I'm running 8000 in 2:1 on my 9800x3D. Slightly better than 6200 1:1 on here, and basically the same as 6400 1:1, but easier to dial in at 8000.
Side note: At 2ghz IF, a single Zen4/5 CCD can saturate 7200 MT DDR5, but the io die is not restricted to that, so x16 and IGPU DMA can still saturate it at 8000.
There isn't a single country in the world that can make advanced chips on its own. It takes machines, raw materials an chemicals from over a dozen different countries.
You're absolutely right! And that is why the last couple of years has been fascinating to watch from a geopolitical and trade standpoint.
Watching China push its weight around earlier in the year with REM and REE was crazy. The US, and the rest of the world for that matter, were brought to our knees. Germany's auto industry and weeks of chips left, the US started shipping F35's to allies _without_ radar. Truly astounding trade power to see.
Made it very clear to every nation that you must have a sovereign story to tell about chip manufacturing.
> A country must feed itself, power itself, and in 2026 it must be able to manufacture its own chips (from ore, to refining, to the foundry)
This is only possible if you (a) fudge the definitions of "everything" a bit and (b) are one of the largest countries in the world. This is only ever going to be possible for US and China. And even then the US really is the "lucky country" with which natural resources it has ended up holding within its borders.
While Intel had bad sales of CPUs for personal computers, due to the memory crysis, they have sold every server CPU they could make, despite the fact that their server CPUs are much inferior in comparison with the AMD server CPUs.
The reason is that AMD could not make enough server CPUs to satisfy the demand for datacenters, due to limited production capacity at TSMC, so Intel, who makes server CPUs in their own fabs, could also sell everything they got.
So as you say, even the older fabs of Intel, which use the somewhat obsolete "Intel 3" CMOS manufacturing process, with which the old Intel Granite Rapids server CPUs are made, have been an important asset in these circumstances.
It's always crazy running into people that are somehow unaware that Wintel has achieved worldwide domination of IT infrastructure (by number of users). This site is like a sealed terrarium of people using macOS on Apple ARM hardware blissfully unaware of the infinite black pit outside the glass.
And yet Intel is the one taking a dump here, because it has no moat. People use Wintel because people use Wintel, the same way people bought Gros Michel bananas because people bought Gros Michel bananas. When they went away, people bought something else.
Still, a lot of that lost share went to AMD and that is ultimately a safe thing for Intel compared to losing share to alternative CPU platforms. More than just safe, mutually beneficial for both companies.
Sure, losing Apple was a huge deal, but ultimately Apple still only has ~20% of the consumer PC market and far less in enterprise, and even less than that in more niche applications since their form factors are so limited, and 0% marketshare in data centers.
I’m not sure how hard Intel is trying to get into the enterprise AI data center scene but Intel ARC cards are definitely floating around in the local LLM scene for their stellar value and I’d be surprised if Intel doesn’t see a massive opportunity to take advantage in that area. They are one of three companies who have reasonably modern GPU designs.
If you pick out one sentence and strip out the rest of the context then all you’re doing is removing meaning from what I was actually saying.
The competitive pressure of AMD versus Intel is mutually beneficial, as long as they don’t totally crush each other into bankruptcy, their small leapfrogs of each other’s x86 technology keeps x86 relevant and dominant.
Key word there is had, while the comment you're replying to is speaking as to today minus ten years roughly.
And yeah, as a millennial I know lots of people who had a shared desktop at home, probably had a laptop in college, used a work laptop, and hasn't had a personal computing device outside an iPad and iPhone for a decade or more. Outside of those who play PC games or other niche hobbies it's getting more and more rare, at least for the people I know.
This is often said, but incorrect. 81% of households have a desktop/laptop. [1] And the breakdown is basically a proxy for wealth. Only 56% of households with earnings of less $25,000 have a desktop/laptop whereas it's 96% for $150k+. For millennials (approximately - age range 35-44) the rate is 87%.
You'd open your phone and nothing would work, not even the connection to the cellular network since your phone company has management components running on AWS.
You might have that problem, but us-east-1 going dark is unlikely to impact my Australian carrier enough to knock cellular service out. But granted, I might notice something is up if I tried to use my phone for much. If us-east-1 vanished at the right time of day I might notice on my phone first, sure.
> You'd open your phone and nothing would work, not even the connection to the cellular network since your phone company has management components running on AWS.
I'm pretty confident, sitting here on a different continent, that I wouldn't have that problem at all.
I think parent is trying to say that the number of services that rely themselves on AWS or some of their internal services is enormous and potentially impacts many.
Nah they've been "out of fashion" for 5+ years, and in that time AMD took their datacenter crown, which was also encroached on by various ARM hyperscaler systems.
PCs are like 5% of computing devices. Very important for business offices but not outside of that niche. And mass amounts of servers can be adapted to whatever is cheapest. Google wouldn't insist on AMD64 - they'd port to ARM if ARM was cheaper. And if you do want AMD64, AMD currently makes better AMD64 chips.
Yep, like technicians. Especially technicians that work on, well, everything. Every industrial machine, robot, PLC, industrial vehicle, mining equipment, power plant, etc etc.
This is repeated every time this topic comes up, but so far it's just a conspiracy theory. An urban legend. An old wives' tale. Are there any documented cases of this happening?
Some do this (Samsung), but the more common tactic is to throw constant nags up if the telemetry is not being sent to the manufacturer.
I did my research and bought one that does not do this (2025 TCL), but, if it turned out that it did, I'd have shamelessly returned it. I'd also have issued a chargeback if the store gave me any crap.
Even with all that, if you install firmware updates over the network, then disconnect WiFi, the TV becomes unusable until you factory reset it. (I did this because its EDID is corrupted, and I have to override it with a linux cmdline. The update did not help.)
This hasn't quite been my experience. I got a Samsung QLED TV about a year ago, and besides some frustration at the initial setup experience, I haven't had any issues with it.
The only "nags" I see is if I accidentally press one of the internet-app shortcuts on the remote (netflix, youtube, etc). If you do that, it will give you a popup saying "can't do this without wifi", which isn't entirely surprising. But I took the remote apart and removed the contacts behind those buttons, so they can't be pressed anymore, and I can't remember the last time I saw that popup.
I obviously have no way of telling if it's secretly connected to the sole low-signal open WiFi near me, but I haven't seen a single advertisement or anything that looks vaguely like it could have originated from the internet.
With TV makers having recently destroyed trust and there not being a good way for consumers to force trust it makes little difference whether they are doing it or not.
If you can't tell whether your disconnected TV is spying on you but you know that you cannot trust it, it's better to treat it as compromised.
probably, but I'll let gp answer. But I want guarantees that they don't do that, until them I'll consider the fact that if they can and it's profitable they will do it
If you want a guarantee, that's a conservative threat model and a fair position. What's not fair is GP saying "doesn't work, all the TVs do XYZ", when not a single TV in the real world has ever actually done XYZ.
Ultimately I'll need to find data to settle how paranoid I might be. I'm still leaning on not enough. But isn't it also not fair to claim not a single tv has done it? Proving an universal is not easy for gp but the inverse of forall is not a not exists.
And I know the world of middle management profit optimization.
I indeed cannot prove it. I also can't universally prove that wombats don't have telekinesis. But the preponderance of the evidence points to neither being true.
I know there are lots of advocacy groups and youtube channels that are testing TV models and would LOVE to be the first to break this story and publicly shame Samsung or whoever. But it simply hasn't happened.
I don't know, I found AI autocomple pretty useful, e.g. if I'm coding manually, but still don't want to type tedious, repetitive log messages or other boilerplate myself.
The two Issues I saw were:
- If it says "tab completion", than that's what it should do, and not more. I eventually disabled the copilot model, not because it wasn't useful, but because VSCode's integration was overzealous and kept on suggesting all kinds of things even if I never asked for it, pulling me out of focus every time. Unlike any of their other completions, btw...
- It was not integrated with the existing completions at all, often creating chaos: The standard completions show you available classes, methods, parameters etc, in your current context, based on the information from the language server. The AI completion happily ignored the language server and showed you suggestions for nonexistent fantasy methods instead.
In general, I think acceptance of AI could be better if we offered more "lower-case AI" options, that let people offload the tedious parts of their workflow instead of taking away the entire process of creating.
As an example, there are lots of image generation models that take a short text prompt and will give you a complete finished work, making thousands of creative decisions in the process, according to their training data. Those models do not have artists as their target group - they are basically targeted at everyone but artists and replace their role. Hence the predictable backlash.
But where are the tiny, locally runnable models that, say, turn a sketch into a lineart, or apply coloring and shading according to given reference images?
I think the OG github copilot version felt magical in that time. But I've never tried working like that with current frontier models. Considering how they train it now it might never work without changes how model respond.
But will they? We've been trained to accept slop at every turn. Food tastes worse and we just accept it (society as a whole - not saying you or I personally). I'm sure Disney has competent product people watching over this. They'll A/B test their way into "maximum amount of ads they can show before losing a critical mass of users". They'll optimize. Folks will get used to that as "the new normal". Disney will adjust from time to time to show more ads as people allow.
I do agree though. It seems like pretty bad behavior to charge 20 dollars AND insisting on showering us with ads.
Problem with that font is they haven't pushed a new build of it in over a year. You can take their build file and build it with a newer Iosevka, but you'd have to learn how that tool works (I've done it, its not user friendly).
Side note: At 2ghz IF, a single Zen4/5 CCD can saturate 7200 MT DDR5, but the io die is not restricted to that, so x16 and IGPU DMA can still saturate it at 8000.
reply