nonplus

Same experience: https://i.imgur.com/Z9BrQGn.png

I don't hold the in browser POC at fault for the trained model (but am amused by the efficacy of the model).

Alternative android/android tv client with sponsor block baked in. https://github.com/polymorphicshade/NewPipe/

> what is an idiot-proof explanation for the need of docker in light of dlls?

I think the key thing is it's not a need. What you are doing is fine if your team can hire the talent they need to maintain (and grow) the product/software to match the companies needs.

OCI standards (what Docker uses) to some extent are trying to make a uniform ecosystem so engineers and architects know what to expect before digging into a project. That should not be compulsory though; other systems that work only make 'bigger' systems from places like the open container initiative better by showing what works in the wild.

I think an opportunity was missed to call this WarHAMing.

I've personally visited the X-10 reactor at Oak Ridge (built in the forties). So in the above, nuclear is not 'technology' until it's producing power for customers?

I'm sure the intent of their statement is salt/SMR/other exotic designs that the NRC has not greenlit (and, I think I'm ok NOT greenlighting designs that don't account for 50 - 80 years of maintenance/operation/risk)

I don't think technology and invention (or discovery for that matter) are synonymous. There has been progress in nuclear technology, even if it's so far non viable.

Re HN: I've had a hn account for 7 years, and I still can't downvote, and I think I like it this way. If I feel strongly enough I can start a dialogue instead.

Conversation threads on hn are also nice in my opinion, they are not compulsory, if you forget a discussion was happening the platform allows that. Instead of throwing it back in your face to encourage engagement.

I would love to see a 2022 follow-up from Agner Fog on this. He has work on C++ compilers as recently as 2021 so I'm sure he has recent real world info on the topic.

https://www.agner.org/optimize/#manuals

There are 2020 updates around MKL (But you may be correct that that content is about 2019 MKL optimizations).

At any rate though, based on Intel's track record I think this content is still relevant and of value to engineers who don't have domain knowledge in compilers or work downstream.

No contract in hand yet, I just recall loosely what it looked like from an offer a few years back, and what James Liu shared a few years back and want to be proactive this time around. I will see if I still have a copy I can reference (It may be vastly different, but it will be a good exercise). I'm not in California /'reasonable' non-competes have been upheld in my state.

Sure, and I have stayed away from FAANG (MAMAA now?) for 15 years because of that. The position/work is such that I am interested in what my options are beyond the "Why bother" stance that I have held until this point.

Edit: I appreciate a response regardless. If I get an overwhelming amount of apathy in these responses that might just be a sign I should change my mind about the position.

This point feels like arguing why any organization would seek density in computing if they can just buy more of something and spread it out. I don't know about you but I've saved a ton of effort on design complexity by not distributing workloads when it can be avoided (but distributed computing is a solved problem).

I recognize what you are calling out/that performance will be the same on some workloads if you distribute or not. I would just point out less manufacturing causes less e-waste/I would rather live in a world where Nvidia sells 50 million 10*0 cards, than 500 million 1030 cards to create the same amount of compute in the world. It's not just the power costs to consider (but it could be there is a reality where running 500 million 1030s for their lifetime wastes so much less power, that the manufacturing costs to the planet are worth it).

There was a pretty strong pivot in the first hour of this being posted from everyone saying "that's to much money!" to other concerns. I felt compelled to respond. I don't think the 1000 of these or so that are made are for everyone, and I have no doubt a v2 will be better.

I'm just ready to give my money to someone who wants to build on openXR, has a vr window manager in development, and the chops to build ambitious hardware (on Linux).

Yeah I get the concern. Following the project for the past 3 months one knows this information; but unless you're following development on discord/matrix or GitHub there's no way to know upgradability is a cornerstone of this project. Seeing 0 upgradability from Apple makes you leery of everyone.

Yeah the Samsung G9 was a silly amount of money on release, and I bought one.

I'm not on the team but yes, they will take lens inserts; there has been some discussion on this on discord.

https://discord.com/channels/603723949586644997/603723949586...

Not really, their break even (and by that, I mean ideal bulk pricing of components) for units sold is under 1000 units. The Simula one is not trying to sell 100k units. If you think thats the case, your expectations are wrong.

Re (software) development: Yes it will need funding, but it's there right now https://github.com/SimulaVR/Simula and is agnostic/works with any openXR headset.

I see a lot of comments on the pricing of this hardware. It's going to be more expensive (for the end user) to build a headset (and FOSS window manager) than a product like the quest. Meta has a suite of products bringing in capitol by selling user PII. Valve has a distribution platform to offset costs.

If you want to see a product that is open end to end and is not part of an ecosystem selling your information, then it has to be funded.

I realize most of the people reading this don't want a moral argument for spending $2k-$4k on a HMD, but thats what I am offering.

We should have multiple openXR compatible headsets to choose from, and specialized window managers that 'just work' in VR. Fund the FOSS projects you want to see in the world so others will follow.

Yeah I've found I care a lot more about the layout than the keys. I've learned I just want ANSI 60. It's minimal enough, the cognitive load is low/all keys that are not first class are one layout/fn press away.

@grahar64 You're doing a great job taking care of your son and navigating a scary reality. Thanks for taking the time to write this, it is now my #1 resource if anyone has questions.

Obligatory reply about linear (silent) cherry switches existing but mostly here to say I'm glad you found a keyboard you like. I think that's the takeaway I try to give people on input devices. Just get something you like and don't worry about what everyone else likes.

Maybe 5 products for clients/b2b. That's pretty much it for launched. Plenty of aborted saas platforms that I couldn't calculate an ROI/sales model that made sense.

Personally buying ip for a project you just started working on as someone expected to be a contributor? Yeah I'm serious (That I take that as a flag now). There were accounts and avenues to buy a domain when ready for that.

It showed they couldn't communicate well, couldn't work in stealth mode, couldn't focus on building/proving out the mvp.

Other factors are just hindsight/observing them the decade after. Two of those factors might be. 1. Repeatedly quitting contracts (if it's not a good fit fine, if it's never a good fit maybe you're missing skills to complete a project).

2. Comments/behaviors that illustrated they would have been a poor manager if we scaled/inability to be wrong.

For me, other developers and contributors have been one of the most important factors for success/completing projects/keeping clients (in the contracting space). So I consider this an important lesson I was fortunate to learn early. I don't know who I would be working with now if I had failed to keep great talent around me (or for that matter, what skills I would not have developed).

Sometime after 2009 I started working on a data mining app that allowed a user to enter medications and cross reference drug interactions. At the time opendata was one of the main information sources, now I think those datasets are managed by openfda.

I brought one person in to work on that with me, they proceeded to buy a domain for the top product name we were considering (canitake.com) within the first day. It was wonderful to get that red flag so soon, I stopped working with them, and abandoned the project. Losing the project was a little unfortunate, the data brought good questions someone might want to ask their doctor about; but knowing what I know now about "doing your own research" a lot of people would not have used this as a way to prepare to talk to a professional and would have interpreted the data themselves.

Not working with that acquaintance again was invaluable, they burned their way through the industry for the next decade. They could have messed up projects I actually cared about.

There is something ironic about looking down on the stack ml practitioners use while trying to bootstrap your public speaking career using ghost/a javascript based service. If you are going to name drop tech stacks you don't think are worth using that cuts both ways.

Damn; this is the most zen rock flattening simulator I have ever used.

Great, I guess I can add a supply chain attack to my list of concerns related to diabetes.

Good to hear I hope it arrives soon! One of the people I know ordered later, and one backed it very early and is still waiting.

That's good to hear, I don't know anyone who ordered the laptop so it's something to hear people online are receiving their products.

Of the two people I know who ordered one, 0 have been delivered.

I'm well into worried this is another eve laptop / failed to deliver a real product situation.

So even if they are not sold out, Purism is not getting any money from me until I hear of delivered products to people I know.

Played with TC for an hour or so this evening, and what I stated above (possibly poorly) still stands. I chose google maps in particular, because it is an application that requires telemetry data to function; but it is reasonable for an individual to not want to be tracked when not using google maps.

If I block infinitedata-pa.gogleapis.com, maps will not function, but google maps will continue to collect telemetry data on my phone if it is running and has permissions. It will save that collected data until a user unblocks essential monitoring in order to use maps (Unless the user clears cache/data, or uninstalls maps, before unblocking).

That is the case I am pointing out, tc is a stopgap (and a welcome/useful one) but it does _not_ provide users a way to prevent _collecting_ of telemetry data to be sent off the device. It just delays the sending until the applications use is more valuable than the users privacy.

Edit: Things that could help with that:

1. Physical kill switches for radios (I know, that's not going to happen from any major arm cpu maker, the SOC is integrated, but it's the most practical solution.).

2. Granular permissions settings for androids network location provider. As an example, A permission that if app is running in the background send spoofed location data back (Once again, it's not that simple telemetry data is coming from many sources, I'm just listing what solves the problem.).