I’m sure Google does monetize the gps data instantly, then throws it away rather than save old data that only costs money to respond to government requests.

This is a case where privacy is economically beneficial to Google.

The device only was for privacy. When the data was stored in the cloud, the government had unrestricted access. By making it device only they need to get your device to get that data.

Btw the CPU in the Lenovo P330 is an e-2174g. I also got an e-2274g.

I got a sff P330 Xeon with integrated graphics for ~$500 two years ago that includes case power supply etc. Far faster than an n100 and even lower power than if you added a GPU to an n100.

I just plugged in a kilowatt to check:

My Lenovo sff workstation running Plex idles at 15 watts- which is 90% of the time. Streaming 4k 52Mbs hevc (This Flash Gordon is my torture test that caused me to upgrade 2 years ago) it’s 18 watts! I was so surprised that I went back and unplugged the Ethernet thinking I put the killawatt on the wrong server.

What’s your budget? I’m a big fan of old Xeon servers.

You have to look at the datasheet. It says 6 minutes at full load and 18 minutes at half. Full data can’t be listed because it’s a non linear curve. The website gives you a calculator at the bottom.

https://www.apc.com/us/en/product/SMC1500-2UC/apc-smartups-c-line-interactive-1440va-rackmount-2u-120v-6x-nema-515r-outlets-smartconnect-port-usb-and-serial-communication-avrgraphic-lcd/

You pay a huge premium for rack mount. You could buy 2-3 regular UPS for the price of one rack mount. I bought a shelf for my rack and have regular UPSs on the shelf. Note that you can only chain 1 ups into another without problems. So it’s best to separate devices onto different UPSs or buy one that supports plugging external batteries.

Watts is peak usage not watt hours. You can have two 1500 watt UPSs but one could last far longer because it has extra batteries but the same 1500 watt inverter as another model.

Yeah I want to completely switch off of Plex but neither is a good solution for my non tech family members. Mother in law is in a retirement center where they use wifi provided for the condos so I can’t access her router. And I would expect her ip to occasionally change on reboots etc. I might try IP ranges or narrow geo blocking.

I thought there was some way to use Jelly on the backend with a Plex client!

Jellyfin over Plex?

How does that work with Roku/smart TVs?

I didn’t see tokens/s anywhere?

So you are leaving your laptop behind as the server?

What’s your budget? I’d build a separate server but that’s me. I would think you could get old hardware practically free that would run your services fine by checking Craigslist and your local recycling center.

With the current Minecraft monthly updates, paper is always behind on the latest features. There are also minor problems that paper introduces with its performance improvements.

Years ago paper was critical for a good Minecraft experience, but a newer PC (newer than 6 years old) runs great on vanilla.

So if you’re planning to have the kid play on Switch or something like that, it’s not going to work.

You can run Geyser (a modified Minecraft server) to let bedrock clients play on your Java server.

I have a lifetime Plex pass but am still annoyed at having to deal with “recommended” every time a device is setup or reset.

The recommended view is useless and there is no way to make library the default view. You have to reset every source. It makes it incredibly annoying helping my family remotely to get to family videos.

No, full models are not loaded into each GPU to improve the tokens per second.

The full Gpt 3 needs around 640GB of vram to store the weights. There is no single GPU (ai processor like a100) with 640 GB of vram. The model is split across multiple gpus (AI processers).

You can. But I don’t think it will help because the igpu is so slow.

https://medium.com/@mayvic/llm-multi-gpu-batch-inference-with-accelerate-edadbef3e239

More gpus do improve performance:

https://medium.com/@geronimo7/llms-multi-gpu-inference-with-accelerate-5a8333e4c5db

All large AI systems are built of multiple “gpus” (AI processers like Blackwell ). Really large AI models are run on a cluster of individual servers connected by 800 GB/s network interfaces.

However igpus are so slow that it wouldn’t offer significant performance improvement.

I love the series but I assume there’s some mistranslation of culture in the idioms of some of the characters. (I watch dubs) Like Senku’s constant “I don’t care a billionth” and “you get 1 billion points.” and Gen’s constant pig latin. Those idioms are cute when used sparingly but it seems like it’s every other sentence.