Enjoyment is not in the act itself. Enjoyment comes from the appreciation of your fellow citizen who needed a thing moved that they were unable to move without help… And pizza.

I don’t find it uncomfortable. You’re there to do a job, you do it, you get your pizza and the appreciation of your fellow human, you go home.

You’re there for a purpose. People want you there for that purpose. You might not be good at it, but you’re there and that’s the important part. At the end of the day, the thing is done.

It’s a social contract, and it’s not exclusive to guys.

This is the way.

You engage in the economy of helping people to move stuff. One person has a dolly, another has a truck, they come, they move the thing, you pay in pizza and at least one IOU of “I’ll be there for you next time”.

When the time comes you are enlisted, you go, you help, you recieve pizza.

The cycle continues.

This is the way.

This is the only way.

Looks a lot like a straw man to me.

I don’t hate all LLMs/ML models/“AI”.

I also don’t use them. I also don’t see a need to use them.

I work in tech and these “AI” things are little more than averaging the most common replies to a given input. Any question that’s been asked, especially those that are asked often, is well suited for an LLM chat bot. That’s good if you want a high level overview of a topic that you don’t know much about. If you want accuracy, you’ll need to dig a bit more, but overall, you’ll get a relatively good idea of the answer you’re looking for. That’s enough for most people.

I’m crazy. I like to deeply understand everything, so the high level, lack of detail, finesse, and accuracy of LLM chat bots is unacceptable for me, so I trend towards doing my own research. If I use an LLM at all, it’s just an augmented search engine. I’m more interested in its references than its response.

There’s also something to be said about making short clips of impossible things that some smaller YouTubers are using “AI” models for. Things that would take a long time for a graphics artist to create, and an editor to composite. Stuff that a small time YouTuber can’t really afford, but they can get good enough but clearly “AI” visuals to spice things up in their videos. The AI is just for display during transitions, it doesn’t affect the content being made in any meaningful way. I’ve seen Paul Hibbert do this a lot and his production team from what I know is basically him and his wife. I’m not even sure if he has an editor, or if he does it himself. He’ll go from a live shot of him talking about a product, do an AI transition, to another live shot of him continuing his discussion. Rather than a cut, or some kind of fade, wipe, or other simplistic transition, the generated images are of him, and only to add visual flavor. He also does some AI generated music stuff, which again, fair. He’s not exactly a musician, and unless he has the money to hire a musician, he either uses AI for the funny transition music or he uses an unrelated and unlicensed (or open licensed) song… It just adds flavor to his content, it builds on what he’s creating, it doesn’t shape what he’s creating. All of his scripts sound genuine, not AI created, so everything should be fine.

That, to me, is a good use of AI as a tool.

There’s too many bad examples for me to really cover them in any detail, but I’m sure everyone has seen the AI generated person being animated by AI, reading an AI generated script, using an AI generated voice sample. Even if the original source material is legit, is buried under so many layers of slop that I can’t bear to watch it.

For me, it entirely depends on what he’s using it for. It’s entirely not about the fact that he uses it. I’ll have to dig into it a bit to see what he’s doing, but from the comments I’ve read so far, it seems like the answer to what he’s using it for, is yes. He’s using it for everything… Not ideal.

some AI coprocessors, you’re right. they can’t be used.

Most of what nVidia is offering, while lacking display outputs, can still be used for gaming/VDI/vGPU use, and in general aren’t particularly bad at it, usually quite the opposite. might take a bit of work to get them going, and nVidia hasn’t made it easy with their whole licensing thing. Putting aside the challenges, it’s usually possible to adapt their SXM datacenter GPU modules to work with more common frameworks like PCIe, and then utilize those as virtual GPUs in a virtual desktop. This is not dissimilar to what they’re doing with the geforce now platform.

The big benefit of the chips is that they come with loads of memory, so cutting them up as a vGPU is actually a really optimal use-case. Most games are memory limited, which is to say they need more memory than most GPUs have, so the limit in performance is from the memory limitations, not the graphics processing limitations.

The AI processors that are GPU based, like almost everything that nvidia is using, can mostly be repurposed, but those that are ML-only, like AI coprocessors that show up in CPUs like the AI optimized CPUs from AMD and Intel, can’t really be re-used, but they’re usually not released as a standalone product. Those may still be able to perform physics processing offloads… so maybe not useless? I’m not sure.

it’s called “Everybody’s Free”

Listen, I work IT at a managed service provider, which means I see a dozen + small to medium business environments daily.

The number of users that I know of, out of the hundreds or thousands that I support, who are actively using “AI” continually to do their jobs… Well, I can likely count them on one hand

There’s a bunch that use whatever free tool they are allowed to access. Which kind of flies under the radar for me, but I still don’t think you’re dealing with double-digit percentages of people using “AI” that actually spend any money on it.

And these are business users, who work at businesses with rather deep pockets at times…

People are not buying the product. “AI” in the business world, isn’t making any money. So from what I’ve seen, this is less a matter of if, and more a matter of when and how bad.

As a tech person, I am not threatened by AI. It can’t do my job. Also, I’m the one who fixes it when things go wrong. And the only thing I’m looking forward to is being able to pick up cheap RAM, and storage when this all comes crashing down as AI companies cash out and liquidate their inventory. Maybe I can pick up a cheap AI server and repurpose it to be a 3D/gaming virtual desktop, so when my friends come over with their cheap laptops and on board graphics for a LAN party, I can just provision them a virtual gaming computer and we can go ham…

I’m going to save my sheckles until then…

I have limited experience with the market but I see it this way: anyone with AI holdings should liquidate at their earliest convenience. I don’t think they’ll be keeping up this charade for a lot longer. Anyone who has tech holdings, either cash out, or be prepared to have that money held up for a while because it will take a long time for the prices to recover, and unless you want to cash out at a loss, you’re going to be holding for a long time. Everyone else, hold onto your butts. Probably everything will probably dip when the fecal matter hits the rotary air movement device. Non tech areas of the market should recover pretty quickly, so meh?

This is obvious to me that the whole thing has failed. That realization hasn’t hit execs yet.

Aches and pains! Enjoy!

Are we supposed to get a second time through our 20s? If so, I got ripped off man.

43 this year too. 42 was a wild ride and 43 isn’t shaping up to be any less crazy.

If I could offer only one piece of advice, sunscreen would be it. The long term benefits of sunscreen have been proven by scientists, where as the rest of my advice has no basis more reliable than my own meandering experience.

And no, there was no copying, nor pasting involved, I just remember that “song” very well. If you don’t know what I’m talking about, it may be worth looking up, because as I’ve gotten older, the advice in that “song” has only become more accurate, more true, and hit home a whole lot harder year after year.

So, in my case, I had a modem router model by a company called SmartRG. Details aside, I pretty much instantly put it into bridged mode so it wouldn’t participate in the IP routing. That modem did modem things only.

The connection went from provider phone line to SmartRG to my firewall.

I was weird and got a set of WAN IP addresses, so I put a router in front of everything to handle that, so the connection went from provider line, to modem, to router, to firewall.

It’s not super relevant, but the router I was using was a Cisco 1911. This is a semi modular enterprise router. The modular part, which will be important later, is in the form of “WIC” modules, or “WAN Interface Card” modules. The 1911 has two.

Anyways, I managed to get a WIC that supported VDSL2 with all the options and configuration that my ISP used. Happened to be the ehwic-va-dsl-m. Long story short, this module would integrate with my router and act as a modem of sorts to “translate” to the provider line. When I implemented this, I basically threw out my SmartRG. The phone line went directly into my router. So the connection was from the provider line, into my router, then to my firewall.

So the modem was “deleted”.

Another instance was for a fiber GPON line. The provider in this case, gave you a modem with a GPON connection, but they didn’t really tell anyone that the GPON interface was just a plain old SFP transceiver. So I pulled the SFP, put it into the firewall and threw out the modem. The provider line went right into their module in my firewall. The modem was effectively “deleted”

The idea of a modem delete is to remove whatever standalone device the provider has converting their signal (DSL, cable, or fiber) into Ethernet, and effectively plug that into your gateway.

It’s not always possible.

I’m currently looking for an option to do a modem delete for a local ISP that’s switched to xgs-pon. They put out a modem router for it that has the transceiver built in, so there’s no way to extract it and plug it into something else.

I’m hopeful I’ll find a SFP+ module like I found for the GPON ISP in my area.

I dunno if they still offer it, but I found that Cisco’s ICND1 was fairly neutral. They use examples from Cisco stuff, naturally, but the majority of the content is around learning and understanding how IP networks function. This is the first half of the CCNA study materials, and honestly, one of the best resources I had, and used, for learning how it all works.

There’s probably a ton more out there now, but at the time when I was learning, it was all CBT Nuggets and pluralsight… I believe a lot has hit YouTube in recent years.

Don’t worry if the information is out of date, this stuff doesn’t change. The updated stuff just has newer vendor specific information, and IPv6.

IPv6 isn’t crazy different in how it behaves, but the mechanisms for local discovery, IP assignment, and whatnot, can vary quite extensively.

Good luck out there

Sorry this took me a bit to get to. Hello!

I’m hoping that not all of that is running on a single pi. I mean, it can, but you might hit limitations when everything is engaged with doing things. I just feel like, that’s a lot for one raspberry Pi…

Anyways, iptables are good to have a general grasp of, but they’re generally GNU/Linux specific. There’s other routing implementations that run on Linux, and hardware appliances generally have their own bespoke, vendor specific stuff. One project I’m aware of is free range routing. There’s a lot more, but this is one that I know of. Using FRR, vs iptables, they’re very different beasts. But you shouldn’t need FRR, it’s a monster in terms of memory use and designed to operate in ISP class networks. You don’t need it. I’m just using it as an example of what is out there.

The best advice I can give about this is that learning the concepts behind routing is more valuable than any specific product. Knowing the difference between an RIB and FIB, and how to structure routes, priorities, costs, etc… All very important. Can you learn that with iptables? Sure, and probably more, since iptables can also function as a low end firewall.

The important thing is that you learn the meaning behind what you’re doing in whatever routing platform you are working with.

I’ve worked with so many different ways of handling routing and firewall work that I get annoyed when vendors come up with dumb marketing terms that leak into the device user interface, for a very common routing, firewall, or VPN technology. I don’t care whether I’m on a router or firewall that’s custom and running open WRT, ddwrt, opnsense, or one from Cisco, Sonic wall, watchguard, Fortinet, Palo Alto, or any of the dozens of other vendors. A VPN is a VPN. IKE and IPsec don’t change because it’s vendor x or y. Don’t start calling the IKE identifier something else.

… Sorry, rant.

Anyways, I don’t really see the vendor’s interface as anything more than a code I have to convert into the industry standard protocol information that everyone uses. It’s a filter by which that vendor portrays the same options that everything else has. Some have quirks. Some are more straight forward. But they all have the same options in the end. Allow the traffic or don’t, do it by port and protocol or by IP. Apply content filters or don’t, use Ethernet, DHCP, pppoe, or something else like ATM or ipx/SPX for signaling. Who cares.

If you understand the concepts, the skills are transferable, no matter what platform you end up using, you’ll know what needs to be done, you’ll just be stuck figuring out how you do it on this platform.

You are very welcome my friend.

An unmanned switch? Nothing concrete.

A managed switch can give you telemetry, like port utilisation, and you can observe how much upstream is in use.

My concern is that you have a 1g switch connecting 2.5g capable devices to a 2.5g capable upstream network. That’s a bottleneck that I would want to eliminate. I know serve the home has a roundup of 2.5g switches that might be useful for you. I’m not saying you should switch to managed either, you may be well served by an unmanaged switch, and it will save you money. The telemetry for managed switches usually requires a system to collect and store it, usually an NMS, or network monitoring/management system.

Some manufacturers build NMS style telemetry into their products, ubiquiti does this to a limited extent. Other vendors may be better or have nothing at all. Something to think about when picking gear, if you like that sort of visibility. NMS usually operates over SNMP, which can become a whole thing; but for monitoring, setting up read only SNMP can be rather easy.

A word of caution. 10G and 2.5/5G were developed independently, and 10G came first. It was expensive which is why 2.5/5g Ethernet became a thing. Because of this checkered past, there’s a lot of 10G equipment that will not support operating at 2.5 or 5gbps. So if you get a 10G switch, check if there’s 2.5G, or 5G capability separately, or included on the 10G ports.

In my experience, most 10G ports are 1 or 10G, with nothing in between. Most 2.5G ports can’t do 10G. So the best idea would be to have a switch with a couple of 10G for fast uplinks and some 2.5G connections for your devices. Unless you can find a unicorn of a switch that supports all speeds on all ports, a switch split between 2.5G and 10G ports is probably your best bet.

Good luck.

Well, I’m not opposed to it. If someone has an operational security issue that they want my take on, I’m happy to take that to DM.

The same promise, or lack thereof, for replies, will apply. Heh. I’m busy, so I can’t spend all my time on Lemmy. I love you all, but Shaka six feet dude.

That’s fair. I feel a little called out, rightly so.

What you want to do is look at the MoCA frequencies, which is ~450-1625mhz or so and see if your splitters include that range. If so, you’re golden, if not, there’s a good chance your MoCA signal will be attenuated by the splitter. Each splitter will cut the power by the number of connected branches, so 2 branches will be half power per branch, 4 will be a quarter. Etc. MoCA can handle some loss, but too much and it will fall over. There are splitters that you can buy that specifically include MoCA frequencies, some that don’t, and they’re will be some that specifically block it. The last type is good for separating MoCA segments, or at service entry points so you don’t end up sharing your network with neighbors.

Each splitter will have a label that specifies what frequencies it’s been tested with and that are validated to work. It should be printed on the splitter. If it’s not, throw it out and buy something that’s not in disrepair.

To clarify what’s going on a bit, the coax is just an antenna line, with no antenna. It can handle many different frequencies of transmissions. Like with the radio in your car, you can “tune into” different radio stations. The other radio stations on air don’t interfere with the one you’re listening to and vice versa. It’s the same idea with coax. Some frequencies are used to send cable TV, others are used for Internet (otherwise known as DOCSIS) and some are used for MoCA. All coax handling gear will support and be tested for some frequencies, and unless otherwise stated, anything outside of that range will be unknown. Most cable splitters support cable TV and DOCSIS frequencies primarily. There are different coax splitters for satellite, which uses all different frequencies, and there’s others that support much broader frequency ranges. Some can connect a wide spectrum but are only validated for a small part of what they can carry.

Your mileage may vary, and it’s really up to what you have and what the manufacturer did with the design of that specific splitter.

I’m sorry the answer isn’t more straight forward.

No argument here. The earlier stuff is still very bandwidth constrained, some pieces are incapable of meeting line rate, like the UDM pro. It has 10G connections but the throughput of the unit is around 6Gbps. Still much faster than most people’s Internet connections (who in their right mind has a 5+ Gbps Internet connection at their house?) but it’s a limitation worth knowing before you buy.

Solid gear otherwise. I haven’t seen how their end of life looks, so I can’t really comment, but most companies just announce that they’re no longer supporting a piece of gear and suggest a replacement. Called an EOL notice, or something similar. EOL being end of life. Usually includes a recommendation for an upgrade to something supported that’s a similar class of device.

Time will tell on that one. I have a UDM pro in a network I manage so I’m waiting for that EOL notice.

Yeah, that can certainly happen. It really depends on how the router handles contention, it’s overall capability and performance, and what loaf balancing algorithms, if any, have been implemented. QoS basically guarantees that some form of traffic management is happening. With it off, it’s really just a guessing game whether something will work well or fall over, as you’ve discovered.

Have a great day.

I see what you’re saying and this is a good inquiry. The reality is that most networks are what we call North/South traffic exclusive. In this context, we use “North” to describe towards the Internet, “South” to be from the Internet, and east/west to be LAN to LAN traffic.

Networks that are primarily or exclusively North South, your contention will always be your ISPs committed speed (the speed they’re allowing you to use). So most of what’s South of that is pretty trivial, as long as it can keep up with, or exceed the speed of the North connection.

That changes if you do any East/West traffic. Whether that’s a home lab, a home server, or even just a NAS, or computer to computer file sharing… Once that traffic is more than a trivial amount of the network traffic, then you see a lot of benefit from wired connections to your computers. The switch backplane can handle a lot more bandwidth than any individual port, and the only way you’ll see that bandwidth is if some traffic is going somewhere other than your router, or the Internet.

To say most home networks are North/South heavy is obvious. Business networks frequently have servers and other LAN resources that are frequently utilized. So East/West traffic is usually non-trivial.

To spin an example, if your ISP is providing a 100mbps committed rate, and you gave full gigabit ethernet inside and at least 802.11ac wireless, with almost all traffic going to the Internet and back, you’re going to see little difference between Wi-Fi and Ethernet. The only major change moving from Wi-Fi to Ethernet is that your ping time will be more consistent and lower overall. It won’t be a huge change, something in the range of 10s of ms, but it’s literally the only thing you’ll notice a difference with.

Another example where it will make a big difference is if you have a NAS or home server, where you have files stored. Compared to a file storage service like drop box or Google drive. The LAN specific traffic will move at line rate, or the speed of whatever storage the data ultimately rests on, whichever is slower. In that context, the East/West traffic benefits greatly from Ethernet, and the full duplex connection between the two devices.

It’s all subjective to how you are using your network. You’ve made a good point, so thanks for that. Have a good day.