Coral Acceletor is only needed if you run setup that does not have GPU or enough CPU. Spare laptop usually has enough power to handle AI detection, but RasPi doesn’t. I run mine in CPU at rack server.
Cameras own detections are limited in my experience, and it is much harder to integrate to anything else, like HomeAssistant for notification & automation
HomeAssistant + Frigate combo is just plain awesome. You can leverage the automations of HA through Frigate’s AI detection, so you get things like notifications.
The best implementation of this that I have seen has been Frends Integration platform (https://frends.com). We use it at work, and have implemented ~500 business and integration processes with it.
This kind of visual programming works well in close to business logic implementations, were it is important to be able to quickly understand the business need behind the process.
How Frends work on the background, it compiles the visual process to C# code, so it is actually quite fast. Many solutions interpreters the process flow while executing, which makes them much slower.
My background is in normal code development, and I still think that there is a place for that. I have always hated having the raw business process in code, because those tend to change, and reimplementing code because process changes, is always really annoying.
Yes, just flip binary directly to the cpu
Oh jeez, rule34 is going to be full of lovenders nsfw now
edit: fucking hell, it had already from before, what’s wrong with you
They don’t even need to do triangulation nowadays, tower has direction and distance, because the way how 4G/5G work. 5G also has MIMO (meaning one burst of data to you carry data to other clients in same direction).
I assume all vpn services accessible from china are run by government and they monitor the traffic
Yeah, pretty much like that, in Azure and paid openai both let you modify the system prompt also. There is also a creativity (temperature) property that can be modified. When too high, it will hallucinate more, if too low, it will give same output everytime.
Retraining the model costs like hundred million and weeks of computing power.
Copilot runs with GPT4-turbo. It is not trained differently than openai’s GPT4-turbo, but it has different system prompts than openai, which tend to make it more easy to just quit discussion. I have never seen openai to say that I will stop this conversation, but copilot does it daily.
Addicted people with poor self-control are abused in these kind of microtransactions. This should be illegal.
Smart metering has been a thing in Nordics for past 20 years, and it has enabled per hour pricing that guides electricity usage away from peak hours (you have to build the network/production for the peak consumption, which is a problem).
This is pretty nonsense rambling.
What US of course needs is proper privacy laws. Electricity usage is protected by GDPR in EU, so it would be illegal for anyone else than you and your grid operator to see your consumption.
On-device AI is not bad. If you can do tricks to actually improve things with AI, it is a tool like any other.
Main problem with game like this, is that you are probably not going to have it running more than 6 months with heavy loads, after that you can scale everything down.
If you have a business that is going to run 10-20 years, you can build complex solutions to optimize the cost.
In this kind of rocket like need of global computing power, the cloud is only real solution.
No one in their right mind would deploy own servers for this kind of load. It fluctuate way too much and in half a year you have unused servers that are junk. Initial purchase price would be millions, and setup would take months.
They are definately running in some cloud, and 500k/month is about what you would expect to host servers for a popular game like this in close to launch.
It is quite long game, and only second game ever to break 2M concurrent players on steam
I would suggest more learn by doing approach. Learning OSI model etc is nice, but it is quite jargon :)
Use some old PC as a server, and get some network cards into it, and use it as firewall/router. Route your home network/NAT/DNS/DCHP through it. Raspberry Pi’s are nice, but their hw is still bit limited.
OPNSense is quite nice and easy free and open source firewall/router solution.
If you want to add bit of flexibility, you can use some virtualization platform like VMware in to the machine, so that you can run OPNSense in it, with some other virtual servers.
Then when you get things working, you can start looking in to VLAN’s, because they are quite important part of enterprise networking. Most cheap switches nowadays support VLAN’s out of the box.
This might change now relatively fast, now that large language models can process code, you could give the function to LLM to rename the function. Iterating over the code and rename all functions and variables.
This won’t of course reproduce exact code, but it makes one really heavy part of reconstruction to human readable much lighter.
Try dungeon AI, it has been out for a while. Uses LLM to generate the adventure:
Same as any library/framework upgrade in code: if you don’t upgrade, you will not eventually get any new features/security updates/asset etc. If you have stopped development, then no point, but if you plan for support game for longer period, it makes sense. Also in the future if UE 7 brings something awesome, upgrade from UE5->UE7 is much harder than two simpler version upgrades.
And ofc this thing is clear scam, so this was just in general
You can add IPS to port to add some security checking, but yes, in general port is never secure or unsecure.