultimately you end up doing the same work but one of them gets paid less.

The website mentions:

"Funkstandards / Frequenzen EnOcean (868 MHz), Low-Power-Funk (2.4 GHz) WLAN (802.11 b/g/n 2,4 GHz)"

So that's neither of the two major protocols (zwave and zigbee) nor any of the up and coming ones (thread).

Am I missing something?

It's 7 years old by now, but there's some literature:

https://research.facebook.com/publications/holistic-configur...

You can see that there's a common backend ("configerator") that a lot of other systems ("sitevars", "gatekeeper", ...) build on top of.

Just imagine that these systems have been further developed over the last decade :)

In general, there's 'configuration change at runtime' systems that the deployed code usually has access to and that can switch things on and off in very short time (or slowly roll it out). Most of these are coupled with a variety of health checks.

Interesting. That almost sounds like mania/hypomania.

(Not saying that it induced a hypomanic episode, just that all of those fit the description)

> A hundered thousand years of safe storage is not part of the calculation of the energy price

People always mention that as an argument. I'm pretty sure that we could figure out a solution if we actually worked on it. Given how far we've come over the last 100 years, I don't see this as a problem that we couldn't solve over the next 100 years.

> without the government subsidizing child care costs

Why not though?

They'll be tax payers soon and I'm sure we'd get more of them if we made it easier to rear them.

There's probably a lot of subsidies that are currently in place that make less sense.

> In the USA, it is common for unions to be industry-wide

A lot of the ones in Germany are as well. https://en.wikipedia.org/wiki/List_of_trade_unions_in_German...

my two cents as someone that worked in Europe and the US:

> time off

I get 21 days in the US, but any holiday that falls on a weekend I am guaranteed to get on a weekday.

Ultimately it ends up somewhere around 30 paid days off I would guess. Back in Germany, if a holiday fell on a weekend, I just lost it.

> government paid health care

In Germany I had to pay 7.5 percent of my salary + my employer had to match that. When I was freelancing I had to do the full 15%.

The US clearly has the worse healthcare system overall, but I pay a LOT less for MUCH better service. That is however just because of my employer and on a population level it's a catastrophe.

That being said, my impression is also very skewed by FANMAG and the average experience is probably worse.

A lot of these providers are part of the Open Compute project [0] that Facebook created in 2011 and all the hardware is open source. At least for FB, all the new hardware is part of that.

[0] https://www.opencompute.org/

Agreed. I've seen thousands of mac minis in CI farms writing 1+ TB / day (all day, every day) and they're slowly starting to die 4 years in. It's hard to write flash to death with that kind of write load. I wouldn't worry about that.

It has 4 Cortex-A7 cores with up to 1.2GHz. Given the kind of stuff e.g. OpenWRT usually runs on, I would say it has more than enough power.

For simple workloads (e.g. serving static files via HTTP) I'm pretty sure it could saturate a gigabit connection.

You can even find pre-made travel routers with the Allwinner H3: https://www.amazon.com/Portable-Allwinner-Quad-core-high-Per...

This one has a gigabit port, Orange Pi Zero is just 100 MBit/s (but adds wifi). The Orange Pi also seems to be a bit more expensive (close though).

This is sadly 'just' homekit, but I was pretty impressed with the polish of the solution:

https://hochgatterer.me/hkcam/

I agree on the comment for rtsp though. Most solutions have multiple seconds of latency which is a bit suboptional.

Another set of mirrors with decent network infrastructure can be found at: http://mirror.facebook.net/

(Not holding my breath for the kind of responses this comment might get ;))

Don't a lot of the bigger schools have billions of dollars in endowments that can easily cover fixed costs with their returns?

Same in California. For people that want the details:

https://california.public.law/codes/ca_bus_and_prof_code_sec....

California Business & Professions Code section 16600 makes clear that any non-compete provision between employer and employee will not be enforceable under California law

From my (German moved to the US) perspective:

1. Stop putting college kids with a bachelor degree in business in charge of software development teams. "Management" is still seen as something very 'fancy' in Germany. Being "a manager" carries a lot of weight.

2. Allow Engineers to advance without forcing them into management. A lot of very talented engineers in Germany are forced in mediocre middle management / product owner positions as the only option to gain some more influence.

3. Pay

Yup, moved from Germany to the US. Would never be able to earn in Germany what I earn here without going into management.

Most PhDs I know that are doing research in labs are on a F1 or J-1.

We've had no issues getting same Whole Foods deliveries via Amazon in Menlo Park. We just had to refresh a few times over the course of 15 minutes or so and always got a slot.