

What sort of wear was your LED light getting? I’d not think that the wires would have any sort of flex or give in them on a well designed lamp. Sort of like my house wiring, it’s not like it’s flexing around all that much.


What sort of wear was your LED light getting? I’d not think that the wires would have any sort of flex or give in them on a well designed lamp. Sort of like my house wiring, it’s not like it’s flexing around all that much.


One of the more fucked up aspects of eminent domain. City/county/state governments can nuke deeds by using eminent domain. It allows them to turn a plot of land, regardless prior restrictions, into things like dumps.
The fucked up part about it is they can also turn public lands into private lands with that same trick.
What’s frustrating is we still need eminent domain for good. It’s basically the only way to build railways and roads. It even ends up being one of the few ways to deploy things like district heating/cooling and new fiber lines.


Given CXMT doesn’t just up their prices after gaining market share too.
They might eventually, but certainly not immediately. In the process they are going to force the other big players to lower their prices to compete.
Also China can just ban the export, or tariff it.
I don’t think China has ever banned an export. It’s pretty rare to tariff an export, basically only happens when it’s a limited good that the government wants to ensure a local supply of. I think the only country I’ve heard of doing that is Greenland due to some british wankery.


Maybe. I mean at least a major part of it is that AI pays a lot more money than the consumer does, and even if the consumer market pulls back, they are banking on the raised prices to stay around even if they can increase capacity.
I think what scares them is that CXMT is rapidly catching up to the state of the art. If they dick around for too long, they run the real risk that China and CXMT will do what China does and sweep the market with really cheap memory they can’t compete with.
These companies still care about non-ai servers, and that’s a big part of the market that could be obliterate pretty quickly.


Hey, can we stop calling everything with a computer “AI”? Order management systems have been a thing long before LLMs were invented (I’ve worked on one). This was perhaps one of the first applications of computing. Humans hand writing an order form in a major grocery store hasn’t been a thing since like the 80s.
Also, I’m like 80% sure this article was barfed out by an LLM. The em-dashes be everywhere.


Regardless, I’d bet anything that you are not able to reencode blurays to 15mbps without substantial quality loss.
It’s crazy how confident you are about something you’ve never done.
Yes, I’ve gotten transparent encodes with 15mbps. Lots of people have.
If you don’t believe me, go to the AV1 subreddit and ask “can you get a transparent encode with only 15mbps”. Or just go read up on svt-av1 and av1an and do it yourself.


Look, this is just an incorrect oversimplification of the problem. It’s popular on the internet but it’s just factually incorrect.
Here’s a thread discussing the exact problem I’m describing
https://www.reddit.com/r/AV1/comments/1co9sgx/av1_in_dark_scenes/
The issue at play for streaming services is they have a general pipeline for encoding. I mean, it could be described as cheaping out because they don’t have enough QA spot checking and special purposing encodes to make sure the quality isn’t trash. But it’s really not strictly a “not enough bits” problem.


I promise streaming services and CDNs employ world-class experts in encoding
They don’t really care about the quality
It’s funny that you are trying to make both these points at the same time.
You don’t hire world class experts if you don’t care about quality.
I have a hobby of doing re-encoding blurays to lower bitrates. And one thing that’s pretty obvious is the world class experts who wrote the encoders in the first place have them overly tuned to omit data from dark areas of a scene to avoid wasting bits in that location. This is true of H265, VP9, and AV1. You have to specifically tune those encoders to push the encoder to spend more of it’s bits on the dark area or you have to up the bitrate to absurd levels.
Where these encoders spend the bitrate in dark scenes is on any areas of light within the scene. That works great if you are looking at something like a tree with a lot of dark patches, but it really messes with a single light person with darkness everywhere. It just so happens that it’s really easy to dump 2mbps on a torch in a hall and leave just 0.1mbps on the rest of the scene.
That will unarguably provide a drastically worse experience on a high-enough quality tv than a 40mbps+ bluray. Like, day and night in most scenes and even more in others.
I can tell you that this is simply false. And it’s the same psuedo-scientific logic that someone trying to sell gold plated cables and FLAC encodings pushes.
Look, beyond just the darkness tuning problem that streaming services have, the other problem they have is a QOS. The way content is encoded for streaming just isn’t ideal. When you say “they have to hit 14mpbs” the fact is that they are forcing themselves to do 14mbps throughout the entire video. The reason they do this is because they want to limit buffering as much as possible. It’s a lot better experience to lower your resolution because you are constantly buffering. But that action makes it really hard to do good video optimizations on the encoder. Ever second of the video they are burning 14mb whether they need those 14mb or not. The way that’d deliver less data would be if they only averaged 14mbps rather than forcing it throughout. Allowing for 40mbps bursts when needed but then pushing everything else out at 1mbps saves on bandwidth. However, the end user doesn’t know that the reason they just started buffering is because a high motion action scene is coming up (and netflix doesn’t want to buffer for more than a few minutes).
The other point I’d make is that streaming companies simply have a pipeline that they shove all video through. And, because it’s so generalized, these sorts of tradeoffs which make stuff look like a blocky mess happen. Sometimes that blocky mess is present in the source material (The streaming services aren’t ripping the blurays themselves, they get it from the content providers who aren’t necessarily sending in raws).
I say all this because you can absolutely get 4k and 1080p looking good at sub-bluray rates. I have a library filled with these re-encodes that look great because of my experience here. A decent amount of HD media can be encoded at 1 or 2mbps and look great. But you have to make tradeoffs that streaming companies won’t make.
For the record, the way I do my encoding is a scene by scene encode using VMAF to adjust the quality rate with some custom software I built to do just that. I target a 95% VMAF which ends up looking just fantastic across media.


Blocky artifacts typically appear in low light situations. There will be situations where it might just be blocky due to not having enough bits (high motion scenes) but there are plenty of cases where low light tuning is where you’d end up noticing the blockyness.


People don’t like hearing this, but streaming services tune their codecs to properly calibrated TVs. Very few people have properly calibrated TVs. In particular, people really like to up the brightness and contrast.
A lot of scenes that look like mud are that way because you really aren’t supposed to be able distinguish between those levels of blackness.
That said, streaming services should have seen the 1000 comments like the ones here and adjusted already. You don’t need bluray level of bits to make things look better in those dark scenes, you need to tune your encoder to allow it to throw more bits into the void.


Yup, the band is already littered with 6g devices. It’d be a stupid purchase.
But also, 6GHz is somewhat of a useless band for carriers. It’s high enough frequency that it’ll get absorbed by most things yet low enough frequency that it’ll struggle to really carry a whole lot of data.


Every CEO thinks like this. CEOs are so incredibly bullish on AI BECAUSE they want to replace people and not tasks.


Just reread it and no, it’s not a BT vulnerability. The “erase flash” command is something that has to be done by software running outside the BT stack. You can even see that inside the slides. The UsbBluetooth software is connected to the device with the flawed bluetooth chipset.
The vulnerability is that if you have this chipset and compromised software, someone can flash the chipset with compromised flash. They even say that it’s not an easy attack to pull off in the article.
In general, though, physical access to the device’s USB or UART interface would be far riskier and a more realistic attack scenario.
In otherwords, the attack is something that can only be pulled off if there’s also a security vulnerability within other parts of the hardware stack.


I just re-read the article and yes, you still need physical access.
The exploit is one that bypasses OS protections to writing to the firmware. In otherwords, you need to get the device to run a malicious piece of code or exploit a vulnerability in already running code that also interacts with the bluetooth stack.
The exploit, explicitly, is not one that can be carried out with a drive-by Bluetooth connection. You also need faulty software running on the device.


I think we are basically already there with ESPs :D.


Security wise, unless you are being specifically targeted by someone, you are almost certainly fine. And if you are being specifically targeted, I think someone hacking your ESPs is the least of your worries. A malicious attacker that knows your physical location can do a lot more scary things than just spying through ESPs.


You’re fine. This isn’t something that can be exploited over wifi. You literally need physical access to the device to exploit it as it’s commands over USB that allow flashing the chip.
This is a security firm making everything sound scary because they want you to buy their testing device.
Not just copper. These server racks are so unfortunately loaded up with parts and components worth a lot of money on the open market. Very unfortunately they have enterprise grade SSDs worth thousands and AI accellerater cards worth 10s of thousands.
And these poor starving companies very unfortunately don’t have the funds to hire a lot of security staff. It’s sadly usually just one guy.