MudMan

It is a replacement for a specific portion of a very complicated ecosystem-wide integration involving a ton of interoperability sandwiched between the natural language bits. Why this is a new product and not an Assistant overhaul is anybody’s guess. Some blend of complicated technical issues and corporate politics, I bet.

So an interesting thing about this is that the reasons Gemini sucks are… kind of entirely unrelated to LLM stuff. It’s just a terrible assistant.

And I get the overlap there, it’s probably hard to keep a LLM reined in enough to let it have access to a bunch of the stuff that Assistant did, maybe. But still, why Gemini is unable to take notes seems entirely unrelated to any AI crap, that’s probably the top thing a chatbot should be great at. In fact, in things like those, related to just integrating a set of actions in an app, the LLM should just be the text parser. Assistant was already doing enough machine learning stuff to handle text commands, nothing there is fundamentally different.

So yeah, I’m confused by how much Gemini sucks at things that have nothing to do with its chatbotty stuff, and if Google is going to start phasing out Assistant I sure hope they fix those parts at least. I use Assistant for note taking almost exclusively (because frankly, who cares about interacting with your phone using voice for anything else, barring perhaps a quick search). Gemini has one job and zero reasons why it can’t do it. And it still really can’t do it.

For context, Meta reported 40 billion in revenue during that period, with 24Bn in expenses and made 13bn for the period. All those numbers are up from the same period last year.

So I’m gonna go with “probably, yeah”.

Is the fuel in this case a bunch of damaged 13 and 14th gen CPUs?

Oh, I absolutely could have. It would lose a couple of cores, but the 13th gen is pretty linear, it would have performed more or less the same.

Thing is, I couldn’t have known that then, could I? Chip reviews aren’t aiming at normalizing for temps, everybody is reviewing for moar pahwah. So is there a way for me to know that gimping this chip to run silently basically gets me a slightly overclocked 13600K? Not really. Do I know, even at this point, that getting a 13600K wouldn’t deliver the same performance but require my fans to be back to sounding noticeable? I don’t know that.

Because the actual performance of these is not to a reliable spec other than “run flat out and see how much heat your thermal solution can soak” there is no good way to evaluate these for applications that aren’t just that without buying them and checking. Maybe I could have saved a hundred bucks. Maybe not. Who knows?

This is less of a problem if you buy laptops, but for casual DIY I frankly find the current status quo absurd.

Whoa, that’s even worse. It’s not just the uncertainty of knowing whether Intel will replace your hardware or the cost of jumping ship next time. Intel straight up owes you money. That sucks.

You’re not wrong, but “we’ve been winging it for decades” is not necessarily a good defense here.

That said, I do think they did look at their performance numbers and made a conscious choice to lean into feeding these more power and running them hotter, though. Whether the impact would be lower with more conservative power specs is debatable, but as you say there are other reasons why trying to fake generational leaps by making CPUs capable of fusing helium is not a great idea.

“Clearly damaged” is an interesting problem. The CPU would crash 100% of the time on the default settings for the motherboard, but if you remember, they issued a patch already.

I patched. And guess what, with the new Intel Defaults it doesn’t crash anymore. But it suddenly runs very hot instead. Like, weird hot. On a liquid cooling system it’s thermal throttling when before it wouldn’t come even close. Won’t crash, though.

So is it human error? Did I incorrectly mount my cooling? I’d say probably not, considering it ran cool enough pre-patch until it became unstable and it runs cool enough now with a manual downclock. But is that enough for Intel to issue a replacement if the system isn’t unstable? More importantly, do I want to have that fight with them now or to wait and see if their upcoming patch, which allegedly will fix whatever incorrect voltage requests the CPU is making, fixes the overheating issue? Because I work on this thing, I can’t just chuck it in a box, send it to Intel and wait. I need to be up and running immediately.

So yeah, it sucks either way, but it would suck a lot less if Intel was willing to flag a range of CPUs as being eligible for a recall.

As I see it right now, the order of operations is to wait for the upcoming patch, retest the default settings after the patch and if the behavior seems incorrect contact Intel for a replacement. I just wish they would make it clearer what that process is going to be and who is eligible for one.

No, I have a DDR5 setup. Which is why my motherboard was way more expensive than 100 bucks.

The problem isn’t upgrading to a entry level AM5 motherboard, the problem is that to get back to where I am with my rather expensive Intel motherboard I have to spend a lot more than that. Moving to AMD doesn’t mean I want to downgrade.

I mean, happy for you, but in the real world a 200 extra dollars for a 400 dollar part is a huge price spike.

Never mind that, be happy for me, I actually went for a higher spec than that when I got this PC because I figured I’d get at least one CPU upgrade out of this motherboard, since it was early days of DDR5 and it seemed like I’d be able to both buy faster RAM and a faster CPU to keep my device up to date. So yeah, it was more expensive than that.

And hey, caveat emptor, futureproofing is a risky, expensive game on PCs. I was ready for a new technology to make me upgrade anyway, if we suddenly figured out endless storage or instant RAM or whatever. Doesn’t mean it isn’t crappy to suddenly make upgrading my CPU almost twice as expensive because Intel sucks at their one job.

The article is… not wrong, but oversimplifying. There seem to be multiple faults at play here, some would continue to degrade, others would prevent you from recovering some performance threshold, but may be prevented from further damage, others may be solved. Yes, degradation of the chip may be irreversible, if it’s due to the oxidation problem or due to the incorrect voltages having cuased damage, but presumably in some cases the chip would continue to work stable and not degenerate further with the microcode fixes.

But yes, agreed, the situation sucks and Intel should be out there disclosing a range of affected chips by at least the confirmed physical defect and allowing a streamlined recall of affected devices, not saying “start an RMA process and we’ll look into it”.

So here’s the thing about that, the real performance I lose is… not negligible, but somewhere between 0 and 10% in most scenarios, and I went pretty hard keeping the power limits low. Once I set it up this way, realizing just how much power and heat I’m saving for the last few few drops of performance made me angrier than having to do this. The dumb performance race with all the built-in overclocking has led to these insanely power hungry parts that are super sensitive to small defects and require super aggressive cooling solutions.

I would have been fine with a part rated for 150W instead of 250 that worked fine with an air cooler. I could have chosen whether to push it. But instead here we are, with extremely expensive motherboards massaging those electrons into a firehose automatically and turning my computer into a space heater for the sake of bragging about shaving half a milisecond per frame on CounterStrike. It’s absurd.

None of which changes that I got sold a bum part, Intel is fairly obviously trying to weasel out of the obviously needed recall and warranty extension and I’m suddenly on the hook for close to a grand in superfluous hardware next time I want to upgrade because my futureproof parts are apparently made of rust and happy thoughts.

I have a 13 series chip, it had some reproducible crashing issues that so far have subsided by downclocking it. It is in the window they’ve shared for the oxidation issue. At this point there’s no reliable way of knowing to what degree I’m affected, by what type of issue, whether I should wait for the upcoming patch or reach out to see if they’ll replace it.

I am not happy about it.

Obviously next time I’d go AMD, just on principle, but this isn’t the 90s anymore. I could do a drop-in replacement to another Intel chip, but switching platforms is a very expensive move these days. This isn’t just a bad CPU issue, this could lead to having to swap out two multi-hundred dollar componenet, at least on what should have been a solidly future-proof setup for at least five or six years.

I am VERY not happy about it.

Yeah, the notion that Lemmy is a Reddit replacement is misguided. It definitely doesn’t have the same Q&A balance Reddit does. It feels a lot more like 90s and early 2000s forums than the large-scale self-service link and customer service churn Reddit encourages.

Which I’m all for. I was never a Reddit guy and I do like it here. But in terms of how bad it is now that Reddit is not happy to host most of the actually useful online content for free… well, that’s a different conversation.

Okay, hear me out, we should make a different service, like a network of computers where people can freely post information on some sort of page that other people can freely access without needing to log in to a million services. Maybe we could also add like little conversation boards to those to allow people to ask and answer questions, too.

Wild, I know, but I think there is some opportunity there.

Yeah, it seems more interesting to reverse engineer why they chose this line of marketing. They are clearly misrepresenting the challenge and cost of running a LLM locally, so… why?

I don’t thin anyone believes that when you put it that way, but… that’s kinda what is being said, yeah.

No, I get it, I’m not saying you’re doing it on purpose. I’m saying the cynical right wing anti-regulation talking points have seeped so deep into the fabric of social disenchantment that they are parroted universally now.

No, artificially crippled regulation being bad at regulating doesn’t change the need for regulation or the collective onus on regulating abusive corporate behavior. No, you don’t just control you. You take collective action. Individual action means squat and does nothing. Nobody has ever “voted with their wallet”. You vote with your votes, and if that isn’t allowed or is rendered ineffective you vote with something pointier.

So no, you don’t -can’t- abstain from participating in that process as a political act. You engage in political action. Especially if the performative pantomime of activisim is arbitrary and driven by false premises and misconceptions.

Look, social media is fundamentally harmful. Including this site. Meta invented a good chunk of it and monetized most of it using anticompetitive practices and abdicating their responsibilities. And they should be broken apart into manageable chunks and regulated within an inch of their lives.

None of which has any bearing on them spending their considerable resources on subsidizing some gaming-focused VR headset. And even if the perceived slight on that particular product was offensive, it certainly isn’t more offensive than what Google does with Gmail. Or what Unilever does with pretty much everything you buy that has some chemicals in it, or whatever else.

I don’t care or want to be an online vigilante, taking valiant ranting action against whatever company is unlucky enough to have been part of an article I read once. I want to pay my taxes, vote in conscience, unleash the hellhounds of the hard end of the social contract and get to buy whatever cool shit I think is cool without having to bear the burden of moral judgement from highly ineffective keyboard warriors. Or, you know, at least only bear the judgement that is based on actual facts, not made up grudges.

Libertarian talking points again. Every instance of a government failing to do their jobs is an indictment on the concept of governments.

There are two things you need for effective governance: effective people in government and effective design of government.

If you don’t have those, your only priority is to get those (so get to it, Americans). If you do have those, your job is to oversee them doing their job.

Neither of those tasks particularly requires getting into spats with random Internet people in obscure corners of social media to flaunt how much you dislike anything associated with a particular brand you’ve chosen as an arbitrary avatar for a specific issue. One could argue for an organized boycott, but that’s not what’s happening here, and if it was it’d be the most ineffectual example of one on record.

Ah, yes, that’s why I keep my walls unpainted. Can’t trust big paint with the not putting lead in that. Never going to happen.

You’re not pushing back against anything, you’re arguing about Meta’s nichest product on the nichest social media of them all. At some point one will have to assume the arguing is the point, more than enacting any change.

Between regulation of the techbro oligarchy or… whatever this thread is, I know where my money goes. The opposite is straight up libertarian talk.