Oh let's be charitable! A parameter measured in ms is mistakenly thought be measured in s. Hilarity ensues.
Real world example: þe Windows registry DWORD time periods seems to invite 10^-3s granularity for totally inappropriate timescales. Perhaps its considered a "best practice" by the dick heads that decide to do these things, who knows? Why bother considering how a sysadmin might actually want to use the knobs and dials and what is an appropriate value for a parameter.
I could probably find a better example but this is recent: Smoothwall has an agent (IDEX) that you install on a Windows domain controller and one of its functions can be to harvest DHCP data and pass it onto the firewall so that it can track sessions. The upload period is a registry DWORD value.
I fixed a "problem" by stopping IDEX trying to upload data a thousand times per second. I will also point out that switching on this functionality and the periodicity setting is only applied by editing the registry - there is no GUI for this. The dReal world example -ocs are clear that you should initially set 1000 as the period.
For me that sort of thing comes under the heading of "you are holding it wrong", potential victim shaming and rubbish engineering.
This is what I suspect. A retry every X milliseconds actually being used against a variable nanoseconds, a broken loop condition that ends up always retrying, something like that.
Not making an accusation, but it reminds me of a chat transcript oddity I saw while using opencode back in August. It replace a syllable of a word with a punctuation mark that has the same pronunciation or HTML entity. It was something like "...the &litude of that waveform..." but I can't remember exactly, nor which model I was using. I just noticed it and thought, that's weird. A few days later I read about text watermarking and figured it may be that.
Happens on paid too. Every once in a while, one of my conversation names will have random Chinese characters in them. I guess distillation goes both ways.
It's impossible to tell the difference between being malicious and being ignorant.
With this particular company, everything else they do is malicious so I won't ever give them the benefit of the doubt.
I tried to use a reusable pod in one of their machines the other day and when I shut it the handle broke off leaving me rather confused. Turns out in the closing head of the machine they stuck in 4 big metal spikes to destroy anything put in there. There is absolutely no reason to do this, none, other than being dicks. Had to get out the epoxy and repair the handle of a friends machine.
Exactly. The whole thing is weird and creepy and wrong.
I understand why a TV would keep track of what I’m watching so they can sell the data. I think it should be illegal. It’s horrible. My TV isn’t connected. But the reason they would do it fits in my brain. I can see how they got there.
How a coffee machine got to running network probes… nothing. It seems like some sort of Internet of Things DEFCON presentation topic made up by putting random words together.
So to think that on top of that they were purposefully causing so much traffic on the local network is just a few steps too far for me to think that part was intentional.
The reason is that probably the coffee machine vendor doesn't know what is going on. They probably got that module from.some third party, like "hey, you want to cheapen your smart coffee machine? - take out network stack!" and then they don't have to hire developers doing all the work to implement a network configuration system and all that stuff, but have a simple API for the App telling you coffee is ready
I think the reason to use a coffee machine is because it's an innocent device that doesn't require any data whatsoever, so nobody would ever suspect it of something so nefarious.
They're explicitly taking advantage of their customers' trust, and deserve to go bankrupt.
The funny part is that the manufactures shitty QA made this a much bigger thing that it could have been. Had the machine done a scan once every week, which is more than enough for their purpose, diffed the result locally and pushed the few kB of data to the internet, then no one would have noticed anything.
Yes, it's malicious and completely unnecessary, but incompetence has potentially made it a PR problem.
I was talking to the guy who used to run our ITSM system at work. He said a team was trying to query the system 10x per second to check for updates, which was causing performance issues. Assuming it was a bug, he went to the team to have them tone it down. He was shocked when they fought with him over it. It was working as designed and they didn't want to check less frequently, despite there being no logical reason to do this.
This may be working exactly as designed, as it costs them effectively nothing to constantly scan.
I don't understand giving the benefit of the doubt to a company that is actively spying on its customers, which in some jurisdictions would be illegal.
All these companies make their money as 'feeders' to the advertising industry, they just sell a different device to consumers in order to achieve it.
Selling devices to consumers is a solved problem. The problem we're currently trying optimise solutions for is selling consumers to the advertising companies.
Maybe they literally just don't care about how much traffic they put on your network. Maybe the thought is "1 tb of traffic internally is very low utilization of the network over the course of a month".
Whatever they're trying to accomplish with these scans would probably be impeded in some way by this much volume. Like it'd compete with the link it's using to phone home or overload whatever on-device processing it does with that data. The man in the article who discovered this called it a bug, I agree with him.
I'm not saying there's no malice, it's just that maliciously spamming the network would get in the way of the malicious spying they actually want to do.
When the cost of the action to you is 0, why not do it as fast and as frequently as possible until you get your intended benefit.
Even the act of engineering rate limiting costs you more than just having this run wild over your customers networks because the vast majority of people buying these machines do not have the inclination or skills to detect this activity.
I just finished watching the race. The start was certainly a callback to the start of the season when cars were dropping like flies and having constant odd problems.
From a hacker's perspective, yeah the feeling of accomplishment in figuring out something manually is great. But from the end user's perspective, if I get to choose between "shit ain't working" vs "it's working flawlessly but AI was involved in getting there", I'm picking the latter every single time.
It's not an unsympathetic position. macOS' current direction is controversial; many people want an alternative. They see Linux less as a hacking curiosity and more of an escape hatch.
Even with years of excellent, tireless reverse-engineering, Asahi's power management for M1 is not on-par with macOS, let alone what AMD/Intel chipsets get on Linux day-1. It's just inevitable that the hacking efforts will fragment into a fast-and-loose vs slow-and-steady mentality. And both of them will be equally right, since there is no official support for any of this hardware.
I get why someone may use AI to try to speed things up. Let’s put aside if that would even be useful at that point on the M4
I don’t understand complaining someone who spent their own time doing something isn’t using AI instead.
Is it all that different from asking why someone painted something when AI could generate an image of a painting for you?
It probably wasn’t intended as such but the sentiment that comes across to me is “why did you bother?”
There are certainly people who would always be anti-AI. But there are a lot of people like me who continually bristle at how an AI booster seems to show up in every discussion implicitly implying you’re dumb for wasting your time by not using it.
You know how people got really sick really fast on HN of people asking “why not Rust” on any discussion vaguely related to programming no matter how little it made sense? Or “A blockchain would fix that”?
> Is it all that different from asking why someone painted something when AI could generate an image of a painting for you?
I don't agree. In this case it's asking AI to analyze the high-res photos of Van Gogh to learn what paint he used, what techniques, how he held his pencil and how did his strokes look like.
Sure, you could then also ask to paint one for you in similar style, but that's not what is being suggested here.
Oh, I’m sure they exist. There’s Christian versions of any genre you can think of. That’s partly why I ended up adding in the 8-track part, to make it more ridiculous.
Maybe I should have made it Mongolian gangster rap.
Nuvia started because a few Apple engineers wanted to make server CPUs and sell them. Apple management didn’t want to do it so they left and started their own company. They were acquired by Qualcomm.
There are talks of John Ternus potentially selling Apple Silicon for server and AI use.
Pretty much soon. Nuvia's design has been incorporated into 8 Elite and X Elite/Plus series. It was generational leaps ahead of 8 Gen 3, where 8 Elite is already close to Apple M1.
I'm looking for a cheap 8 Elite phone to root it and run a special kernel module that let me run LXC containers and run arbitrary Linux ARM64 program. And run a couple of services on it, maybe I can port Kubernetes whatsoever
> here are talks of John Ternus potentially selling Apple Silicon for server and AI use.
At the very least they ought to consider doing a macOS-based server chip. Given how many providers are racking Mac Minis these days. It's mostly for CI testing for code deployed to Apple platforms, so it very much supports their business.
The Server market is enormous with a much higher average selling price. It would make perfect sense for Apple to make servers if it meant more revenue and profit than they could make not doing it. There are rumors that Apple is planning a M8 based server in 2029.
Apple tried to address the server market 20 years ago during the PowerPC days (and later Intel). Their Xserves were really well
designed machines and worked well. And yes, they also ran Linux (Yellowdog, IIRC).
It failed.
But I think that they learned something really valuable. As a company, they don’t have a good understanding the server market or how to sell to it. Their premium product style just didn’t have a place in the race to the bottom server market.
We had two in my grad school lab that we used as web servers. But that was the problem - it was a premium designed product that wasn’t any better of a web server than a plain (cheaper) Dell 1U box.
Maybe things would be different now that they have a really unique chip. ARM as a whole ecosystem has more buy-in from other companies than PowerPC did. Their performance per watt and GPU design might make more of a difference this time around.
But I can’t see Apple being able to release and sell a data center centric server. I think others might have a better chance.
A lot of Apple's advantages come from tightly-linked memory and compute in a fully integrated design. I somewhat doubt that advantage can fully translate to servers, where features like hot-swappable memory are standard. Maybe they should skip mid-scale and head straight to mainframes?
Hot-swappable memory is probably much more common in mainframes than mainstream servers. A typical x86 server still requires you to power it off before replacing DIMMs. Meanwhile, NVIDIA and AMD have both recently done some rather Apple-like server processors that solder CPU, GPU, and RAM to the same package/module.
If Apple cared about the server market, they would have added syncookies into their TCP stack. Seems like they should have done it during the time of Xserve... it would have been somewhat straightforward to pull it in from upstream BSD back then. But they still haven't (last I checked).
Apple doesn't have the business to business support organization to make this kind of thing work either.
Libdispatch is a burden to use for a server because of its book keeping overhead, and their history with kqueue makes it hard to convince someone to use it.
epoll isn’t very fast, kqueue is indeed much faster as a primitive.
MacOS’s implementation is an order of magnitude or more slower than (say) FreeBSD though, so its not a valid comparison that the syscall simply exists.
Another example of a syscall performing differently would be gettimeofday() in OpenBSD vs linux, its an extremely cheap operation in Linux but quite costly in OpenBSD despite being identical at the kernel boundary.
Sure, they can, but they've done servers several times in their past and I think it's reasonable to guess that results will be similar. Sure, they have a very nice CPU, but they don't have a good reputation for enterprise support and that's more important to selling servers than having the best cpus. If their cpus were 10x as good as the competition, maybe, but they're not, especially in a server context where power and cooling are abundant.
The rumors are that it’s for internal use. It’s already known that they have custom M2 or M3 servers that they don’t sell to the public. Source: multiple episodes of the Accidental Tech Podcast that discuss articles about this.
> There are rumors that Apple is planning a M8 based server in 2029.
The issue is that the current Apple way of designing hardware doesn't really work for servers. An individual 1U server can easily cost $50,000 or more. Soldering everything together so that if a single memory chip goes bad you have to replace the entire machine instead of that one stick is going to make customers unhappy. Likewise, they expect to be able to add storage or network ports, and expect the software to have enough openness that the new systems will work with their existing enterprise SAN or what have you.
Obviously Apple could make open server hardware, but then people might start wondering why they can't make open laptop hardware.
They could use sockets for the memory. Easily justifiable because there is much more available space in a server than their other products. Just look at how bare the Apple Silicon Mac Pro mainboard looks.
The problem was never space. They solder the RAM even in the laptops that have 8GB and the same memory bandwidth as ordinary PC laptops with modular RAM, and DIMMs take up less space on the board because the memory chips can then be attached double-sided to a DIMM that only needs as much space on the board as it takes for an edge connector.
The primary advantage of soldering compared to DIMMs is that the traces can be slightly shorter without routing them through an edge connector, but if you need that then you can use CAMM.
They solder the memory because then you can't upgrade it. Strong evidence for this is that they do the same thing with the SSD, for which there is no even plausible other justification.
The main reason is definitely just overcharging people for upgrades but size also includes thickness. SOCAMM would add at least the thickness of an extra PCB. I don't really agree with Apple's obsession to make their devices as thin as possible (they're thin enough!) but it would be a valid concern for them.
> Strong evidence for this is that they do the same thing with the SSD
Worse, they have a proprietary module that they use in the Mac Mini. Couldn't have just used M.2.
That still doesn’t overcome the issue that they are selling every single one they can already. They have no need for additional chip sales.
All the capacity at all the fabs is booked up for years. They can’t make more if they want to.
So any hypothetical sales of OEM style guts would only cut into the number of Macs they can sell to existing customers. And there’s no way they would ever do that.
Servers need a lot of RAM. Apple can either make 42 iPhones with 12GB or one server with 512GB RAM.
42 people with iPhone will buy airpods, some subscriptions and apps. They also be vendor locked into Apple ecosystem and almost guaranteed to bring more iPhone sale in 3 years.
Apple must be dumb to canibalize their own market.
They tried it with various open systems that allowed licensed powerpc clones of their systems and they ended it because it reduced their profits as people bought the clones instead of mac hardware. I don't see them redoing that anytime soon.
Still seems buggy.
reply