[-] utopiah@lemmy.ml 4 points 1 month ago* (last edited 1 month ago)

True yet still not OK.

That's also why a lot of us do try to avoid, as much as is realistically feasible, to provide any data to any company that should store it. Hence why a lot of questions here are about self hosting, no cloud, etc. It's not paranoia, it's because companies cut corners and as you correctly point out, fail to keep us safe. So it's not about Tile specifically, they are just yet another poor example. Let's not defend them nor this kind of practices. If people in the Privacy community are OK with that, we have a rather deep problem.

[-] utopiah@lemmy.ml 4 points 1 month ago

Care to unfold a bit more what's hilarious? Which metrics from the article are wrong or irrelevant for example? You might disagree with the conclusion, and maybe rightly so, but are you saying the data itself, e.g number of companies funded is false? Or it does not matter and something else could help better understand the situation?

[-] utopiah@lemmy.ml 4 points 1 month ago* (last edited 1 month ago)

"Increased Spending on Equipment" is not evidence of progress. In fact there are numerous examples in the past of fraud, e.g https://www.ft.com/content/1e3fe107-1b6e-43dd-8f04-e3c88502c36b (cf "“Big Fund”, which raised $51bn in its last two funding rounds." 2 years ago, setup 10 years ago)

China is indeed pouring money on the problem and they are making significant progress. Yet it's not competitive in terms of performance and, much harder to evaluate, it seems not to be competitive in terms of economics. To make a processor a lot of low quality ones are discarded, leading to the idea of "yield" (cf https://en.wikipedia.org/wiki/Semiconductor_device_fabrication#Device_yield ). So, even if one CPU/GPU/TPU is genuinely produced, in "full autonomy" (so without e.g ASML unique machinery) and it actually on independent benchmarks comparable in terms of performance to the state of the art produced outside, it's still impossible to evaluate how viable the production process is. Maybe there yield is very high and thus producing those chips is efficient and thus cheap, maybe the chip used in the benchmark is the single existing one and thus is prohibitively expensive.

I recommend the 2022 Chip War https://en.wikipedia.org/wiki/Chip_War:_The_Fight_for_the_World%27s_Most_Critical_Technologyon the topic, it is quite interesting.

[-] utopiah@lemmy.ml 4 points 2 months ago

Yes if

  • the environment you would work on is Linux based, obviously (which it often is when servers are involved, even with Microsoft due to Azure cloud and containers)
  • you master the command line, i.e you know a bit of e.g bash, can write your own scripts that do basic functions
  • you understand how the OS works, i.e permissions, services, package managers, etc

but not really if you are mostly clicking through buttons of the window manager and/or would work in a Microsoft environment with its own set of tools, conventions, etc.

Which brings up obvious suggestions :

  • do improve your mastery of the command line
  • apply to jobs that put Linux forward (but that might bias to a sysadmin position, which might not be what you prefer)
[-] utopiah@lemmy.ml 4 points 3 months ago

Not sure I understand either but when I need to tinker with devices from another network through the Internet I usually rely on Tailscale or setup my own dedicated VPN using e.g OpenVPN.

[-] utopiah@lemmy.ml 4 points 5 months ago

I’ve broken my Nvidia driver 4x this week

Genuinely confused by that statement... been using an NVIDIA for years, both closed (to play and work) and open drivers (to test only) and beside having the "wrong" version for CUDA and some graphical bug in specific situation, e.g ALT-Tab out of game or resuming from a game leading to some minor visual glitches, I've never encountered even a reboot. I also have relatively recent drivers but I don't even know which version I have (checked out of curiosity : Driver Version: 525.147.05 CUDA Version: 12.0).

So... I don't get it, what leads you and others to such situation? Are you reverse engineering the drivers? Are you overclocking? Are you changing some specific parameters that are not stable?

I'm asking because this is so different from my experience that I don't get it.

[-] utopiah@lemmy.ml 4 points 6 months ago

Quite curious, I'm a VR gamer (and developer) and so far I've just had no problem with Proton and SteamVR, including for officially non supported games.

[-] utopiah@lemmy.ml 4 points 8 months ago

chromium being poisoned last year

Can you please expand on that? I don't use chromium except when I have compat tests to run but still curious.

[-] utopiah@lemmy.ml 4 points 11 months ago

thanks for sharing a screenshot of ncdu, should help others discover it

for the visualization itself IMHO Disk Usage Analyzer gives aesthetically pleasing results, not a fan of the UX but it works well enough to identify efficiently large files or directories

[-] utopiah@lemmy.ml 4 points 11 months ago

Well, as somebody working on this I can tell you it's up to you what you decide to use. Of course you can rely on Meta or Google but there are alternatives like Monocle or Lynx XR with software (not AGI... not even going to address that, assuming it's a joke) that is also open-source and self-hostable e.g GraphHopper for directions, Mistral or LLaMA for LLMs and countless others.

So... can you use self hosted AI with open-source hardware today? Yes.

Will most people decide to do so? Sadly not.

Are we trapped? Absolutely not, in fact I suggest you help on either or both side, open-hardware, open-source software and open models.

Please don't fuel fears.

[-] utopiah@lemmy.ml 4 points 1 year ago* (last edited 1 year ago)

It's for everyone. People who are tech oriented can dig deeper by implementing or modifying what I suggest but overall anybody can understand the problems, see that solutions are available and what a next step could be. I would say it's for people who want to do better with tech regardless of their current knowledge.

Edit: I give weekend workshops for 11-12 years old kid so I believe the material is rather accessible but always happy to hear suggestions to do better!

[-] utopiah@lemmy.ml 4 points 1 year ago

Downvoted because phrased as a technical solution. There might be a technical solution one day but until then, if it ever happens, it's a moral problem. By phrasing it otherwise we diminish the value and efforts of countless people, including RMS, who did invest their time in FLOSS for an ideal. Again it might happen but until then we must bet on what is right, not an idealized future that prompts idleness because it is genuinely dangerous.

view more: ‹ prev next ›

utopiah

joined 2 years ago